Primary resistance to CD19-directed chimeric antigen receptor T-cell therapy in T-cell/histiocyte-rich large B-cell lymphoma

TO THE EDITOR:

We read with interest the recent report in Blood by Griffin et al,¹  which provided a detailed description of the topographical immune landscape of T-cell/histiocyte-rich large B-cell lymphoma (T/HRLBCL) and implicated the programmed cell death protein 1 (PD-1)/programmed cell death ligand 1 (PD-L1) pathway as a key driver of immune escape in this disease. T/HRLBCL is an uncommon variant of diffuse large B-cell lymphoma (DLBCL) that frequently presents at advanced stages with extranodal involvement in young to middle-aged men.²  T/HRLBCL shares molecular and genomic characteristics with nodular lymphocyte-predominant Hodgkin lymphoma,^3,4  and the T/HRLBCL environment is also reminiscent of classical Hodgkin lymphoma. For instance, in contrast to most DLBCLs, which appear histologically as sheets of malignant B cells, T/HRLBCL is characterized by rare tumor cells scattered among a dense background of reactive T cells and macrophages (histiocytes).

A recent genetic and quantitative spatial analysis has provided important insights into the complex interactions occurring at the tumor-immune interface in T/HRLBCL.¹  Specifically, investigators identified recurrent PD-L1 copy gains associated with high PD-L1 expression on malignant B cells often surrounded by an abundance of PD-L1–expressing macrophages and PD-1⁺ T cells. Interestingly, we have also identified a subset of DLBCLs similarly characterized by PD-L1 gene alterations, an inflammatory microenvironment, and responsiveness to PD-1 blockade therapy.^5,6  It is therefore not surprising that 3 of 5 T/HRLBCL patients in the aforementioned study achieved objective responses to anti–PD-1 immunotherapy.¹  Collectively, these findings suggest that PD-L1 is a dominant immune checkpoint that mediates the dysfunction of endogenous T cells in T/HRLBCL.

The impact of the T/HRLBCL immune environment on the fate of adoptively transferred chimeric antigen receptor (CAR) T cells is not known. This question is highly relevant because, although CD19-directed CAR T-cell therapy has transformed the treatment of relapsed/refractory (r/r) DLBCL,^7,8  its efficacy in uncommon DLBCL subtypes, such as T/HRLBCL, is unknown, which represents a critical knowledge gap. Between July 2017 and December 2019, we identified 9 patients with r/r T/HRLBCL treated with axicabtagene ciloleucel (axi-cel) or tisagenlecleucel (tisa-cel) CD19-directed CAR T-cell therapy at our institutions (patient tissue sections were obtained from institutional review board–approved institutional biorepositories in accordance with the Declaration of Helsinki). Seven patients received commercial CAR T-cell therapy, and 2 were treated on clinical trials of US Food and Drug Administration (FDA)-approved anti-CD19 CAR T cells for investigational indications. Patient characteristics are provided in Table 1. Patients were all male with a median age of 42 years and had received 1 to 5 prior treatments. CD19 expression was present on lymphoma cells in all evaluable pretreatment biopsies. Baseline metabolic tumor volume (MTV), serum lactate dehydrogenase, ferritin, and C-reactive protein were assessed in 7 of 9 patients (Table 1). Prior to CAR T-cell infusion, patients received lymphodepleting chemotherapy with FDA-recommended doses of fludarabine and cyclophosphamide. All patients were administered a single CAR T-cell infusion at a standard dose (axi-cel, 2 × 10⁶ viable CAR⁺ T cells per kilogram; tisa-cel, 0.6-6.0 × 10⁸ viable CAR⁺ T cells).

Cytokine release syndrome (CRS) occurred in all patients (grade 1-2), and immune effector cell-associated neurotoxicity syndrome (ICANS) was observed in 6 patients (grade 1-4), as assessed by American Society for Transplantation and Cellular Therapy (ASTCT) guidelines.⁹  Five patients were administered tocilizumab and 2 received corticosteroids. Response assessments using positron emission tomography/computed tomography (PET/CT) imaging were performed between days 30 and 90 following CAR T-cell infusion. Remarkably, imaging demonstrated progressive disease by day 90 in all 9 patients. Eight of 9 patients had confirmatory biopsies. CD19 expression was maintained on lymphoma cells in all 5 assessable cases.

