Clinical outcomes of chimeric antigen receptor T-cell therapy following autologous hematopoietic stem cell transplantation in 38 patients with refractory/relapsed primary or secondary central nervous system lymphoma

Abstract

Background

Several reports have indicated that chimeric antigen receptor (CAR) T-cell therapy following autologous hematopoietic stem cell transplantation (ASCT) is a promising strategy for refractory/relapsed (r/r) central nervous system lymphoma (CNSL), but the number of reported cases is limited.

Methods

The cohort in this retrospective study consisted of 38 patients with r/r CNSL who received CAR T-cell therapy following ASCT at our center between January 2019 and April 2024. Group comparisons of continuous variables were tested using the unpaired Student’s t-test or the Mann–Whitney U-test, while categorical variables were analyzed using Fisher's exact test. The Kaplan–Meier method was employed to estimate survival curves, and group comparisons were performed using the log-rank test.

Results

The cohort comprised 38 patients with r/r CNSL, all of whom had active CNS involvement. After therapy, the best overall response rate (ORR) of all patients was 78.9%. Subgroup analysis found that a lower ORR was observed in patients with lactate dehydrogenase levels above the upper limit of normal (60.0% vs. 91.3%, P = 0.039). With a median follow-up of 37.5 months, the estimated 1-year overall survival (OS) and progression-free survival (PFS) rates were 72.8% and 57.4%, respectively. The risk factors associated with PFS was no response to current therapy (adjusted hazard ratio: 22.87, P < 0.001). The incidence rates of severe cytokine release syndrome and immune effector cell-associated neurotoxicity syndrome were both 13.2%. Among the 25 patients with secondary CNSL (SCNSL), the best ORRs were 91.7% for those with CNS lesions only and 61.5% for those with CNS and systemic lesions (P = 0.160), while the estimated 1-year PFS rates were 83.3% and 38.5%, respectively (P = 0.030).

Conclusions

CAR T-cell therapy following ASCT shows promising efficacy for r/r CNSL patients. Besides, SCNSL patients with CNS and systemic lesions have inferior treatment efficacy compared to those with CNS lesions only.

Electronic supplementary material

The online version of this article (10.1007/s00262-024-03855-7) contains supplementary material, which is available to authorized users.

Background

Central nervous system lymphoma (CNSL) is typically classified into primary and secondary subtypes. Primary CNSL (PCNSL) refers to lymphoma lesions that occur in the brain tissue, meninges, eyes (vitreoretinal space), and spinal cord, generally without the involvement of areas outside the central nervous system (CNS) [1]. The overall incidence of PCNSL is approximately 0.5/100,000, showing an increasing trend over time [2, 3]. Secondary CNSL (SCNSL) represents metastatic involvement of the CNS in systemic lymphomas, especially in highly aggressive lymphomas [4]. A retrospective study indicated that 19% of patients with Burkitt’s lymphoma had CNS involvement at initial diagnosis, and 6% experienced CNS relapse after disease remission [5, 6]. Additionally, the overall probability of CNS involvement in patients with systemic diffuse large B-cell lymphoma (DLBCL) is approximately 5%, but it can reach up to 15% in high-risk patients [7]. Traditionally, CNSL is associated with an extremely poor long-term prognosis, and salvage treatment options and efficacy for patients with refractory or relapsed (r/r) disease are very limited. Following a positive response to high-dose chemotherapy, favorable long-term survival can be achieved in patients with r/r CNSL who undergo autologous hematopoietic stem cell transplantation (ASCT) [8]. Nevertheless, for patients resistant to high-dose chemotherapy, the challenge lies in determining appropriate salvage treatment options.

Chimeric antigen receptor (CAR) T-cell therapy arms the patient’s immune system and immune cells to specifically recognize and target malignant tumor cells. Preclinical and clinical reports have indicated that CAR T-cells have intracranial activity and potential therapeutic effects on r/r CNSL with controllable cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) [9–11]. However, the issues of therapeutic resistance and relapse remain unresolved. Ghafouri et al. reported that the median progression-free survival (PFS) of patients with r/r CNSL treated with CAR T-cells alone was only 4.5 months [12]. Moreover, a meta-analysis showed that patients with r/r CNSL who achieved remission using CAR T-cell therapy had a relapse rate of up to 38% [13]. Therefore, to overcome the issue of poor long-term PFS, experts are resorting to combination treatment strategies involving CAR T-cell therapy, such as ASCT, nivolumab, and lenalidomide. In a cohort of 25 patients with r/r large B-cell lymphoma who underwent combined therapy with ASCT and CAR T-cells, the overall response rate (ORR) was 92.0%, with a 2-years PFS rate of 62.3%. The patients receiving the combined therapy had enhanced CD19 CAR T-cells expansion and reduced long-term exhaustion formation compared to patients receiving CD19 CAR T-cell monotherapy [14]. Xu et al. found that eight patients with r/r CNSL who underwent ASCT combined with CAR T-cell therapy had a more satisfactory clinical prognosis compared to nine patients with r/r CNSL who received CAR T-cell therapy alone [15]. Similarly, our earlier report with a small sample described the clinical outcomes of 11 patients with r/r CNSL who received CD19 and CD22 CAR T-cell therapy following ASCT. This treatment strategy achieved an impressive 1-year PFS rate of 74.6% [16]. These findings suggest that CAR T-cell therapy following ASCT is a promising treatment strategy and could be considered an alternative option for patients with r/r CNSL. However, the limited sample size raises concerns about the reliability and stability of the results and restricts the feasibility of subgroup analysis.

