Tumor burden, inflammation, and product attributes determine outcomes of axicabtagene ciloleucel in large B-cell lymphoma

Abstract

ZUMA-1 demonstrated a high rate of durable response and a manageable safety profile with axicabtagene ciloleucel (axi-cel), an anti-CD19 chimeric antigen receptor (CAR) T-cell therapy, in patients with refractory large B-cell lymphoma. As previously reported, prespecified clinical covariates for secondary end point analysis were not clearly predictive of efficacy; these included Eastern Cooperative Oncology Group performance status (0 vs 1), age, disease subtype, disease stage, and International Prognostic Index score. We interrogated covariates included in the statistical analysis plan and an extensive panel of biomarkers according to an expanded translational biomarker plan. Univariable and multivariable analyses indicated that rapid CAR T-cell expansion commensurate with pretreatment tumor burden (influenced by product T-cell fitness), the number of CD8 and CCR7+CD45RA+ T cells infused, and host systemic inflammation, were the most significant determining factors for durable response. Key parameters differentially associated with clinical efficacy and toxicities, with both theoretical and practical implications for optimizing CAR T-cell therapy. This trial was registered at www.clinicaltrials.gov as #NCT02348216.

Conflict of interest statement

Conflict-of-interest disclosure: F.L.L. reports scientific advisory roles for Novartis, Celgene/Bristol-Myers Squibb, Kite/Gilead, GammaDelta Therapeutics, Calibr, Amgen, Wugen, and Allogene; consultant with grant options for Cellular Biomedicine Group, Inc.; and research funding from Kite/Gilead. J.M.R. and A.X. report employment with Kite (a Gilead Company) and stock or other ownership in Gilead Sciences. S.S.N. reports personal fees from Kite, Merck, Bristol-Myers Squibb, Novartis, Celgene, Pfizer, Allogene Therapeutics, Cell Medica/Kuur, Incyte, Precision Biosciences, Legend Biotech, Adicet Bio, Calibr, and Unum Therapeutics; research support from Kite, Bristol-Myers Squibb, Merck, Poseida, Cellectis, Celgene, Karus Therapeutics, Unum Therapeutics, Allogene Therapeutics, Precision Biosciences, and Acerta; royalties from Takeda Pharmaceuticals; and intellectual property related to cell therapy. C.A.J. reports honoraria from Kite, Novartis, Celgene, BMS, Precision Biosciences, Nkarta, and Lonza; consultancy or advisory role for Kite, Novartis, Celgene, BMS, Precision Biosciences, Nkarta, and Lonza; speakers' bureau participation for AXIS and Clinical Care Options; research funding from Pfizer; and travel support from Kite, Novartis, BMS, Celgene, Precision Biosciences, Nikarta, and Lonza. D.B.M. reports consultancy or advisory roles for Kite, Novartis, Celgene, Juno, Bristol-Myers Squibb, Adaptive Biotechnologies, Pharmacyclics, and Janssen; research funding from Kite, Novartis, Celgene, Juno, Bristol-Myers Squibb, Adaptive Biotechnologies, and Pharmacyclics; patents, royalties, or other intellectual property from Pharmacyclics; and travel support from Kite, Novartis, Juno, Celgene, Bristol-Myers Squibb, Adaptive Biotechnologies, Pharmacyclics, and Janssen. A.G. reports consulting or advisory roles with Kite, Amgen, Atara, Wugen, and Celgene and honoraria from Kite. O.O.O. reports honoraria from Kite; consultancy or advisory role for Kite, Pfizer, Spectrum Pharmaceuticals, and Bayer; and research funding from Kite. P.M.R. reports consultancy or advisory roles for Kite, and Curis and has received research funding from Seattle Genetics. Y.L. reports a consultancy or advisory role for Janssen, Celgene, Bluebird Bio, Legend, Sorrento, Gamida Cells, Vineti, Novartis, Kite, and Juno Therapeutics and has received research funding from Janssen, Celgene, Bluebird Bio, Merck, Kite, and Takeda. M.S. reports employment with Kite and stock or other ownership in Gilead Sciences. M.B. reports previous employment at Kite, stock or other ownership in Gilead Sciences, and consultancy or advisory role for various confidential companies. W.Y.G. reports previous employment with Kite and current employment with A2 Biotherapeutics and stock or other ownership in Gilead Sciences and A2 Biotherapeutics. J.S.W. reports previous employment at, travel, accommodations, and expenses from, and patents, royalties, and other intellectual property with Kite, and stock or other ownership in Gilead Sciences. A.B. reports employment with Gilead Sciences, stock or other ownership in Gilead Sciences and Kite, and consultancy or advisory role for Gilead Sciences; and travel support from Gilead Sciences. L.J.L. declares no competing financial interests.

© 2020 by The American Society of Hematology.

Figures

Graphical abstract

Figure 1.

