Neutrophil activation and clonal CAR-T re-expansion underpinning cytokine release syndrome during ciltacabtagene autoleucel therapy in multiple myeloma

Yang, Shuangshuang; Xu, Jie; Dai, Yuting; Jin, Shiwei; Sun, Yan; Li, Jianfeng; Liu, Chenglin; Ma, Xiaolin; Chen, Zhu; Chen, Lijuan; Hou, Jian; Mi, Jian-Qing; Chen, Sai-Juan

doi:10.1038/s41467-023-44648-3

Download PDF

Article
Open access
Published: 08 January 2024

Neutrophil activation and clonal CAR-T re-expansion underpinning cytokine release syndrome during ciltacabtagene autoleucel therapy in multiple myeloma

Nature Communications volume 15, Article number: 360 (2024) Cite this article

3943 Accesses
3 Altmetric
Metrics details

Subjects

Abstract

Cytokine release syndrome (CRS) is the most common complication of chimeric antigen receptor redirected T cells (CAR-T) therapy. CAR-T toxicity management has been greatly improved, but CRS remains a prime safety concern. Here we follow serum cytokine levels and circulating immune cell transcriptomes longitudinally in 26 relapsed/refractory multiple myeloma patients receiving the CAR-T product, ciltacabtagene autoleucel, to understand the immunological kinetics of CRS. We find that although T lymphocytes and monocytes/macrophages are the major overall cytokine source in manifest CRS, neutrophil activation peaks earlier, before the onset of severe symptoms. Intracellularly, signaling activation dominated by JAK/STAT pathway occurred prior to cytokine cascade and displayed regular kinetic changes. CRS severity is accurately described and potentially predicted by temporal cytokine secretion signatures. Notably, CAR-T re-expansion is found in three patients, including a fatal case characterized by somatic TET2-mutation, clonal expanded cytotoxic CAR-T, broadened cytokine profiles and irreversible hepatic toxicity. Together, our findings show that a latent phase with distinct immunological changes precedes manifest CRS, providing an optimal window and potential targets for CRS therapeutic intervention and that CAR-T re-expansion warrants close clinical attention and laboratory investigation to mitigate the lethal risk.

FOXO1 is a master regulator of memory programming in CAR T cells

Article Open access 10 April 2024

FOXO1 enhances CAR T cell stemness, metabolic fitness and efficacy

Article Open access 10 April 2024

The present and future of bispecific antibodies for cancer therapy

Article 06 March 2024

Introduction

Over a decade, chimeric antigen receptor modified T cells (CAR-T) therapy emerges to be an attractive approach against hematological malignancy, marking a cancer treatment paradigm shift. The overall therapeutic effectiveness of CAR-T is quite encouraging, but its toxicity varies individually from clinically manageable to life-threatening. Cytokine release syndrome (CRS) is the most common adverse effect after CAR-T infusion. Severe CRS (Grade ≥ 3) is reported in nearly 46% of B-cell acute lymphoblastic leukemia (ALL)¹ and 13% of B-cell lymphoma² patients receiving anti-CD19 CAR-T therapy, and 41% of multiple myeloma (MM)³ patients infused with anti-BCMA CAR-T.

CRS is theoretically an indispensable inflammatory response during CAR-T therapy. Appropriate CRS is thought to be helpful to mitigate cancerous cells. However, excessive CRS can result in vital organ injury, and often put patients at risk. Hence, an in-depth understanding of the CRS mechanism of action is warranted to properly manage severe immune response. Serological tests show that CRS occurs with a robust elevation of diverse cytokines⁴, of which, IL-6, IL-1β and GM-CSF, mediated by the monocyte/macrophage signalings, play crucial roles in cytokine storm development^5,6,7. However, the comprehensive chronological pathophysiologic landscape of CRS upon CAR-T therapy hasn’t yet been fully elucidated.