In this report, we share the clinical outcomes of 38 patients with r/r CNSL treated using CAR T-cell therapy following ASCT, and analyze the risk factors associated with treatment efficacy.

Methods

Study design

This was a retrospective study, focusing on analyzing the clinical outcomes of CAR T-cell therapy following ASCT in 38 patients with r/r CNSL from January 1, 2019, to April 30, 2024. All patients with r/r CNSL were confirmed to have residual active CNS involvement by pathology, magnetic resonance imaging (MRI), or cerebrospinal fluid (CSF) examination at the latest assessment before treatment, regardless of the presence of systemic lesions. Patients with previous primary or secondary CNS involvement who achieved complete remission through other therapy strategies but opted to receive the mentioned regimen for further consolidation were excluded from this study. Twenty-nine of the eligible patients participated in a clinical trial (ChiCTR-OPN-16009847) conducted at our center, while the remaining nine patients received approved CD19 CAR T-cells products following ASCT. The mentioned clinical trial was approved by the Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, and conducted in accordance with the principles of the Declaration of Helsinki. Informed consent was obtained from all enrolled patients in the clinical trial.

Therapy procedure

A detailed procedure of the therapy can be found in a previous study [17]. In summary, all patients received autologous hematopoietic stem cell (HSC) collection and peripheral blood mononuclear cell apheresis. Approved or clinical trial CAR T-cells products were subsequently prepared. The approved CAR T-cells products included Relmacabtagene Autoleucel (relma-cel) and Axicabtagene ciloleucel (axi-cel), while the clinical trial products were provided by Wuhan Bio-Raid Biotechnology Co., Ltd. The clinical trial utilized third-generation CD19 or CD22 CAR lentiviral vectors, incorporating two costimulatory domains (CD28 and 4-1BB) and a CD3-zeta activation domain. Bridging therapy, if necessary, was allowed during CAR T-cells preparation, the selection of bridging therapy was based on institutional experience, the patient’s condition, and prior therapy experiences. Patients then received HSC infusions after undergoing a conditioning regimen (the BEAM-based protocol and thiotepa-based protocol) and subsequently received an infusion of approved CD19 CAR T-cells alone or a “cocktail” infusion of clinical trial CD19 and CD22 CAR T-cells within 2–6 days.

Data collections and assessment of response and toxicity

Baseline characteristics, serum inflammatory markers, treatment-related adverse events, radiographic and CSF findings, and clinical outcomes were collected. Baseline assessments were performed prior to treatment initiation. The objective response was evaluated according to the International PCNSL Collaborative Group Response Criteria and Lugano Response Criteria for B-Cell Lymphoma [18, 19]. The objective response was graded as complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD). Therapy response was assessed using MRI with and without contrast enhancement, CSF detection, and positron emission tomography/computed tomography (PET/CT, for SCNSL patients with systemic involvement or PCNSL patients with clinical indications) at 1, 3, and 6 months, and 1 year after treatment.

PFS was defined as the time interval from ASCT to recurrence, progression, death, or the last follow-up point. Overall survival (OS) was defined as the time from ASCT to death, or the follow-up endpoint. The last follow-up point was April 30, 2024. SCNSL patients experienced CNS or systemic relapse or progression, defined as disease progression. CRS and ICANS assessments and grading were repeated according to the American Society for Transplantation and Cellular Therapy (ASTCT) CAR T-cell therapy toxicity criteria [20]. ICANS refers to new or worsening neurological signs and symptoms occurring within 28 days of CAR T-cells infusion. Other adverse events occurring within 30 days were assessed and graded according to the Common Terminology Criteria for Adverse Events, Version 5.0.

Statistical analysis

Continuous variables were expressed as means ± standard deviation (SD) or median (range). Differences between groups were tested using the unpaired Student’s t-test for continuous variables with Gaussian distributions and the Mann–Whitney U-test for continuous variables with non-Gaussian distributions. The categorical variables were described as absolute numbers and percentages, and group comparisons were performed using Fisher's exact tests. The lower and upper limits of the 95% confidence interval (CI) of the proportions were calculated using the Wilson method without correction for continuity. The median follow-up time was calculated for surviving patients. The probability rates of PFS and OS were analyzed using the Kaplan–Meier method, and group comparisons were performed using the log-rank test. Cox regression was performed to test the association between a covariate of interest and survival; and factors with a P-value < 0.1 in the univariate analysis were included in the multivariate analysis. A P-value of < 0.05 indicated statistically significant differences. All statistical analyses were performed, and charts were generated using IBM SPSS Statistics for Windows, Version 26.0, GraphPad Prism 8.0, and Adobe Illustrator software.

