Oligoclonal T-cell expansion in a patient with bone marrow failure after CD19 CAR-T therapy for Richter-transformed DLBCL

TO THE EDITOR:

CD19-directed CAR T-cell therapy (CAR-T) has emerged as a promising treatment modality for multiple B-cell malignancies.¹ However, its utility is hampered by a unique toxicity profile that classically includes cytokine release syndrome (CRS) and neurotoxicity. Hematological toxicity represents a vexing and common side effect.^2-5 Cytopenia can occur long after lymphodepleting chemotherapy and resolution of CRS and is often prolonged and biphasic. Severe bone marrow (BM) failure has been described, which can predispose to severe infections and high non-relapse mortality.^6-8 Overall, the underlying pathomechanisms of CAR T-cell–related hematotoxicity remain poorly understood.

We present the case of a 57-year-old patient who received tisagenlecleucel for diffuse large B-cell lymphoma (DLBCL) transformed from chronic lymphocytic leukemia (CLL; Richter’s Syndrome). Disease manifestations included multilevel lymphadenopathy and extranodal disease of the duodenum, pleurae, and BM (70% infiltration). A combination of venetoclax, cyclophosphamide, and dexamethasone was applied as bridging therapy. The clinical course was notable for severe CRS (grade 3) and several episodes of infection (Figure 1A). Strikingly, the patient developed profound and prolonged neutropenia and transfusion dependency (Figure 1A), despite the application of growth factor (eg, granulocyte colony-stimulating factor), thrombopoietin agonists (eg, eltrombopag), and anti-cytokine therapy (eg, tocilizumab and anakinra). On ¹⁸fluorodeoxyglucose positron emission tomographic/computed tomographic imaging, a partial response was noted according to Lugano criteria (Figure 1B).⁹ Anti-lymphoma therapy was escalated with pembrolizumab, deepening the radiographic response (supplemental Figure 3, available on the Blood Web site). Eventually, neutrophil count recovery occurred around day 50, coinciding with decreased BM infiltration of the underlying LBCL (supplemental Figure 1). However, the patient ultimately relapsed and died of pneuomogenic sepsis on day 165. We performed further studies to understand the nature of CAR T-cell–related BM failure.

We performed sequential CAR T-cell immunoprofiling using a 2-step staining method with a biotinylated CD19 protein.¹⁰ We performed flow cytometry utilizing a panel of surface markers to profile peripheral blood mononuclear cells (PBMCs) in pre-CAR T- (day −8) and post-CAR T-cell (day +31) therapy samples, BM samples (days +27 and +47), and in a healthy donor (HD) control. For serum cytokine profiling, samples were collected serially across 3 time points (days 0, 4, and 14) and explored using a 92-protein multiplex proximity extension assay (Immuno-Oncology Panel; Olink Bioscience). We performed coupled single-cell RNA sequencing (scRNA-seq) and single-cell T-cell receptor sequencing (scTCR-seq) on pretreatment and posttreatment PBMC samples (days −8 and +31) and a HD control. Barcoding was performed using TotalSeq C hash-tag oligonucleotides (Biolegend). Stained cells were analyzed with the Chromium Single-Cell V(D)J Reagent Kit (10x Genomics). Pooled scRNA-seq libraries were sequenced (Illumina Novaseq), analyzed with the Seurat alignment algorithm, and visualized in t-distributed stochastic neighbor embedding plots (supplemental Materials). Gene Set Enrichment Analysis (GSEA) and Genomatix were used to interpret involved pathways and functions of differentially expressed genes.

On BM biopsy, we observed persistent DLBCL infiltration with new upregulation of PD-L1 and a lymphocytic infiltrate consisting of CD8⁺CD57⁺ T cells (Figure 1C; supplemental Figure 1). Furthermore, we noted migration of CAR T-cells to the BM (supplemental Figure 2). Reflecting cytokine storm and coincident sepsis, the patient displayed high serum levels of C-reactive protein, procalcitonin, and interleukin-6 during the first 2 weeks after CAR T-cell infusion (Figure 1D). The observed cytokine signature was consistent with acquired aplastic anemia, including downregulation of EGF and CD40-L (Figure 1E).¹¹ Notably, we observed a four-fold downregulation of the hematopoietic growth factor interleukin-7,¹² whereas soluble interferon-γ, PD-L1, and FAS ligand were upregulated.¹³ Peak CAR T-cell expansion was detected on day 21 (Figure 1F). Flow cytometry showed a shift from central memory (CM; 28% to 2.6%) to effector memory (EM; 72% to 88%) CD4⁺ T-cells after CAR T-cell infusion (supplemental Figure 4). In contrast, the CD8⁺ T-cell population displayed lower EM, but increased terminal effector (TEF) cells, suggesting EM-to-TEF differentiation. The exhaustion marker PD-1 was markedly increased after treatment, especially in CD8⁺ T cells (supplemental Figure 5). CD57 expression, usually a marker of T-cell activation and clonal expansion in aplastic anemia¹⁴ and T-cell large granular lymphocytic leukemia (T-LGLL),¹⁵ was increased in both CD4⁺ and CD8⁺ T cells after treatment, particularly in CD8⁺ T cells (Figure 2A; supplemental Figure 6). CD57 expression was similar between both CAR and non-CAR–bearing CD4⁺ and CD8⁺ T cells, suggesting equivalent clonal expansion in both populations (Figure 2B; supplemental Figure 6).