Ciltacabtagene autoleucel (Cilta-cel, previously known as LCAR-B38M) is a biepitopic CAR-T product targeting BCMA antigen, and has achieved remarkable success in the treatment of relapsed/refractory (r/r) MM patients in China^3,8,9 and United States of America (US)¹⁰. In light of it, Cilta-cel successively obtained approval from the US FDA, the European Medicines Agency, and the regulatory agency of Japan¹¹ recently.

In the present work, peripheral blood (PB) samples are collected from r/r MM patients who received Cilta-cel. Longitudinal cytokine profiling and gene transcriptomes are performed to characterize the dynamics of acute inflammatory events in response to CAR-T. We reveal a kinetic change in the axis of signaling activation-cytokine release-clinical manifestation, showing a stepwise evolution of CAR-T-related acute inflammation. We also point out that multiple factors may be able to collectively trigger CAR-T re-expansion, and subsequently prompt a toxic reaction that is probably severer than tumor-alone-driven CRS.

This study provides a solid rationale for a timely and precise application of targeted drugs, and proposes a notion that CAR-T re-proliferation requires more close monitoring and effective medical management to alleviate the lethal severity.

Results

Patient characteristics, clinical response and study design

The samples investigated were obtained from 26 Cilta-cel-exposed r/r MM patients, of whom, 16 cases were from phase I Legend-2 trial as we reported previously³, and 10 from phase II CARTIFAN-1 trial⁹. Patients’ demographic data are summarized in Table 1. Statistically, there were no significant differences of age, gender, MM subtype and other baseline characteristics between the two subpopulations from the above-mentioned separate trials (Table 1).

Table 1 Baseline characteristics

Full size table

As of the manuscript preparation, the overall response rate of the 26 cases was 88.5%, with 80.8% obtaining complete responses (CR). The progression-free survival (PFS) rates were 39.2% at 3 year and 26.1% at 5 year; the overall survival (OS) rates were 53.1% at 3 year and 45.5% at 5 year (Supplementary Fig. 1a~b).

PB specimens were collected at the given time points. Serum was separated from PB for cytokine profiling. Purified mononuclear cells of PB (PBMCs) were assigned to three experimental cohorts as shown in Fig. 1a and Supplementary Data 1.

**Fig. 1: Overview of the study design and dynamic changes of CAR-T kinetics.**

Characterization of CRS

Within 30 days post Cilta-cel administration, CRS was observed in all patients at a median onset time of 6 days (range, 1~10 days). Fever signaled as an earliest event in 65.4% cases, appearing as early as 6 days at average. In the other 34.6% patients, hypotension, hypoxemia or organ malfunction presented earlier than fever in CRS outbreak. 34.6% patients had Grade 1~2 CRS and 65.4% suffered Grade 3 or worse (Fig. 1b). In accordance with clinical symptom, CRS grading was adjusted and the change of CRS grading aligned with the kinetic amount of circulating Cilta-cel (Fig. 1b~c). CD8:CD4 ratio of CAR-T increasingly elevated as CRS progressed, reaching the peak at day 16~20 (Fig. 1d). Based on the two typical T cell immunophenotying markers CCR7 and CD45RA¹², cytotoxic CAR-T in CRS was enriched for effector memory (CCR7^- CD45RA⁻) and effector memory T cells re-expressing CD45RA (CCR7^- CD45RA⁺), particularly accumulated at day 10~12 (Fig. 1e).