Results

Baseline characteristics

This study included 38 r/r CNSL patients with active CNS involvement who received CAR T-cell therapy following ASCT at our center between January 1, 2019 and April 30, 2024. This cohort comprised 13 patients with PCNSL and 25 patients with SCNSL. The median age of the enrolled patients (23 male and 15 female) was 47 years (range: 19–67). The pathology of most patients was DLBCL, except for one case of Burkitt’s lymphoma and one case of intravascular large B-cell lymphoma. Baseline assessments were performed prior to treatment initiation. Neurological symptoms were exhibited by 25 patients (65.8%). Parenchymal involvement was observed in 29 cases (76.3%), leptomeningeal involvement in one case (2.6%), spinal cord involvement in two cases (5.3%), and ≥ 2 the mentioned lesions were present in six patients (15.8%). Nine patients (23.7%) were confirmed by MRI detection to have CNS lesions ≥ 1 cm. Thirteen of 25 (52.0%) SCNSL patients accompanied with CNS and systemic involvement, while no PCNSL patients had systemic lesions. Patients priorly experienced a median of five (range: 2–8) therapy lines, with two and three of these patients having previously received ASCT and CAR T-cell therapy, respectively. 89.5% of patients had disease in SD or PD status. During CAR T-cells preparation, 21 patients (55.3%) received bridging therapy to reduce the disease burden. The therapy regimen consisted of Bruton’s tyrosine kinase (BTK) inhibitors, high-dose methotrexate (HD-MTX)/high-dose cytarabine (HD-Ara-C) based chemotherapy, temozolomide, rituximab alone, and WBRT, with several patients having accepted two of the above options. The conditioning regimen included the BEAM-based protocol (28.9%) and thiotepa-based protocol (71.1%). After HSC infusion, nine patients received an approved CD19 CAR T-cells products (Relma-cel: five cases; Axi-cel: four cases) infusion alone, while the remaining 29 patients received a “cocktail” infusion of clinical trial CD19 and CD22 CAR T-cells products. The median infusion doses of CD19 CAR T-cells and CD22 CAR T-cells were 3.2 × 10⁶/kg and 4.0 × 10⁶/kg, respectively. Additional baseline characteristics are summarized in Table 1.

Table 1.

Baseline characteristics of eligible patients

Characteristics	All patients (N = 38)
Age, median (range)	47 (19–67)
Male/Female, n (%)	23/15 (60.5/39.5)
Diagnosis
PCNSL	13 (34.2)
SCNSL	25 (65.8)
Pathology, n (%)
DLBCL, GCB	11 (28.9)
DLBCL, non-GCB	25 (65.8)
Others	2 (5.3)
ECOG, n (%)
0–1	13 (34.2)
≥ 2	25 (65.8)
LDH, median (range)	209 (123–428)
Neurological symptoms, n (%)	25 (65.8)
CNS lesions, n (%)
Parenchymal	29 (76.3)
Leptomeningeal	1 (2.6)
Spinal cord	2 (5.3)
The mentioned lesions ≥ 2	6 (15.8)
CNS lesions ≥ 1 cm, n (%)	9 (23.7)
Systemic involvement in SCNSL (N = 25), n (%)	13 (52.0)
Disease status, n (%)
PR	4 (10.5)
SD	8 (21.1)
PD	26 (68.4)
Prior therapy lines, median (range)	5 (2–8)
Prior ASCT, n (%)	2 (5.3)
Prior CAR T-cell, n (%)	3 (7.9)
Relapsed disease, n (%)	17 (44.7)
Bridging therapy*, n (%)	21 (55.3)
Conditioning regimen, n (%)
TBC	27 (71.1)
BEAM	11 (28.9)
CAR T-cell products, n (%)
CD19 (approved)	9 (23.7)
CD19 and CD22 (clinical trial)	29 (76.3)
CAR T-cell dose (× 106/kg), median (range)
CD19 (N = 38)	3.2 (1.0–12.8)
CD22 (N = 29)	4.0 (1.1–8.9)

PCNSL: Primary Central Nervous System Lymphoma; SCNSL: Secondary Central Nervous System Lymphoma; DLBCL: Diffuse Large B-cell Lymphoma; GCB: Germinal Center B-cell like; Non-GCB: Non-Germinal Center B-cell like; ECOG: Eastern Cooperative Oncology Group; LDH: Lactate Dehydrogenase; PR: Partial Response; SD: Stable Disease; PD: Progressive Disease. ASCT: autologous hematopoietic stem cell transplantation; CAR: Chimeric Antigen Receptor; TBC: Thiotepa, Busulfan, Cyclophosphamide; BEAM: Carmustine, Etoposide, Cytarabine, Melphalan. Approved CD19 CAR T-cells included Relma-cel and Axi-cel, clinical trial CD19 and CD22 CAR T-cells were provided by Wuhan Bio-Raid Biotechnology Co., Ltd. *Bridging therapy: including BTK inhibitor, HD-MTX/HD-Ara-C based chemotherapy, Temozolomide, Rituximab alone, and WBRT; Several patients have accepted two of the above options

Objective response

Out of the 38 patients included in our study, 30 patients finally responded to the treatment, with 24 achieving CR and 6 achieving PR, while the remaining eight patients did not respond to the therapy. The best ORR and CRR for the enrolled cases were 78.9% (95%CI: 63.7%-88.9%) and 63.2% (95%CI: 47.3%-76.6%), respectively. As of April 30, 2024, the longest duration of response among 30 responsive patients has been nearly 58 months. After achieving a response, five patients (one CR patient and four PR patients) subsequently experienced relapse or disease progression within 1 year; the cumulative relapse or progression rate of the 30 responders was 18.1% (95% CI: 1.6%–49.3%). One patient received a “cocktail” infusion of CD19 and CD22 CAR T-cells as salvage therapy, but resulted in uncontrolled disease and developed bone marrow involvement. Another patient received salvage therapy with approved CD19 CAR T-cells (Relma-Cel) and had a complete remission of the disease, but the remission was not long-lasting and ultimately died of disease progression one year later.