After computational assignment of single cells,¹⁶ consistency was observed between gene expression–defined cell types and established surface protein expression profiles (Figure 2C). GSEA revealed upregulation of genes related to immune- and cell-cycling pathways in T cells (supplemental Figure 7A). In myeloid cells, immune response and inflammation-related pathways were upregulated, whereas genes related to antigen processing and presentation were downregulated (supplemental Figure 7B). Pathway analysis of top differentially expressed genes mirrored these findings (supplemental Figure 8). Although the pretreatment sample displayed multiclonal TCR usage, the posttreatment sample was oligoclonal, with a large, dominating T-cell clone (Figure 2E). Concomitantly, the Gini index of T-cell clone size distribution dramatically increased after CAR T-cell infusion, reaching a higher level than HD controls and similar to patients with T-LGLL with obvious T-cell clonal expansion (Figure 2F).¹⁷ Among the top 200 clones, only 14 were shared between pretreatment and posttreatment samples, indicating post–CAR T-cell clonal drift (Figure 2D). Most of the top 10 pretreatment clones declined in size, and only 1 post–CAR T-cell clone originated from the top 10 clones of the paired pretreatment sample (clone 4; Figure 2G). Of note, the dominant post–CAR T-cell clone originated from a pretreatment clone ranking of 11, with the relative clone size dramatically increasing from 1.3% to 76%. Both CAR and non–CAR-bearing T cells constituted the dominant clone. To understand the mechanism of dynamic clonal dominance, we compared gene expression profiles of the dominant clone versus all the other T-cell clones. In the dominant clone, we found elevated cell-cycle–related gene expression, and lower expression of genes involved in cell apoptosis, immune response, and inflammation (Figure 2H-I). Among the 500 T cells with highest CAR expression, we found decreased expression of genes involved in STAT signaling (P < .001) and inflammatory response (P = .03; supplemental Figure 9).

Our case illustrates the multifactorial nature of CAR T-cell–related hyporegenerative BM failure. First, the patient presented with underlying BM infiltration, affecting hematopoietic stem and progenitor cell function at baseline⁵ and predisposing for local inflammatory processes within the BM niche.^4,18 Indeed, lymphocytic infiltrates including both CAR and non-CAR T cells were detected in the BM, perhaps a manifestation of “innocent bystander” killing.^19,20 Concomitantly, neutropenia resolved after effective lymphoma treatment with pembrolizumab, which corresponded with decreasing BM infiltration. Second, the patient developed severe CRS and the associated cytokine patterns have been implicated in CAR-T–related hematotoxicity.^2,4 Finally, we describe oligoclonal T-cell expansion, with clonality levels rivalling those of a reference T-LGLL population.¹⁷ Among the similarities to acquired aplastic anemia were effector memory CD8⁺CD57⁺ T-cell expansion and T-cell receptor oligoclonality.^14,21 These findings were mirrored on a single-cell level, revealing dominant clones both in CAR and non–CAR-bearing T-cell populations. Predisposed subclones appear to have been present before product manufacture. Although the stimulus for clonal (CAR) T-cell expansion ultimately remains unclear, high-grade CRS and an inflammatory micromilieu due to BM infiltration represent potential culprits. Patients with underlying CLL may be at particular risk for clonal expansion phenomena,^22,23 and the disease carries its own risk of autoimmune neutropenia.²⁴ B-cell–depleting therapies such as rituximab have also been linked to clonal expansion and neutropenia, most likely related to T-cell imbalances facilitated by diminished T-cell/B-cell interactions.^13,25 Importantly, the unique constellation of clinical features in our patient (eg, history of CLL, marrow involvement, high-grade CRS) makes it difficult to broadly extrapolate our findings to the majority of patients presenting with prolonged cytopenia after CAR T-cell infusion. Still, these hypothesis-generating findings introduce oligoclonal T-cell expansion as a potential contributing pathomechanism of CAR-T–related hematotoxicity, inviting future study.

This study was performed with institutional review board approval (LMU Project 19-817) in accordance with the 1964 Declaration of Helsinki and its later amendments. The patient provided informed, written consent.