To examine whether the baseline clinical and laboratory characteristics had an impact on CRS, we compared several key parameters between severe (grade > 2) and mild CRS (grade ≤ 2) groups. The patients subjected to combination lymphodepletion (Cyclophosphamide + Fludarabine) experienced more severe CRS than those receiving Cyclophosphamide alone (15/18 versus 2/8, p = 0.008), which was consistent with the observation previously stated¹³. At baseline, endogenous lymphocyte count was significantly lower in the severe CRS group than that in the mild group (p = 0.016). However, there was no statistical difference in the neutrophils (p = 0.545) and monocyte (p = 0.199) count between mild and severe CRS groups. With respect to the CAR-T, the mean frequencies of CD8⁺ and CD4⁺ CAR-T in the mild CRS group were respectively 59% (0.46 × 10⁶/kg) and 41% (0.25 × 10⁶/kg), and those in the severe group were respectively 62% (0.43 × 10⁶/kg) and 38% (0.32 × 10⁶/kg). Neither CD8⁺ nor CD4⁺ reached proportionally and numerically significant differences between the two CRS groups. The killing capacity of CAR-T was evaluated by the apoptotic proportion of BCMA-expressing tumor in co-culture. Though phase I and phase II adopted different effector-to-target (E:T) ratios in vitro, there was no evident tumor apoptotic disparity between severe and mild CRS cohorts.

In addition, the individual therapeutic response to CRS in terms of CRS symptoms, grades, management, and outcome was summarized in Supplementary Data 2. We found no significant difference in clinical outcome between the patients with severe and mild CRS, suggesting clinical prognosis did not correlate with CRS grade.

Dynamic changes of cytokine profile

To gain insight into the development of CRS, patients’ clinical presentation separated the entire 21-day inflammatory reaction into 5 slots, including baseline (before infusion), latent (day 3~5), fever (day 6~9), acute aggravation (day 10~15, AA), and resolving (day 20~21) periods. Since a robust cytokine secretion at the AA period, we collected individual samples two times (Day 10~12 and Day 13~15), denoted as AA1 and AA2, respectively to gain adequate information. Based on these time slots, serological data were accordingly assessed to draw an overall landscape of cytokine profile.

As compared with the baseline, the inflammatory molecules substantially up-regulated at the late fever period, reaching a high level at the acute aggravation period (Fig. 2a). Several well-studied cytokines, such as IL-6, Granzyme B, IL-10, G-CSF and CXCL10, rocketed up by 50~100 folds at the peak, indicating a highly active inflammatory response at the periods spanning from day 6 to 15 (Fig. 2b). By the means of unsupervised clustering, we identified five cytokine clusters (CCs), among which, CC3, CC4, and CC5 displayed a positive correlation with mean CRS grade, whereas CC2 exhibited a negative correlation (Fig. 2c). Notably, among the cytokines positively correlated with CRS severity, sIL-2Rα, sTNF RII (both belonging to CC4), and Granzyme B (belonging to CC5) exhibited the strongest correlations (Fig.2a, d Supplementary Data 3). To interpret the correlation of cytokine expression and CRS level, we adopted the computational methodology CONNECTOR¹⁴ and partitioned the 26 patients into two groups: Group1 (n = 8) and Group2 (n = 18) (Fig. 3a~b). Further analysis of the mean log₂FC of cytokines revealed a distinct cytokine kinetic mode of Group1 from that of Group2 (Fig. 3c~d). The patients in Group1 displayed a higher peak level of CC3, CC4, and CC5 cytokines, as well as the whole, as compared with Group2. Based on this grouping, we can clearly distinguish the temporal changes in CRS grading during the treatment course. The patients in Group1 experienced severe CRS which peaked around day 12, whereas the patients in Group2 having a relatively mild CRS with a low peak around day 9 (Fig. 3e). The temporal dynamics of cytokine secretion was highly consistent with CRS grade regardless of individual discrepancy.

**Fig. 2: Dynamic changes of cytokines.**

**Fig. 3: Dynamic correlation between cytokine expression and CRS level by CONNECTOR.**

The above results demonstrate that the timing of cytokine production aligned with the clinical symptoms, revealing a highly regular systemic inflammatory reaction upon CAR-T-tumor contact.

