Severe cytopenia after chimeric antigen receptor‐T cell driven by large granular lymphocytes and responsive to steroids

Summary

Immune effector cell‐associated hematotoxicity (ICAHT) is a common toxicity associated with an important morbidity after chimeric antigen receptor (CAR)‐T‐cell therapy. Multiple factors seem to be involved in the development of severe ICAHT, making its management difficult. Here, we report three cases of severe ICAHT after axicabtagene‐ciloleucel (axi‐cel) for diffuse large B‐cell lymphoma showing an expansion of large granular lymphocyte in the bone marrow with a CD3/CD57‐positive non‐CAR‐T immunophenotype. We show that it is possible to treat them with low‐dose steroids, obtaining a striking resolution of cytopenias with no deleterious impact on the underlying malignancy.

Within the past decade, chimeric antigen receptor (CAR)‐T‐cell therapy has been established as a key treatment for relapsed/refractory B‐cell malignancies. ¹ However, this breakthrough in terms of efficacy came with some specific toxicities such as cytokine release syndrome (CRS) and immune effector cell‐associated neurotoxicity syndrome (ICANS). ² , ³ In addition to CRS and ICANS, cytopenias are frequent after CAR‐T‐cell infusions. Rejeski et al. described three clinical phenotypes of cytotoxicity in that context: the ‘quick recovery’, the ‘intermittent recovery’ and the ‘aplastic’ ⁴ phenotype. The latter tends to be more severe, prolonged, and resistant to growth factors, thus it is associated with higher non‐relapse mortality. ⁵ , ⁶ These unique features led to the concept of immune effector cell‐associated hematotoxicity (ICAHT), ⁷ opening a new field of research into its pathogenesis.

Multiple factors seem to be involved in the development of severe ICAHT. For example, an inflammatory state with a profound modification of the cytokine environment has been described at a systemic and local bone marrow level. ⁵ , ⁸ , ⁹ , ¹⁰ Moreover, pre‐infusion characteristics of the bone marrow can influence the risk of severe ICAHT, like pre‐existing clonal haematopoiesis and haematopoietic stem cell (HSC) reserve. ¹¹

Recently, prolonged cytopenias following CD19 CAR‐T‐cell therapy have been linked in some patients to an oligoclonal expansion of CD8 T cells in the bone marrow. This subset showed different characteristics from CAR‐T cells and expressed a ‘large granular lymphocyte (LGL) like’ phenotype (CD57+GranzymB+) ¹² as well as a high expression of CXCR1. Interestingly, the MD Anderson group was able to show an interferon (IFN)‐γ signature on that oligoclonal population, thereby postulating that prolonged ICAHT was driven by IFN‐induced HSC dysfunction. ¹³

Here we report three cases of severe ICAHT after axicabtagene‐ciloleucel (axi‐cel) in whom we demonstrated an oligoclonal expansion of ‘LGL like’ T cells in the bone marrow. Written informed consent was obtained from each patient. The study was approved by the local institutional review board. We treated them with low‐dose steroids, obtaining a striking resolution of cytopenias.

CASE 1

A 71‐year‐old woman with stage III diffuse large B‐cell lymphoma (DLBCL) relapsed 3 months after a first line with rituximab, cyclophosphamide, adriamycin, vincristine, and prednisone. She was referred to our institution to receive axi‐cel in the second line. A combination of rituximab, dexamethasone, cytarabine, and carboplatin was administered as bridging therapy, but she was progressive before lymphodepletion with fludarabine and cyclophosphamide. The CAR‐HEMATOX score was low (=1) with an absolute neutrophil count (ANC) of 2.05 G/L, haemoglobin 9.5 g/dL, and platelets 278 G/L.

Axi‐cel infusion was complicated with a grade 2 CRS treated with two doses of tocilizumab on day +3 and +5. No ICANS occurred. The ANC decreased below 0.50 G/L on day +3 and recovered above 0.50 G/L on day +13 (Figure 1).