Subgroups analysis was conducted according to the baseline variables. The ORR and CRR remained consistent across subgroups of diagnosis, gender, age, pathology, ECOG score, neurological symptoms, CNS lesions size, disease status, bridging therapy, conditioning regimen, CAR T-cells products, and Cmax of CAR19 transgene (Table 2 and Supplementary material 1). However, a lower ORR was observed in patients with LDH levels above the upper limit of normal (60.0% vs. 91.3%, P = 0.039). Patients with > 4 prior therapy lines had a lower CRR than patients with ≤ 4 prior therapy lines (45.0% vs. 83.3%, P = 0.020).

Table 2.

Subgroups analysis of ORR according to baseline variables

Variables	Patients no.	OR	ORR (%)	P value
Diagnosis				0.689
PCNSL	13	11	84.6
SCNSL	25	19	76.0
Gender				1.000
Male	23	18	78.3
Female	15	12	80.0
Age				0.106
≤ 50	22	15	68.2
> 50	16	15	93.8
Pathology				1.000
DLBCL, GCB	11	9	81.8
DLBCL, non-GCB	25	19	76.0
Others	2	2	100
ECOG				0.222
0–1	13	12	92.3
≥ 2	25	18	72.0
LDH level				0.039
Normal	23	21	91.3
> ULN	15	9	60.0
Neurological symptoms				1.000
Yes	25	20	80.0
No	13	10	76.9
CNS lesions ≥ 1 cm				1.000
Yes	9	7	77.8
No	29	23	79.3
Disease status				0.251
PR	4	3	75.0
SD	8	8	100
PD	26	19	73.1
Prior therapy lines				0.238
≤ 4	18	16	88.9
> 4	20	14	70.0
Bridging therapy				0.426
Yes	21	18	85.7
No	17	12	70.6
Conditioning regimen				1.000
TBC	27	21	77.8
BEAM	11	9	81.8
CAR T-cell products				0.650
CD19 (approved)	9	8	88.9
CD19 and CD22 (clinical trial)	29	22	75.9
Cmax of CAR19 transgene				0.232
High	19	17	89.5
Low	19	13	68.4

ORR: Overall Response Rate. The median value was used as cut off to distinguish high or low Cmax of CAR19 transgene. Group comparisons were performed using Fisher's exact tests, a two-tailed P-value of < 0.05 indicated statistically significant differences

Survival and correlation with clinical variables

As of April 30, 2024, the median follow-up time was 37.5 months. A total of 15 patients died: 12 due to lymphoma progression, two due to severe pulmonary infection, and one due to veno-occlusive disease and acute liver failure. The median OS of all patients was not reached, with an estimated 1-year OS rate of 72.8% (95% CI: 55.3%-84.4%). The median PFS of all patients was not reached, with an estimated 1-year PFS rate of 57.4% (95% CI: 40.1%-72.4%) (Fig. 1A). Group comparisons of OS and PFS were performed according to diagnosis (PCNSL vs. SCNSL), bridging therapy (yes vs. no), conditioning regimen (TBC vs. BEAM), CAR T-cells products (approved vs. clinical trial), and clinical response (OR vs. NR). The probability rates of PFS and OS were comparable between different groups, categorized by diagnosis, bridging therapy, conditioning regimen, and CAR T-cells products (Fig. 1B-E). The median OS and PFS of the responsive patients were not reached, while for the unresponsive patients were 5.9 and 2.5 months, respectively, both with a significant difference. The estimated 1-year OS rates of the responsive and unresponsive patients were 86.4% and 14.6%, respectively (P < 0.001) (Fig. 1F).

Fig. 1.

Survival curves. A The OS and PFS of all patients. The estimated 1-year OS and PFS rates were 72.8% and 57.4%, respectively. B Comparison of OS and PFS between PCNSL and SCNSL patients. C Comparison of OS and PFS between patients with and without bridging therapy, the median OS in patients with and without bridging therapy were unreached and 13.0 months (P = 0.054), while the median PFS were unreached and 7.6 months (P = 0.078), respectively. D Comparison of OS and PFS between different conditioning regimens. E Comparison of OS and PFS between different CAR T-cells products. Patients received CD19 and CD22 CAR T-cells “cocktail” infusion following ASCT were from a clinical trial conducted in our center; approved CD19 CAR T-cells products included relma-cel and axi-cel. F Comparison of OS and PFS between responsive and unresponsive patients. The median OS and PFS in responsive patients were not reached, but in unresponsive patients were 5.9 and 2.5 months, with significant difference (P < 0.001). The probability rates of PFS and OS were analyzed using the Kaplan–Meier method, and group comparisons were performed using the log-rank test. OS: overall survival; PFS: Progression-free survival; HSCs: Hematopoietic stem cells; PCNSL/SCNSL: Primary/Secondary central nervous system lymphoma; TBC: Thiotepa, Busulfan, and Cyclophosphamide; BEAM: Carmustine, Etoposide, Cytarabine, and Melphalan; OR: Objective response (CR + PR); NR: No response (SD + PD)

The Cox regression analysis to determine the clinical risk factors for PFS included indicators with a P-value ≤ 0.1 from the univariate analysis. Univariate  Cox regression analysis found a P-value of ≤ 0.1 for bridging therapy and clinical response (Fig. 2). After adjusting for multivariate COX regression analysis, the adjusted hazard ratio (HR) for bridging therapy patients versus no bridging therapy patients was 0.52 (95%CI: 0.18–1.47, P = 0.216), and the adjusted HR for no-responsive patients versus responsive patients was 22.87 (95%CI:5.75–91.06, P < 0.001). Thus, treatment resistance was an independent risk factor for PFS in patients with r/r CNSL receiving CAR T-cell therapy following ASCT.