CAR T-cell expansion commensurate with baseline TB is associated with durable responses after axi-cel. Analysis of peak (A) and 3-month CAR T-cell expansion (B) by response status. (C) The association of baseline tumor and probability of durable response was assessed by logistic regression. (D) Scatter plot of baseline TB and peak CAR T-cell levels. (E-F) Analysis of peak CAR T-cell levels normalized to TB by response status. Logistic regression analysis evaluating the association of durable response with peak CAR T-cell levels (G) and peak CAR T-cell levels (H) normalized to TB. P values were calculated using Kruskal-Wallis and Dunn’s tests. CR, complete response; NR, no response; PR partial response.

Figure 2.

Baseline systemic inflammation is negatively associated with both CAR T-cell expansion relative to pretreatment TB and the rate of durable responses. (A-C) Peak CAR T-cell expansion and peak CAR T-cell expansion normalized to TB were analyzed by quartile (Q) analyses of proinflammatory and myeloid activation markers. (D-F) Logistic regression analysis evaluating the association of durable response with baseline proinflammatory and myeloid activation markers. Bar graphs show medians per quartile, and Spearman’s correlation was used to calculate R and P values for all bar graphs.

Figure 3.

Higher expansion rate of product T cells measured before infusion (doubling time) is associated with greater in vivo CAR T-cell levels and efficacy and correlates with T-cell phenotype. Logistic regression analysis showing association of response (A) or durable response (B) with doubling time. Doubling time by peak CAR T-cell expansion and peak CAR T-cell expansion normalized to TB (C) or by CAR area under the curve (AUC) (D) were analyzed by quartile analyses. Scatter plots show association of doubling time with specified T-cell populations (E-H) and CD4:CD8 ratio (I). Bar graphs show medians per quartile, and Spearman’s correlation was used to calculate R and P values for all bar graphs and scatter plots.

Figure 4.

The proportion of T cells with a more juvenile phenotype in the apheresis material directly associates with a lower product doubling time. Association between T-cell phenotypes in apheresis material pre-gated on live CD45+ cells and product phenotype (A-C) or product doubling time (D-H). Spearman’s correlation was used to calculate R and P values.

Figure 5.

The number of CD8 and CCR7+CD45RA+T cells commensurate with TB is critical to achieving durable response after axi-cel. (A-F) Logistic regression analysis of response (right) and durable response (middle) and quartile analysis of peak CAR T-cell levels and peak CAR T-cell levels normalized to TB (left) by the number of CD8 T cells (A-C) or the number of CD8 T cells normalized to TB (D-F). (G) The number of CD8 T cells among patients with low TB (below median, left) and high TB (above median, right) by response. (H-P) Logistic regression analysis of response (right), durable response (middle), and quartile analysis of peak CAR T-cell levels and peak CAR T-cell levels normalized to TB (left) by the number of CCR7+CD45RA+ T cells (H-J), the number of CCR7+CD45RA+ T cells normalized to TB (K-M), or CD4:CD8 ratio (N-P). P values were calculated using Kruskal-Wallis and Dunn’s tests for box plots. Spearman’s correlation was used to calculate R and P values for bar graphs.

Figure 5.

The number of CD8 and CCR7+CD45RA+T cells commensurate with TB is critical to achieving durable response after axi-cel. (A-F) Logistic regression analysis of response (right) and durable response (middle) and quartile analysis of peak CAR T-cell levels and peak CAR T-cell levels normalized to TB (left) by the number of CD8 T cells (A-C) or the number of CD8 T cells normalized to TB (D-F). (G) The number of CD8 T cells among patients with low TB (below median, left) and high TB (above median, right) by response. (H-P) Logistic regression analysis of response (right), durable response (middle), and quartile analysis of peak CAR T-cell levels and peak CAR T-cell levels normalized to TB (left) by the number of CCR7+CD45RA+ T cells (H-J), the number of CCR7+CD45RA+ T cells normalized to TB (K-M), or CD4:CD8 ratio (N-P). P values were calculated using Kruskal-Wallis and Dunn’s tests for box plots. Spearman’s correlation was used to calculate R and P values for bar graphs.

Figure 6.

Factors differentially associated with toxicities: TB, inflammatory markers, and key product attributes. Logistic regression analysis of the probability of grade ≥3 NE (left) or CRS (right) and baseline TB (A), the number of peak CAR T-cells normalized to TB (B), the number of CCR7+CD45RA+ CD8 T cells normalized to TB (C), baseline IL-6 (D), day 1 ferritin (E), day 1 CCL2 (F), baseline LDH (G), and IFN-γ (H).  P values were calculated using logistic regression.

Figure 7.

TB, LDH, and pro-inflammatory markers measured before CAR T-cell infusion associate differentially with clinical outcomes in multivariable analysis. (A) Cluster analysis summarizing the strength of association between covariates from the 2 major categories: product attributes and pretreatment tumor/inflammatory markers. (B-D) Top covariates differentially associated with efficacy and NEs (B), efficacy and CRS (C), and NEs and CRS (D) by random forest analysis (n = 97-101 patients per parameter [supplemental Table 9]). (E) Summary of multivariable findings. CRP, C-reactive protein.