FIGURE 1.

(A) Neutrophil and platelet curves (case 1). CAR‐T, CAR‐T‐cell infusion; CS, corticosteroids; LD, lymphodepletion; TOCI, tocilizumab. (B–D) Bone marrow aspirate smears showing large granular lymphocytes respectively, for cases 1, 2, and 3. (E) Bone marrow immunophenotyping of case 1.

Following initial blood count recovery, the patient developed severe pancytopenia (ANC 0.02 G/L, Hb 8.7 g/dL, platelets 9 G/L) starting from day +19. Despite the use of granulocyte colony‐stimulating factor (G‐CSF), thrombopoietin (TPO) agonists, and erythropoietin (EPO), the ANC did not recover, and she was dependent on red blood cells and platelet transfusions.

After an extensive screening, no infectious cause, in particular viral, and no haemophagocytic syndrome were found (Table 1; Table S1). Bone marrow aspirate at day +34 was hypocellular and contained 43% lymphocytes with a morphological aspect of LGL. On immunophenotyping, lymphocytes represented 80% of bone marrow cells, 93% being T lymphocytes with 61% of CD3⁺CD8⁺CD57⁺. In contrast, anti‐CD19 CAR‐T cells represented 0.9% of T lymphocytes in the bone marrow, and were not detected in the peripheral blood. An oligoclonal T‐cell population was also detected in the bone marrow by T‐cell receptor (TCR) sequencing. Next generation sequencing found a DNMT3A variant without any mutation on STAT genes.

TABLE 1.

Summary of the three cases.

Patient no.	1	2	3
Age (years)	71	81	38
Gender	F	M	M
Lymphoma
Subtype	DLBCL	DLBCL	DLBCL
Stage	III	IV	III
CAR‐HEMATOX score	1	1	3 a
H‐score during aplasia	Probability <1%	Probability <1%	Probability <1%
First‐line regimen/bridge	R CHOP/R DHAC	R CHOP/R GemOx	R CHOP/R ICE
Radiotherapy	No	No	No
Auto‐HSCT	No	No	No
Lymphodepletion
Fludarabine × 3 days	30 mg/m2	30 mg/m2	30 mg/m2
Cyclophosphamide × 3 days	500 mg/m2	500 mg/m2	500 mg/m2
CAR‐T product	Axi‐cel	Axi‐cel	Axi‐cel
CRS max grade	2	2	1
ICANS max grade	0	1	0
Tocilizumab administered	8 mg/kg × 2	8 mg/kg × 2	8 mg/kg × 2
Dexamethasone administered	0	10 mg × 2	0
Anakinra administered	0	0	0
PET response day +30	CR	CR	PD
PET response last follow‐up	CR	CR (month +6)	PR (month +6)
Blood immunophenotyping
Lymphocyte count (G/L)	0.570 (77% of all leucocytes)	0.300 (52% of all leucocytes)	0.230 (76% of all leucocytes)
Characteristics	55% CD3+CD57+ 0% Anti‐CD19 CAR‐T	60% CD3+CD8+CD57+ 3.8% Anti‐CD19 CAR‐T	NA 0.9% Anti‐CD19 CAR‐T
Bonne marrow aspiration
Day (after CAR‐T infusion)	Day+34	Day+30	Day+42
Cytology	Hypocellular. 43% LGL	Hypocellular. 11% LGL	Hypocellular. 60% LGL
Leucocyte immunophenotyping	70% Lymphocytes: 57% CD3+CD8+CD57+, CD7 low; 0.9% anti‐CD19 CAR‐T	15% Lymphocytes: 33% CD3+CD57+, CD7 low; 0% anti‐CD19 CAR‐T	42% Lymphocytes: 45% CD3+CD8+CD57+, CD7 low; 1.8% anti‐CD19 CAR‐T
T‐cell clonality	Oligoclonal T‐cell population	Oligoclonal T‐cell population	NA
Lymphoid NGS panel	No mutation detected	NA	NA
Myeloid NGS panel	DNMT3A variant	NA	NA
Neutrophil count (day after CAR‐T infusion)
<0.5 G/L	Day +19	Day +21	Day +23
>0.5 G/L	Day +41	Day +36	Day +48
Start of steroids (day/ANC)	Day +36/0.030 G/L	Day +34/0.490 G/L	Day +43/0.020 G/L
Corticosteroid
Dose (mg/kg)	0.5	0.5	0.5
Days from introduction to ANC >0.5 G/L	4	2	4
Duration (days)	43	36	15