Dysregulated cytokines and their clinical relevance

In the Cilta-cel treatment, the top major adverse effects were hepatic dysfunction and coagulation dysfunction, both accounting for 92.3% with the majority being asymptomatic. Correlation analysis of inflammatory factors and end organ damage identified a constellation of cytokines was positively related to aspartate aminotransferase (AST) and alanine aminotransferase (ALT) elevation. Granzyme B and IFN-γ, known to cause liver impairment¹⁵, showed the strongest correlation (Pearson’s r > 0.8) (Supplementary Fig. 2a). Overall, CRS induced hepatic damage was transient and reversible since supportive care was effective in most patients.

Meanwhile, a small set of molecules, such as Angiopoietin-1 (Ang-1), PDGF-AA, PDGF-AB/BB and EGF, were noticeable to be down-regulated as CRS progressed (Fig. 2a) and inversely associated with the severity of coagulation dysfunction (Supplementary Fig. 2b). These cytokines play pivotal roles in stabilizing endothelial integrity^16,17 which was indispensable for preventing patients from coagulation system disruption^18,19. By contrast, Angiopoietin-2 (Ang-2), an angiogenic factor, showed positive correlation with CRS grading (Fig. 2a). The ratio of Ang-2 and Ang-1, an indicator of endothelial activation^13,20,21, was responsively elevated as coagulation dysfunction turned worse (Supplementary Fig. 2b). It implied that cytokines could pose destructive effect to the endothelial steady-state condition, representing a pathophysiological feature of CAR-T induced cytokine storm.

Immune cells-mediated signaling regulation of cytokine release

To explore the upstream signaling responsible for cytokine regulation, transcriptome sequencing was performed in the PBMCs at different periods. Three algorithms collectively identified 10,917 differentially expressed genes (DEGs) (Supplementary Fig. 3a), which were assigned into eight clusters with consistent expression trends using soft clustering method (Fig. 4a). We noticed that clusters 2 and 7, which represented the signaling activated at the earlier phase (day 3~9), were significantly enriched for a variety of biological processes involved in neutrophil, T lymphocyte, platelet, monocyte/macrophage activation, cytokine signaling up-regulation, as well as excessive oxidative stress (Fig. 4a~b, Supplementary Fig. 3b). In addition, B cells affiliated genes were significantly enriched in the cluster 6, accompanied by the gene transcription being decreasingly down-regulated since lymphodepletion and Cilta-cel therapies (Fig. 4a~b). Moreover, correlation analysis showed that clusters 1 and 5, characterized by T cell activation and cell cycle, were positively correlated with CRS severity, and cluster 7 was negatively correlated with CRS severity (Supplementary Fig. 3c). Considering CRS is a dynamic process, these data suggested that neutrophils and platelets were activated in the early stage of CRS, and so was oxidative stress. Distinctly, at the CRS aggravation stage, T-cell activation and proliferation were augmented.

**Fig. 4: Signaling involving CRS by RNA sequencing.**

To better understand the timing of immune cell activation cascade, gene set enrichment analysis (GSEA) was performed to compare each time point to the baseline. Interestingly, the neutrophils responded strongest as early as day 3~5 (latent period), while immune response from macrophages and T cells started at the latent period, and turned strong at the fever period of day 6~9 (Fig. 4c), suggesting that neutrophils acted as a prominent pioneer in initiating inflammatory reaction, whereas macrophages and T lymphocytes were more engaged in the centre of subsequent cytokine cascade. As the immune cells responded, the key inflammatory signaling pathways TNF, IL-6/JAK-STAT3, JAK/STAT, IL-2/STAT5 and complement/coagulation cascades were significantly upregulated accordingly (Fig. 4c), which were well-known for their central roles in cytokine storm^{19,22,23,24,25}. Similar conclusions were drawn by gene set scores (Supplementary Fig. 4). These data were predictive of potential targets and appropriate time for the use of signaling blockers.

