Genomic features of CAR T treatment failure in patients with lymphoma

CD19-directed chimeric antigen receptor (CAR-19)-reprogrammed autologous T-cells are innovative immunotherapies for heavily pretreated patients with diffuse large B-cell lymphoma (DLBCL); however, across CAR-19 products, ~60% of patients fail to achieve complete tumor eradication and prolonged remission. Inflammatory markers and clinical factors are associated with impaired responses, but tumor-intrinsic resistance drivers are largely undefined.

Here, we summarize the key genomic features underlying anti-CD19 CAR T-cell treatment failure in patients with lymphoma based on the recent article published by Jain, et al., in Blood in 2022.¹

Methods

Whole-genome sequencing (WGS) analysis was performed on samples from patients with relapsed or refractory (R/R) DLBCL who had received treatment with CAR-19 T cells (R/R lymphoma cohort, n = 49). To increase the statistical power, WGS data from untreated DLBCL cases from the Pan-Cancer Analysis of Whole Genomes (PCAWG) were also added to the comparative analysis (PCAWG cohort, n = 50).

Results

• Follow up and clinical outcomes of 47 out of 49 LBCL patients were included in the progression-free survival (PFS) and overall survival (OS) analysis (excluded cases: relapse-only sample with transitory complete remission [n = 1] and early therapy-related death [n = 1] not evaluable for PFS).
• Demographics, disease characteristics, and responses to CAR-19 for all patients are summarized in Table 1.
• The median follow-up was 17.3 months.
• Median OS and PFS for the entire cohort were 11.6 months and 8 months, respectively.
• The progression-free response was observed in 36.7% of patients.

Table 1. Patient baseline characteristics*

CAR-19, CD19-directed chimeric antigen receptor; DLBCL, diffuse large B cell lymphoma; ECOG, Eastern Cooperative Oncology Group; HDT/ASCR, high-dose therapy with autologous stem-cell rescue; IPI, International Prognostic Index; TCLL, transformed chronic lymphocytic leukemia; TFL, transformed follicular lymphoma. *Adapted from Jain, et al.1
Characteristics, % (unless otherwise stated)	All patients (N = 49)
Median age (range), years	65 (44–79)
Sex
Male	77.6
Female	22.4
Disease
DLBCL	81.6
TFL	16.3
TCLL	2.0
Stage at apheresis
I/II	16.3
III/IV	83.7
IPI at apheresis
1–2	30.6
3–5	69.4
ECOG Performance Status at apheresis
0–1	77.6
2–3	22.4
Prior treatment regimens
Median (range)	2 (1–6)
Salvage Chemotherapies
Platinum compounds	81.6
Cisplatin	22.4
Carboplatin	34.7
Oxaliplatin	30.6
Melphalan	22.4
Previous HDT/ASCR	22.4
Bridging therapy
Yes	69.4
No	30.6
CAR-19 Product
Axicabtagene ciloleucel	91.8
Tisagenlecleucel	4.1
Lisocabtagene maraleucel	4.1
Cytokine release syndrome
Grade 0	18.4
Grade 1–2	73.5
Grade 3–5	8.2
Immune effector cell-associated neurotoxicity syndrome
Grade 0	32.7
Grade 1–2	36.7
Grade 3–4	30.6
CAR-19 Outcome
Response without progression	36.7
Response with progression (relapse)	46.9
Refractory disease	16.3

Treatment resistance

Overall, markers of genomic complexity (APOBEC and chromothripsis) and specific genomic alterations (RHOA) were associated with resistance to CAR-19 immunotherapy for aggressive B-cell lymphomas. Further details on the genomic complexity and specific genome alterations are summarized below.

Mutational signature gene association

Twelve single base substitution (SBS) mutational signatures were involved in the R/R lymphoma cohort.

• Presence of APOBEC (SBS2 and SBS13) signatures was associated with significantly worse PFS, with 12/13 (92%) patients progressing (all patients, p = 0.0023; considering only de novo DLBCL [n = 40], p = 0.045).
• Among additional mutational signatures tested, SBS18 was associated with post-CAR-19 progression in 9/11 (81%) patients (all patients, p = 0.0396; considering only de novo DLBCL [n = 40], p = 0.045).
• This SBS signature reflects genomic damage from oxygen-radical stress.

Deletion of 3p21.31 (RHOA) association

• TP53 deletion was the most frequent in 49 R/R patients and mono-allelic or biallelic loss of TP53 was highly prevalent (59.2%; p = 0.76) but did not predict the poor outcome.
• Focal deletions of 3P21.31 containing RHOA were found to be significantly enriched in the R/R patient’s cohort and independently predicted poor outcomes (p = 0.0013).
• Loss of RHOA was also associated with poor OS (p = 0.023).

Structural variants

WGS identified four main complex structural variants and complex events:

• Double minutes (n = 6)
• Chromothripsis (n = 23)
• Chromoplexy (n = 18)
• Templated insertion (n = 11)

Only chromothripsis showed worse PFS (p = 0.026) after CAR-19 treatment, with 18/22 (81.8%) cases (R/R DLBCL) experiencing early progression; however, this was not associated with OS. All de novo DLBCL with double minutes rapidly progressed and died after CAR-19 (p = 0.017 for PFS; p = 0.0011 for OS).

Conclusion

Despite unprecedented overall response rates to CAR-19 in heavily pretreated patients with DLBCL, a significant number of patients often fail to achieve survival outcomes. Genomic complexity and alterations appear to promote an immunosuppressive tumor microenvironment, limiting CAR-19 efficacy. The results discussed here are the foundation of further functional studies that can establish mechanistically how complex lymphoma genomes promote an environment hostile to CAR T cells and possibly other emerging immunotherapies.