Abbreviations: ANC, absolute neutrophil count; CAR, chimeric antigen receptor; CR, complete remission; CRS, cytokine release syndrome; F, female; HSCT, haematopoietic stem cell transplantation; ICANS, immune effector cell‐associated neurotoxicity syndrome; LGL, large granular lymphocyte; M, male; NGS, next‐generation sequencing; PD, progressive disease; PET, positron emission tomography; PR, partial remission.

^a ANC <1.2 G/L, ferritin 650–2000 ng/mL, Hb <9 g/dL.

Methylprednisolone was initiated at 0.5 mg/kg on day +37. The ANC increased above 500/μL by day +4 and platelets >30 G/L by day +7 after steroid introduction without a requirement for transfusion (Figure 1; Table S2). Importantly, we find a significant reduction of LGL absolute count in the peripheral blood from 0.398 G/L at steroid initiation to 0.2 G/L at day 7. All growth factors were stopped at day +43. Corticosteroid tapering was initiated after 7 days of treatment, and it was stopped at day +80 after CAR‐T‐cell infusion.

Every positron emission tomography‐computed tomography (PET‐CT) after CAR‐T cells showed a complete remission (CR) and at 6 months she had no significant cytopenia.

CASE 2 AND 3

Case 2 was an 81‐year‐old man with stage IV DLBCL (Table 1) with no marrow involvement treated with axi‐cel in second line. He developed an aplastic ICAHT similar to case 1. Bone marrow aspiration was hypocellular with 11% lymphocytes (Figure 1). Flow cytometry revealed a majority of T cells with a significant proportion of CD3⁺CD57⁺ LGL. An oligoclonal expansion of T cells was confirmed by TCR sequencing. After 13 days of severe aplasia despite the use of G‐CSF, EPO, and TPO agonists, the patient needed daily platelet transfusion and had low ANC. We then started methylprednisolone. After 2 days, cytopenia improved quickly without any need for transfusion and we were able to taper steroids (Figure S2A). At the last follow‐up, the patient had no severe cytopenia and remained in CR.

Case 3 was a 38‐year‐old man with stage III DLBCL who received axi‐cel in the second line. He developed severe ICAHT non‐responsive to growth factors (G‐CSF, EPO, and TPO agonists) on day +23 after CAR‐T‐cell infusion. At that time, the patient had persistent adenopathy and a progressive disease on the day +30 PET‐CT.

Bone marrow showed similar features to case 1 and 2 without lymphoma cells (Figure 1). Due to this severe cytopenias, we were not able to start any salvage therapy and so we started methylprednisolone. His blood count improved in a few days allowing the administration of glofitamab (Figure S2B). At the last follow‐up, he had no cytopenia and was in partial remission.

DISCUSSION

We describe three cases of severe aplastic ICAHT after treatment with axi‐cel for DLBCL. All patients had LGL in their bone marrow visible by optical microscopy (Figure 1). More importantly, in all cases, we were able to confirm by flow cytometry the presence of an activated T‐cell population different from anti‐CD19 CAR‐T cells expressing CD3⁺CD57⁺ and mostly CD8⁺ (Figure 1; Figure S1). In cases 1 and 2, we were able to demonstrate an oligoclonal expansion of T cells. Rejeski et al. ¹² had already reported a similar case, but the underlying disease (i.e. chronic lymphocytic leukaemia/Richter' syndrome with bone marrow involvement) made difficult any extrapolation.