CAR-T re-expansion in three patients

The majority of patients experienced CAR-T expansion once, but three cases (RJ-25, 27 and 31) presented re-expansion of Cilta-cel following the first CAR-T peak (Supplementary Fig. 5a, b, Fig. 5). Re-expansion onset time in RJ-25 and RJ-27 was half a year after CAR-T infusion whereas that in RJ-31 was only 22 days. In parallel, RJ-25 and RJ-27 had very limited cytokines produced (Supplementary Fig. 5c, d). By contrast, RJ-31 suffered life threatening CRS. Such discrepancies prompted our interest to figure out the potential causes. We found, at the half of year, RJ-25 was diagnosed with herpes zoster eye’s disease, and RJ-27 was tested positive for human rhinovirus with on-going complete responses (Supplementary Fig. 5e). Viral infection in both cases was symptomatic and most likely owing to persistent hypoimmunoglobulinemia as plasma cells remained deficient (Supplementary Fig. 5f~h). Since T lymphocytes are the key guardians of adaptive immunity, it was speculated that CAR-T might took part in immune response to pathogen invasion and increasingly proliferated again. With respect to RJ-31, this patient consequently died of hepatic failure due to fatal CRS during the second CAR-T amplification. To deeply comprehend the origin of the excessive CRS in RJ-31, this special case was investigated in details.

**Fig. 5: Fatal cytokine storm in RJ-31 following the first CRS.**

Case study of a fatal cytokine storm following the primary CRS

RJ-31, a 61-year-old male, presented the first severe CRS (1^st_CRS) on day 9, and relieved soon after Tocilizumab administration. The adverse events (AEs) of hepatic impairment (Grade 3) and coagulation dysfunction (Grade 2) were transient during the primary immune reaction and relieved on day 12 (Fig. 5a). A second CRS (2^nd_CRS) appeared on day 20 with progressing jaundice (Fig. 5a), which lasted 18 days until he died of liver failure and hepatic encephalopathy in spite of the absence of fever, hypotension at the initial stage. The Cilta-cel in the blood stream presented two peaks on day 11 and day 26~28, respectively, and the ratio of CD8⁺ to CD4⁺ circulating CAR-T was significantly higher at the second episode compared with that at the 1^st_CRS (Fig. 5b, Supplementary Fig. 6a). Cytokine profile confirmed that severe inflammatory storm consecutively occurred twice in RJ-31. The 1^st_CRS was manifested by a transient increase mainly in Granzyme B, IL-6, IL-10, while 2^nd_CRS exhibited a durable and robust rise of more cytokines, such as IL-8, IFN-γ, CX3CL1, sIL-1 RII, sVEGF R1, Ang-2 (Fig. 5c). Compared with the common signaling activation on day 8, GSEA pathway enrichment showed a more fierce T cell immune activity, platelet degranulation and a stronger potency of various inflammatory signaling involving Erk, IKK/NF-κB, Toll-like receptor pathways and oxidative stress, mitochondrial damage on day 22, concomitant with highly active glycolysis, pyruvate metabolic and fatty acid metabolic processes (Fig. 5d). A widespread down-regulation of cytokine transcriptional signaling since day 26 might result from corticosteroids intervention (Fig. 5d).

Hepatic biopsy was obtained on day 38. The transgene copy number of Cilta-cel in the liver was 9.5 times higher than that in the matched PB of day 38 (Fig. 5e). Immunohistological staining showed a massive cytotoxic CD8⁺ T cell infiltration containing toxic granzyme B and TIA-1, high KP-1 (CD68) expression (Supplementary Fig. 6b~g), reflective of a T cell mediated immune response in the liver, which should be stronger than that in the PB owing to a higher accumulation of CAR-T. Since CAR-T were normally activated by the targeted antigen, we hypothesized the presence of BCMA-expressing cells may act as a stimulus. Albeit no imaging evidence to support tumor infiltration in the liver, immunohistochemistry indeed revealed a scattered distribution of plasma cells expressing CD138 and BCMA (Supplementary Fig. 6h~n), indicating that the 2^nd_CRS was probably initiated in the liver with cytotoxic Cilta-cel accumulation, which was most likely relevant to the exposure of localized BCMA antigen.