Fig. 2.

 Univariate COX regression analysis for PFS. The P value and crude HR were calculated using univariate COX regression analysis. The dot plot represented as crude HR, and the error bar represented 95% confidence interval. Variables with P value ≤ 0.1 were used for subsequent multivariate  analysis. * DLBCL non-GCB and two patients with other pathology. The median value was used as cut off to distinguish low or high CD19 CAR T-cells number and Cmax of CAR19 transgene. HR: Hazard ratio; ECOG: Eastern Cooperative Oncology Group; ULN: Upper limit of normal; CRS: Cytokine release syndrome; ICANS: Immune effector cell-associated neurotoxicity syndrome

CRS and ICANS

After therapy, 37 experienced CRS of any grade, and 10 experienced ICANS of any grade, corresponding to incidence rates of 97.4% (37/38) and 26.3% (10/38), respectively. Regarding CRS severity, grades 1–2 accounted for 84.2% (32/38) of cases, while grade 3–4 accounted for 13.2% (5/38). For ICANS, grades 1–2 accounted for 13.2% (5/38) of the cases, while grades 3–4 accounted for 13.2% (5/38). Clinical manifestations of ICANS included cognitive and consciousness disorders, and seizures (Fig. 3A). The median onset time of neurological symptoms was day 5.5 after the first infusion of the CAR T-cells. These symptoms improved after receiving glucocorticoids and supportive therapy for a median duration of 6 days. No deaths occurred owing to severe ICANS. Other common adverse events occurring within 30 days were summarized in Supplementary material 2. All patients experienced severe cytopenia within 30 days. Of these, 92.1% experienced prolonged neutropenia, 84.2% experienced prolonged anemia, and 76.3% experienced prolonged thrombocytopenia. The incidence rates of severe forms of these conditions were 63.2%, 42.1%, and 44.7%, respectively.

Fig. 3.

CRS and ICANS A The incidence rates, severity, and symptoms of CRS and ICANS. B The correlation of inflammatory markers and CRS severity. C The correlation of inflammatory markers and ICANS. Each dot plot represents an individual, the lines are at the median with interquartile range (for indicators with  non-Gaussian distributions) or the mean with standard deviation (for indicators with Gaussian  distributions). *** indicates P < 0.001, * indicates P < 0.05, ns indicates no significance. IL-6: Interleukin-6; CRP: C-reactive protein; PCT: Procalcitonin

The inflammatory markers interleukin-6 (IL-6), ferritin, C-reactive protein (CRP), and procalcitonin (PCT) showed varying degrees of elevation after therapy. Correlation analysis of inflammatory markers and adverse events indicated that the peak levels of IL-6, ferritin, CRP, and PCT were significantly higher in patients with CRS grades 2–4 than in those with CRS grades 0–1 (P < 0.05). Additionally, the peak level of IL-6 was significantly higher in patients with any grade of ICANS than in those without ICANS (P < 0.05) (Fig. 3B-C).

Outcomes comparison of SCNSL with CNS lesions only and those with CNS and systemic lesions

Among the SCNSL group, 13 out of 25 patients had both CNS and systemic lesions. Baseline characteristics comparisons indicated that SCNSL patients with CNS and systemic lesions had a significantly higher percentage of ECOG scores ≥ 2 compared to those with CNS lesions only (92.3% vs. 41.7%, P = 0.011) (Table 3). Out of 12 SCNSL patients with CNS lesions only, ten cases achieved CR, and one case achieved PR. Among the 13 SCNSL patients with CNS and systemic lesions, six cases showed CR, and two cases showed PR. While the best ORR and CRR of SCNSL with CNS lesions only were higher than those with CNS and systemic lesions, this difference was not statistically significant (ORR: 91.7% vs. 61.5%, P = 0.160; CRR: 83.3% vs. 46.2%, P = 0.097) (Fig. 4A). All patients experienced any grade of CRS, the overall incidence rates of severe CRS in SCNSL with CNS lesions only and those with CNS and systemic lesions were 16.7% (2/12) and 0 (P = 0.220), respectively. The overall incidence rates of any grade of ICANS in SCNSL with CNS lesions only and those with CNS and systemic lesions were 8.3% (1/12) and 30.8% (4/13) (P = 0.322), respectively. One patient occurred severe ICANS in SCNSL with CNS and systemic lesions, while no patient occurred it in those with CNS lesions only.

Table 3.