Recently, in a large cohort of patients suffering from ICAHT, Strati et al. highlighted a frequent pattern of T‐cell oligoclonal infiltration of the bone marrow. Those cells were CD8⁺Granzym⁺CXCR1^hi and expressed a high IFN‐γ signature. ¹³ In our cases, we hypothesized that aplasia was partly induced by abnormal T‐cell expansion and cytokine production in the bone marrow, thereby we introduced steroids seeking a positive immunomodulating effect. In all three cases, aplasia resolved and patients became transfusion independent within a few days after steroids initiation, then we were able to completely taper steroids without relapse of ICAHT. Of note, we cannot exclude that growth factors contributed to that recovery. In a recent Chinese retrospective study, low‐dose steroids appeared safe and efficient in the context of prolonged ICAHT after an anti‐CD19 CAR‐T‐cell treatment for acute lymphoblastic leukaemia. ¹⁴ In that recent work, there was no safety signal or excess of relapse. However, prospective data are warranted to recommend a wider use of steroids. Given the CD28 co‐stimulatory domain of axi‐cel and the lesser importance of CAR‐T‐cell persistence in DLBCL, we assumed that low‐dose steroids would be safe. In fact, cases 1 and 2 were still in CR after 6 months. Case 3 was already showing early CAR‐T‐cell failure when we started steroids, but interestingly, the quick haematological recovery that it induced allowed an effective salvage with a bispecific antibody.

Unfortunately, we were not able to evaluate CXCR1 expression or cytokine production like IFN‐γ. In fact, whilst we show in one patient a significant decrease by day 7 after steroid initiation, additional analysis of LGL and cytokine kinetics before and after steroids will be important in the newly treated patients to evaluate if, in addition to the decrease in absolute number of LGL, steroid also induce a modification of their cytokine production. We also acknowledge that the LGL population could be a bystander and that the causal relationship between clonal T‐cell expansion and aplastic ICAHT remains uncertain. Nevertheless, this is the first clinical report using daily practice techniques to detect ‘LGL like’ expansion in the bone marrow, driving an effective treatment with low‐dose steroids for aplastic ICAHT.

AUTHOR CONTRIBUTIONS

AC: wrote the original draft, clinically involved for the three cases. AM: wrote case descriptions and figures with AC, clinically involved. AB: contributed to figures, correction of the original draft, clinically involved. LS: flow cytometry interpretation, correction. LR: correction, clinically involved. EC: correction, clinically involved. EB: correction, clinically involved. NS: correction, clinically involved. ZM: correction, clinically involved. CS: correction, clinically involved. MM: final correction. FM: correction, insights on translational aspects, clinically involved.

CONFLICT OF INTEREST STATEMENT

AC reports lecture honoraria from Takeda and congress invitations from Kite Pharma, MSD. LS reports a congress invitation from Novartis. MM reports grants and lecture honoraria from Janssen, Sanofi, Maat Pharma, and JAZZ Pharmaceuticals, lecture honoraria from Celgene, Amgen, BMS, Takeda, and Pfizer, grants from Roche, all outside the submitted work. FM reports lecture honoraria from Therakos/Mallinckrodt, BMS, MSD, Sanofi, Novartis, Astra Zeneca, and JAZZ Pharmaceuticals, all outside the submitted work. The other authors declare no competing financial interests.

ETHICS APPROVAL STATEMENT

The study was approved by the local institutional review board.

Supporting information

Data S1.

Capes A, Morin A, Banet A, Suner L, Ricard L, Corre E, et al. Severe cytopenia after chimeric antigen receptor‐T cells driven by large granular lymphocytes and responsive to steroids. Br J Haematol. 2025;206(1):180–185. 10.1111/bjh.19822

DATA AVAILABILITY STATEMENT

All data supporting our finding are already reported in the manuscript.