A dominant clone of Cilta-cel was identified in RJ-31 at the 2^nd_CRS

Unexpectedly, RNA sequencing of PBMCs in RJ-31 uncovered a dramatic skewing of TCR repertoire to a dominance possessing TRAV13-1~J20/TRBV12-5~J2-1, which took up 89% of the entire TCR repertoire on day 28 (Fig. 6a). TCR diversity (inverse Simpson index) and clonality (clonality index) indicated that RJ-31 uniquely showed a deep reduction of TCR diversity and a remarkable increase of T cell clonality since day 22 (Fig. 6b). The other cases all showed a transient, light change of TCR diversity and clonality, followed by a quick return to the original status (Fig. 6b), reminiscent of the common findings previously described^26,27. Lentiviral vector integration site analysis further revealed a high abundance of a clone with monoallelic integration at the downstream (nearly 14 kbp away) of ETS1 gene without affecting its transcription confirmed by quantitative PCR (Fig. 6c, Supplementary Fig. 7), ruling out the association of Cilta-cel dominance establishment and lentiviral insertion.

**Fig. 6: Identification of a dominant CAR-T clone in RJ-31.**

To functionally evaluate the property of the dominant clone, purified CAR-positive cells at the 2^nd_CRS (day 28), as well as the matched control in the 1^st_CRS (day 11) were harvested for single-cell RNA sequencing. Overall, CAR-T in the 2^nd_CRS was enriched for the cytotoxic CD8⁺ effectors (Fig. 6d, Supplementary Fig. 8a~b). As a comparable control, a small dominant CAR-T clone, TRAV19~J34/TRBV4-1~J2-7 (referred to as Major_A) in the 1^st_CRS was selected for comparison with the dominant TRAV13-1~J20/TRBV12-5~J2-1 (referred to as Major_B) identified 73% in the 2^nd_CRS (Fig. 7e). This investigation showed that the Major_B possessed a stronger cytotoxic capacity respective of lytic cell composition (Supplementary Fig. 8c) and higher toxicity score (Fig. 6f~g, Supplementary Fig. 8d). Additionally, the significant DEGs in Major_B were involved in Toll-like receptor signaling pathway, apoptotic mitochondrial changes and calcium iron related processes (Fig. 6h), which was in line with the finding generated by PBMCs transcriptome sequencing. Meanwhile, the entire CAR-T population could be representative of the clone when the dominance took up 73%~78% during day 28~38 (Figs. 5b, 6c, e).

**Fig. 7: Development of CRS after Cilta-cel therapy.**

HHV7 activation and TET2 mutation might be collectively responsible for clonal amplification of Cilta-cel in RJ-31

Given CAR-T clonotype rarely skewing to a single dominance upon tumor stimulation, we speculated that the dominant clone was contributed by another factor other than BCMA-expressing cells. To mine the unknown peptide, the TCR clone was shown highly matched with herpesvirus5 (HHV5) in the ImmuQuad Biotech database. Nevertheless, transcriptome sequencing of RJ-31 PBMCs surprisingly found herpesvirus7 (HHV7) but not HHV5 was activated starting from the resolving period of the first CRS. Back to the aforementioned database, HHV7 was not included as a potential target for TCR alignment indeed. Though HHV5 and HHV7 only shared 38.6% homology at peptide level, HHV7 was hypothesized as a stimulus of the CAR-T dominance formation. This hypothesis was verified by a transient positivity of the HHV7 fragment between day 20 and 26 by RT-qPCR, which was exactly consistent with the course of 2^nd_CRS (Supplementary Fig. 9a~b). Hence, HHV7 infection was probably the key stimulus for the corresponding CAR-T clone re-expansion.