Baseline characteristics comparison of SCNSL with CNS lesions only and those with CNS and systemic lesions

Characteristics	SCNSL (N = 25)	CNS lesions only (N = 12)	CNS and systemic lesions (N = 13)	P value
Age, median (range)	43 (19–67)	46 (19–67)	42 (23–67)	0.574
Gender, n (%)				0.411
Male	16 (64.0)	9 (75.0)	7 (53.8)
Female	9 (36.0)	3 (25.0)	6 (46.2)
Pathology, n (%)				0.145
DLBCL, GCB	8 (32.0)	5 (41.7)	3 (23.1)
DLBCL, non-GCB	15 (60.0)	5 (41.7)	10 (76.9)
Others	2 (8.0)	2 (16.7)	0
ECOG, n (%)				0.011
0–1	8 (32.0)	7 (58.3)	1 (7.7)
≥ 2	17 (68.0)	5 (41.7)	12 (92.3)
LDH, median (range)	207 (131–428)	214.5 (164–338)	195 (131–428)	0.611
Neurologic symptoms, n (%)	16 (64.0)	9 (75.0)	7 (53.8)	0.411
CNS lesions, n (%)				0.645
Parenchymal	17 (68.0)	9 (75.0)	8 (61.5)
Leptomeningeal	1(4.0)	0	1 (7.7)
Spinal cord	2 (8.0)	0	2 (15.4)
The mentioned lesions ≥ 2	5 (20.0)	3 (25.0)	2 (15.4)
CNS lesions ≥ 1 cm, n (%)	3 (12.0)	2 (16.7)	1 (7.7)	0.593
Disease status, n (%)				0.813
PR	2 (8.0)	1 (8.3)	1 (7.7)
SD	6 (24.0)	2 (16.7)	4 (30.8)
PD	17 (68.0)	9 (75.0)	8 (61.5)
Prior therapy lines, median (range)	5 (2–6)	5 (2–6)	4 (2–6)	0.769
Relapsed disease, n (%)	14 (56.0)	8 (66.7)	6 (46.2)	0.428
Bridging therapy, n (%)	13 (52.0)	8 (66.7)	5 (38.5)	0.238
Conditioning regimen, n (%)				0.097
TBC	17 (68.0)	6 (50.0)	11 (84.6)
BEAM	8 (32.0)	6 (50.0)	2 (15.4)
CAR T-cell products, n (%)				1.000
CD19 (approved)	6 (24.0)	3 (25.0)	3 (23.1)
CD19 and CD22 (clinical trial)	19 (76.0)	9 (75.0)	10 (76.9)

Fig. 4.

Response and survival of SCNSL patients with CNS lesions only and those with CNS and systemic lesions. A The clinical response of SCNSL patients with CNS lesions only and those with CNS and systemic lesions, the ORRs were 91.7% and 61.5%, respectively (P = 0.160). B–C Comparison of OS and PFS between SCNSL patients with CNS lesions only and those with CNS and systemic lesions, the estimated 1-year PFS rates were 83.3% and 38.5%, respectively (P = 0.030). The probability rates of PFS and OS were analyzed using the Kaplan–Meier method, and group comparisons were performed using the log-rank test

With a median follow-up time of 39.4 months for SCNSL with CNS lesions only and 30.9 months for SCNSL patients with CNS and systemic lesions, the median OS and PFS of SCNSL with CNS lesions only were not reached, while the median OS and PFS of SCNSL with CNS and systemic lesions were 20.5 and 8.8 months, respectively. The estimated 1-year OS rates of SCNSL with CNS lesions only and those with CNS and systemic lesions were 83.3% (95%CI: 48.2%-95.6%) and 67.1% (95%CI: 34.2%-86.2%), respectively (P = 0.067), and the estimated 1-year PFS rate were 83.3% (95%CI: 48.2%-95.6%) and 38.5% (95%CI: 14.1%-62.8%), respectively (P = 0.030) (Fig. 4B-C).

The kinetics of CAR T-cells

CAR transgene copy numbers in peripheral blood were quantified by droplet digital PCR technology to better understand the CAR T-cells expansion in vivo. The median peak levels of CD19 and CD22 CAR transgene copy numbers in vivo were 6611.5 (range: 38–214,676) and 4359 (range: 61–42,188) copies/µg DNA, respectively, with a median time to peak of 8 (range: 1–13) and 10 (range: 1–21) days after CAR T-cells infusion, respectively (Fig. 5A-B). Further analysis of CD19 CAR T-cells expansion between approved and clinical trial products indicated that the peak CD19 CAR transgene copy number was higher in approved products than in clinical trial products, at 43,705 (range: 3793–214,676) and 5208 (range: 38–55,128) copies/µg DNA (P < 0.0001), respectively (Fig. 5C-D). The median durable time of CD19 and CD22 CAR transgene copies were 56 and 63 days, and the CD19 CAR transgene copies were comparable between approved and clinical trial products, with 64 and 53 days (P = 0.450), respectively (Fig. 5E-F).

Fig. 5.

The kinetics of CAR T-cells in peripheral blood. A–B. The median peak levels and time to peak of CD19 and CD22 CAR transgene copy numbers. C–D. The median peak levels and time to peak of CD19 CAR transgene copy numbers between approved and clinical trial products, the peak transgene copy number of approved CD19 CAR T-cells were superior to that of clinical trial CD19 CAR T-cells. Each dot plot of A-D represents an individual, the lines are at the median with interquartile range, and group comparisons were performed using the Mann–Whitney U-test. E The lasting time of CD19 and CD22 CAR transgene in vivo, F Comparison of the lasting time of CD19 CAR transgene in approved and clinical trial products. The median lasting time of CAR transgene was assessed by the Kaplan–Meier method, and group comparisons were performed using the log-rank test

Discussion

CAR T-cell therapy following ASCT is a novel and emerging treatment approach that has shown initial success in treating patients with r/r CNSL [15, 16]. This study aims to analyze the clinical outcomes of 38 patients with r/r CNSL who underwent CAR T-cell therapy following ASCT, further confirming the efficacy and safety of this combined strategy in a large sample size. The combined therapy resulted in controllable side effects, with no patients dying of severe CRS or ICANS. The estimated 1-year PFS and OS rates were 72.8% and 57.4%, respectively, which demonstrates that CAR T-cell therapy following ASCT offers a durable response for patients with r/r CNSL. Furthermore, subgroup analysis revealed a correlation between the LDH level at treatment initiation, the number of prior therapy lines, and the response rate. Additionally, SCNSL patients with CNS and systemic lesions had poorer clinical outcomes than those with CNS lesions only, with a significantly reduced PFS rate.