To investigate potential mechanisms underlying CAR T-cell therapy resistance, we performed a kinetic analysis of CAR T-cell expansion in the peripheral blood of 3 T/HRLBCL patients with available material. Here, we observed clear evidence of CAR T-cell expansion and contraction in a time-dependent manner (Figure 1A-B). Interestingly, PD-1 was highly coexpressed on a large proportion of circulating CAR T cells in all 3 patients, particularly at peak expansion (Figure 1A-B). These data suggest that failure of CAR T-cell therapy in T/HRLBCL is not likely due to poor CAR T-cell expansion, and argue that other factors, such as acquired CAR T-cell dysfunction, could be responsible for the poor clinical outcomes observed.

Given the striking PD-1 expression on CAR T cells described herein, the extent and cellular distribution of PD-L1 expression in the T/HRLBCL environment was defined through multispectral immunofluorescence (mIF) microscopy on 3 available pretreatment biopsies. As shown in Figure 1C, mIF analysis revealed strong CD19 expression on Pax5⁺ lymphoma cells, as well as a prominent T-cell and macrophage infiltrate (Figure 1C-E). Numerous PD-1⁺ T cells were also identified in close proximity to lymphoma cells throughout the tumor microenvironment (Figure 1F). PD-L1 expression was particularly abundant on CD68⁺ macrophages that were often juxtaposed to sparsely distributed lymphoma cells, which also expressed PD-L1 (Figure 1G-H). PD-L1 fluorescence in situ hybridization (FISH) demonstrated PD-L1 gene amplification in 1 of 4 T/HRLBCL cases (Figure 1I). At the time of lymphoma progression following CAR T-cell therapy, tissue was available for mIF analysis in 1 T/HRLBCL case. Here, lymphoma-involved bone marrow showed preserved CD19 expression on malignant B cells that were PD-L1⁺ (Figure 1J) and surrounded by numerous PD-1–expressing T cells (Figure 1K). Given these findings, 5 T/HRLBCL patients were treated with anti–PD-1 therapy following CAR T-cell progression, and 2 achieved objective responses, including the patient with a PD-L1 gene amplification (Table 1).

In conclusion, we report that r/r T/HRLBCL is highly resistant to CD19-directed CAR T-cell therapy. This observation is striking, as reported response rates to CAR T-cell therapy in r/r DLBCL are as high as 83%.⁸  Therefore, we believe our findings represent a true signal of inherent CAR T-cell resistance in T/HRLBCL, likely owing to the unique immune environment that defines this disease. Indeed, consistent with a recent publication,¹  we observed ubiquitous expression of PD-L1 on tumor-associated macrophages, PD-L1 gene alterations within malignant B cells, and robust PD-1 expression among T cells in the T/HRLBCL environment. Moreover, we identified striking PD-1 upregulation on peripheral blood CAR T cells from T/HRLBCL patients over time. Thus, it is interesting to speculate that PD-1/PD-L1 interactions not only promote the dysfunction of endogenous tumor-reactive T cells in T/HRLBCL as previously suggested,¹  but may also contribute to CAR T-cell resistance by inhibiting the function of adoptively transferred CAR T cells. However, as this study is limited by sample size, patient heterogeneity, and limited assessment of specimens following CAR T-cell therapy progression, our results should be considered hypothesis-generating.

For instance, we cannot entirely rule out CD19 antigen loss as a contributor to CAR T-cell therapy failure, as CD19 expression was not assessed in all T/HRLBCL cases at progression. However, CD19 loss is uncommon among lymphomas progressing early after CAR T-cell therapy, as was the case in our patients. High disease burden has also been linked with axi-cel CAR T-cell therapy failure in DLBCL,¹⁰  and, interestingly, the mean baseline MTV in T/HRLBCL patients included in this study (n = 7) was higher than that of DLBCL patients (n = 13) who achieved durable responses to CAR T-cell therapy at The University of Chicago (Figure 1L). Although not all T/HRLBCL patients exhibited high baseline MTV, disease burden may have also contributed to CAR T-cell therapy failure in a fraction of cases. Despite the noted limitations of this study, our observations are nevertheless important in alerting physicians to the preliminarily poor activity of CAR T-cell therapy in T/HRLBCL, and in stimulating the exploration of alternative treatments, such as PD-1 blockade therapy, in this disease.