Clinical and infusion product characteristics influencing the outcomes and toxicity of CAR T-cell treatment with axicabtagene ciloleucel for LBCL

Chimeric antigen receptor (CAR) T-cell therapy represents a novel treatment for the management of several hematological malignancies, including large B-cell lymphoma (LBCL). Despite high clinical efficacy, treatment response and outcomes can be variable. Challenges faced include primary treatment failure, relapse, and significant toxicities, including cytokine release syndrome (CRS) and neurological events (NE) (e.g., immune effector cell associated neurotoxicity syndrome [ICANS]). The mechanisms underlying this variance in clinical response remain unclear, with no risk factors clearly identified among those commonly suspected (e.g., performance status, cytogenetic factors, and disease stage).

Here we summarize the results of two studies exploring factors that influence the efficacy and toxicity of CAR T-cell treatment with axicabtagene ciloleucel (axi-cel) for LBCL. Frederick Locke and colleagues provide findings from a comprehensive analysis of covariates and biomarkers not previously reported from the ZUMA-1 LBCL trial (published in Blood Advances, October 2020)¹. Meanwhile, Qing Deng and colleagues present their findings on how transcriptomic and molecular features are associated with outcomes and efficacy (published in Nature Medicine, October 2020)².

Study 1: Biomarker and clinical correlates of CAR T-cell response and toxicity¹

Samples and data from the multicenter phase I/II ZUMA-1 trial (NCT02348216) were used for the current analysis. Two-year results from this study have been reported previously on the Lymphoma Hub. In summary:

• A total of 101 patients with diffuse LBCL (DLBCL), primary mediastinal B-cell lymphoma (PMBL), or transformed follicular lymphoma (tFL) were treated with axi-cel in phase II.
• At a median follow-up of 27.1 months, 58% of patients remained in complete response (CR).
• Grade ≥ 3 CRS occurred in 11% of patients, and Grade ≥ 3 NE occurred in 32% of patients.

Study design and methods¹

• Durable response (DR) was defined as an ongoing response ≥ 1 year after axi-cel infusion.
• Relapse was defined as disease progression in patients who achieved a CR or partial response (PR).
• CAR T-cell quantities were measured by quantitative PCR prior to infusion.
• Serum cytokine levels were measured at baseline (pre-conditioning), Day 0 (the day of axi-cel infusion), or the day after axi-cel infusion.
• Lactate dehydrogenase was measured using routine laboratory testing.
• Tumor burden was estimated as the sum of product diameters of ≤ 6 index lesions (Cheson 2007 criteria).
• Infusion product fitness was estimated by the doubling time in vitro.
• CAR T-cell expansion in vivo was measured following infusion.

Key findings²

Efficacy

The factors most strongly associated with DR to CAR T-cell treatment are represented in Table 1. A higher expansion rate of product T-cells before infusion (doubling time) was correlated with greater in vivo CAR T-cell levels. The strongest positive correlate of DR was peak CAR T-cell levels normalized to pretreatment tumor burden, whereas high tumor burden and pro-inflammatory state were negatively associated with DR.

Table 1. Factors associated with DR¹

CAR, chimeric antigen receptor; DR, durable response; IL-6, interleukin-6; LDH, lactate dehydrogenase; TB, tumour burden.
Factor	Probability of DR	p value
Peak CAR T-cells (cells/µl)	Increased	0.0159
Peak CAR T-cells/tumor burden (108cells/mm2)	Increased	0.0017
Higher baseline tumour burden	Decreased	0.0259
Increasing baseline ferritin (pg/mL)	Decreased	0.123
Increasing baseline LDH (pg/mL)	Decreased	0.0251
Increasing baseline IL-6 (pg/mL)	Decreased	0.0237
Increased number of infused CD8 cells, normalized to TB	Increased	0.0108
Increased number of CCR7+CD45RA+ T-cells infused	Increased	0.0301

Toxicities

Factors associated with increased risk of ≥ Grade 3 CRS and NE are represented in Table 2.

High tumour burden and proinflammatory state were positively correlated with an increased risk of significant toxicity.

Table 2. Tumor biomarkers correlated with increased risk of significant toxicities¹

CCL2, C-C motif chemokine ligand 2; CRS, cytokine release syndrome; IL-6, interleukin-6; LDH, lactate dehydrogenase; NE, neurological events.
Factor associated with toxicity	Grade ≥ NE, p value	Grade ≥ CRS, p value
Baseline tumor burden (mm2)	0.0164	Not significant
Day 1 ferritin (ng/mL)	0.062	Not significant
Baseline LDH (pg/mL)	0.0006	0.00275
Day 1 CCL2 (pg/mL)	0.003	0.0774
Baseline IL-6 (pg/ml)	Not significant	0.0391

Study limitations¹

The authors identified several limitations, including:

• Other potential immune covariables, such as axi-cel product immunogenicity and tumor microenvironment, were not considered.
• ZUMA-1 trial patients may not be entirely representative of those treated with standard of care axi-cel due to trial inclusion criteria and protocol.
• Adverse event management may be different between patients as this was physician-led.
• Common Terminology Criteria for Adverse Events (CTCAE) criteria was used for grading adverse events; new criteria now established may cause conflict when extrapolating current findings to future patients.

Study 2: Molecular and transcriptomic correlates of CAR T-cell therapy response and toxicity²

Study design and methods²

A biological, observational study of 24 patients with LBCL (16 DLBCL, 6 tFL, and 2 PMBL) receiving standard of care axi-cel. Analyses included the following:

• Single-cell transcriptome profiling of axi-cel infusion products.
• Clinical and molecular assessments of treatment efficacy by PET/CT and cell-free DNA sequencing of patient plasma samples.
• Clinical toxicity grading.

Primary end points were as follows:

• ICANS Grade ≥ 3.
• A 3-month clinical assessment of tumour burden by PET/CT.

Key findings²

• At 3 months after CAR T-cell treatment, 13 patients (50%) had progressive disease (PD), one patient (4%) was in PR, and nine patients (28%) had a CR.
• While patients with CR received CAR T-cell infusions enriched with memory CD8 T-cells, CAR T-cell infusion products from patients with PD/PR had an enrichment of exhausted CD8 and CD4 T cells, which was associated with a poor molecular response as measured by cell-free DNA sequencing.
• Patients with an early molecular response, within 7 days, were more likely to achieve a CR at their 3-month follow-up, as observed by PET/CT, compared with those without (p = 0.008).
• Infusions containing a monocyte-like cell subset were found to be at a significantly increased risk of ICANS Grade 3-4 (p = 0002).

Conclusion

CAR T-cell therapy continues to hold promise for patients with LCBL, both treatment naïve and those with refractory disease. Together, these studies confirm that the response to treatment and risk of side effects is affected by CAR T-cell infusion characteristics, tumour burden, host systemic inflammation at baseline, and molecular response to treatment. Large clinical trials are needed to validate these findings, and to determine if further product optimization and treatments such as immune modulation can improve outcomes and safety.