We did not observe any uncontrollable adverse reactions associated with the present therapy. Most patients experienced fever, with a small proportion experiencing hypotension and hypoxemia. Neurotoxicity manifested as cognitive disorders, consciousness disorders, and seizures. The incidence rates of grade ≥ 3 CRS and ICANS were both 13.2%. Previous clinical trials of CAR T-cell therapy excluded patients with r/r CNSL due to the higher possibility of severe CRS and ICANS. However, in a real-world multicenter cohort study by Bennani et al., comparable incidence rates of grade ≥ 3 CRS (13% vs. 6%), grade ≥ 3 ICANS (31% vs. 33%), and use of tocilizumab and corticosteroids were observed after axi-cel infusion in patients with lymphoma involving and not involving the CNS [21]. This indicates the general safety of CAR T-cell therapy in patients with r/r CNSL. Subsequent clinical studies have reported varying incidence rates of severe CRS and ICANS in CNSL patients undergoing CAR T-cell therapy. Two large-scale studies focusing on r/r CNSL patients found that the incidence rates of severe CRS were 8% and 16%, and the incidence rates of severe ICANS were 20% and 44%, respectively [22, 23]. Furthermore, a meta-analysis demonstrated that the incidence rates of grade ≥ 3 CRS and ICANS in r/r CNSL patients receiving CAR T-cell therapy were 11.1% and 12.0%, respectively [13]. Thus, compared to using CAR T-cell therapy alone for patients with systemic B-cell non-Hodgkin lymphomaor CNSL, our research suggests that CAR T-cell therapy following ASCT in the r/r CNSL population has a comparable profile of CRS and ICANS. However, direct comparisons are not possible owing to differences in evaluation criteria across studies.

Hematologic toxicity also requires close attention. Previous studies have documented varying incidence rates of grade ≥ 3 cytopenia within 30 days following CAR T-cell therapy alone, with thrombocytopenia occurring in 28% to 65% of cases, anemia in 16% to 77%, and neutropenia in 59% to 95% [24]. Cytopenia can persist for several months to years after CAR T-cells infusion, with prolonged neutropenia ranging from 9% to 70% and thrombocytopenia ranging from 9% to 67% [25, 26]. Compared with these data for CAR T-cell therapy alone, the combined strategy of CAR T-cell therapy following ASCT is associated with higher incidence rates of both early and prolonged hematologic toxicity. However, variations in trial design, CAR T-cell products, and disease types may contribute to differences in the reported hematologic toxicity rates [27, 28].

Currently, salvage therapies for patients with r/r CNSL mainly include HD-MTX/HD-Ara-C based chemotherapy, BTK inhibitors, and immunomodulatory drugs. A phase Ib clinical study of ibrutinib combined with HD-MTX ± rituximab involving 15 patients with r/r CNSL demonstrated that 12 patients achieved clinical remission, resulting in an ORR of 80.0%, a CRR of 53.3%, and a median PFS of 9.2 months [29]. Another study, including 45 patients with r/r primary vitreoretinal lymphoma (PVRL)/PCNSL, reported that treatment with rituximab combined with lenalidomide yielded an ORR of 35.6%, with median PFS times for patients with PCNSL and PVRL of 3.9 and 9.2 months, respectively [30]. However, the long-term efficacy of these salvage strategies appears limited. Recently, CAR T-cell therapy has emerged as the preferred option for patients with r/r CNSL who have failed ≥ 2 lines of therapy. Several studies have demonstrated the clinical efficacy of CAR T-cell therapy alone for patients with r/r CNSL. For example, in a cohort of 25 patients with r/r PCNSL from the LOC network, CD19 CAR T-cells infusion (tisa-cel or axi-cel) resulted in an ORR of 80.0%, with median PFS and OS from leukapheresis being 8.4 and 21.2 months, respectively [23]. Additionally, a multicenter cohort study reported an ORR of 68.0% in 61 patients with r/r SCNSL treated with CAR T-cell salvage therapy. The median PFS and OS were 3.3 and 7.6 months, respectively [22]. In the present study, rather than utilizing CAR T-cell therapy alone, we employed CAR T-cell therapy following ASCT for patients with r/r CNSL. Our results indicate that the median OS and PFS have not yet been determined, with estimated 1-year OS and PFS rates of 72.8% and 57.4%, respectively. These findings suggest that CAR T-cell therapy following ASCT prolongs the remission duration of patients with r/r CNSL. Similarly, Xu et al. [15] found that the combined treatment of ASCT and CAR T-cell therapy significantly improved the disease prognosis of patients with r/r CNSL compared to CAR T-cell therapy alone. Hence, CAR T-cell therapy following ASCT offers patients with r/r CNSL advantages in terms of disease response duration and survival compared to either CAR T-cell therapy alone or traditional salvage chemotherapy.

The long-term duration can be explained by the following factors: Firstly, the conditioning regimen before transplantation can reduce disease burden and eliminate lymphocytes inhibiting CAR T-cells expansion and function. Secondly, ASCT can eliminate inhibitory immune cells and cytokines in the tumor immune inhibitory microenvironment. Thirdly, CAR T-cells can purify transplants and avoid collection contamination [16, 31, 32]. However, compared to our preliminary data [16] focusing on CD19 and CD22 CAR T-cells “cocktail” infusion following ASCT in 13 patients with CNSL, which demonstrated an ORR of 81.8% and a 1-year PFS rate of 74.6%, the present study showed inferior outcomes. The variations in outcomes can primarily be attributed to larger sample sizes, longer follow-up times, and differences in the baseline characteristics. Our current study focused on analyzing outcomes of patients with r/r CNSL who had active CNS involvement at treatment initiation, a factor previously established as being associated with inferior PFS in r/r SCNSL patients [33]. Additionally, the current study included patients who received commercially approved CAR T-cell products (Relma-cel and Axi-cel), whereas our prior report did not. Analysis revealed no significant differences in response and survival rates between patients receiving approved versus clinical trial products. This suggests that clinical results may be reproducible with different approved CAR T-cell products, thereby supporting the global practice of this combined treatment. Besides, our findings also demonstrate that r/r CNSL patients with ≤ 4 prior therapy lines achieved a superior CRR when they underwent CAR T-cell therapy following ASCT, consistent with observations in other hematologic malignancies treated with CAR T-cell therapy. Therefore, early initiation of combined therapy involving CAR T-cell therapy following ASCT should be conducted, leading to improvements in response and survival among patients with r/r CNSL.

Our work emphasizes the importance of utilizing bridging therapy prior to CAR T-cell therapy to improve outcomes. In our study, patients who received bridging therapy before CAR T-cells infusion exhibited a higher estimated 1-year OS rate compared to those without bridging therapy (85.0% vs. 58.8%, P = 0.054). A previous report indicated that administering radiotherapy as a bridging therapy before CAR T-cell therapy for r/r aggressive B-cell lymphoma could improve the PFS of patients [34]. In the largest cohort of PCNSL treated with CAR T-cells, a poor outcome was observed in patients who did not respond to bridging therapy, with a 1-year relapse-free survival of only 24% [23]. Generally, bridging therapy holds promise in reducing tumor burden, potentially enhancing the efficacy of CAR T-cell therapy. Further research can explore which types of bridging therapies can optimize prognosis. Furthermore, we compared the outcomes of SCNSL patients with CNS lesions only and those with CNS and systemic lesions, the estimated 1-year PFS rates were 83.3% and 38.5%, respectively (P = 0.030). A previous study found that after receiving CAR T-cell treatment, SCNSL patients with CNS relapse only had a longer median OS compared to those with both CNS and systemic relapse (not reached vs. 6.4 months, P = 0.07) [22]. Generally, SCNSL patients with both CNS and systemic lesions exhibit higher tumor burden, which serves as a risk factor for PFS in patients undergoing CAR T-Cell therapy, this observation could help to explain the disparity [22, 35].

Collectively, our results suggest that CAR T-cell therapy following ASCT can be considered a viable treatment option for patients with r/r CNSL who have failed ≥ 2 therapy lines, with adverse reactions being controllable. These findings contribute to the growing evidence supporting the use of CAR T-cell therapy following ASCT in patients with r/r CNSL. However, it is important to recognize the limitations of our study, notably its single-arm design. Further validation of these findings should be pursued through randomized controlled trials and multicenter clinical studies.

Abbreviations

ASCT

Autologous hematopoietic stem cell transplantation

BTK

Bruton’s tyrosine kinase

CI

Confidence intervals

CNS

Central nervous system

CNSL

Central nervous system lymphoma

CR

Complete response

CRP

C-reactive protein

CRR

Complete response rate

CRS

Cytokine release syndrome

CSF

Cerebrospinal fluid

DLBCL

Diffuse large B-cell lymphoma

ECOG

Eastern cooperative oncology group

HR

Hazard ratio

HSC

Hematopoietic stem cell

ICANS

Immune effector cell-associated neurotoxicity syndrome

LBCL

Large B-cell lymphoma

LDH

Lactate dehydrogenase

MRI

Magnetic resonance imaging

ORR

Overall response rate

OS

Overall survival

PCT

Procalcitonin

PCNSL

Primary central nervous system lymphoma

PD

Progressive disease

PFS

Progression-free survival

PR

Partial response

PVRL

Primary vitreoretinal lymphoma

SCNSL

Secondary central nervous system lymphoma

SD

Stable disease

Author contributions

J.W. and W.L. were responsible for analyzing data and writing the paper. Y.C., Y. Y., Z.S., M.Z. and Y.Z. took care of the patients. F.M. and X.Z. revised and edited the manuscript for submission. Y.X. provided funding acquisition and supervision. All authors were involved in patient recruitment and management in the clinical trial. All authors reviewed the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (No. 82270183 and 82370196).

Data availability

The data and materials cannot be shared publicly due to ethical constraints. No datasets were generated or analyzed during the current study.

Declarations

Conflict of interest

The authors declare no competing interests.

Ethical approval

This study was approved by the Ethics Committee of the Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (No: TJ-IRB20160314).

Consent to participate

The clinical trial was conducted in accordance with the principles of the Declaration of Helsinki, and informed consent was obtained from all eligible patients.

Consent for publication

All authors have agreed the submission.