Along with the engineered cells predominately with the CD8⁺ subset outgrew (Fig. 5b), CD8⁺ endogenous (CAR-negative) T cells preferably proliferated as RJ-31 suffered from viral infection (Supplementary Fig. 9c). It was reminiscent of the viral infection of RJ-25 and RJ-27, both of whom also had a high CD8⁺ CAR-T re-expansion. At the same time point, the endogenous T cells exhibited a higher CD8⁺ fraction compared to that at the early treatment phase (Supplementary Fig. 9c). Taken together, the three cases shared a common feature, which was, a high CD8 CAR-T and endogenous T cell subpopulation level at virus infection. This phenomenon indicated a competent immunity against the exogenous microbe. However, we noticed that the T clonotype of RJ-25 and RJ-27 kept the polyclonal state (Supplementary Fig. 5i). This triggered us to explore additional cause potentially contributing to the accumulation of a single CAR-T clone.

Interestingly, PBMCs transcriptome sequencing showed a heterozygous mutation in TET2 gene from day 20 on. This newly identified abnormality was validated by targeted DNA sequencing, which demonstrated the variation (C > T) of TET2 localized at the genomic site of Chr4:106197167, resulting in an early translational termination of the catalytic structural domain of TET2 protein (referred to as TET2^mut, Fig. 6i, Supplementary Fig. 10a~b). TET2 mutation is commonly considered as an initiating event of clonal hematopoiesis^28,29, and dysfunction of TET2 protein causes an exacerbation of immune inflammation, particularly characterized by an IL-6 elevation³⁰. To precisely determine if a low abundance of TET2^mut existed before the 2^nd_CRS, ddPCR was performed and revealed a small TET2^mut clone already existed in PBMCs at pre-treatment indeed and mainly concentrated in CD3 positive host T cells (Fig. 6j, Supplementary Fig. 10c). This variation was also detected in the CAR-T on day 11 with low variant allele fraction (VAF) of 0.40%, and on day 28 with higher VAF of 38.07% by ddPCR (Fig. 6j). Moreover, in the liver biopsy on day 38 when the case RJ-31 died of the lethal toxicity, TET2^mut was found to have similar VAF levels to that of day 28 peripheral CAR-T (Supplementary Fig. 10d). It was worth noting that TET2^mut VAF of the CD3-negative subset was merely 0.4% at a late time point (day 33), which was basically close to the VAF (0.1%) at pre-treatment (Fig. 6j, Supplementary Fig. 10c). It suggested that the somatic TET2^mut present at leukapheresis was more carried by T lymphocytes, consequently potentiating the engineered T cells clonal expansion, and TET2 mutation was unable to drive non-T hematopoietic cells clonal expansion under the circumstance of microbial infection.

In sum, virus activation, a rare somatic mutation of TET2, along with plasma antigen stimulation in the RJ-31 liver might collectively contributed to the clonal amplification of Cilta-cel, prompting a second severe CRS and conferring an excessive inflammation, and fatal damage to the liver function.

Discussion

This study uncovered an orchestrated chronological process of inflammation in response to CAR-T. The CRS pathophysiological processes investigated in this work are summarized in Fig. 7. We pointed out that CAR-T could be activated by either targeted tumor antigen or TCR-specific pathogen. In the common setting, CAR bound tumor cells to produce pro-inflammatory factors that initially lead to neutrophils activation, followed by participation of other immune cellular components. In a viral invasion setting, the viral peptide presented by antigen presenting cells could allow CAR-T to be activated in a classical TCR/MHC manner. Other factors, such as target antigen exposure in the malignant or normal cells might reinforce CAR-T expansion, and genetic abnormality (such as somatic mutation of TET2) in CAR-T might confer a proliferating advantage as well.

CRS is a result of an activation of a multicellular network. Other than T lymphocyte and monocyte/macrophage, the primary cytokine producers recognized^5,6,7, neutrophils were found the earliest activated after Cilta-cel infusion. As innate immune effectors, neutrophils work at the first line of response for the acute inflammation³¹ and take part in the pathophysiology of CRS in COVID-19^32,33 and acute GVHD;³⁴ whereas, the role of neutrophils in CAR-T related CRS remains unclear. Our finding provides compelling evidence that neutrophilic immunostimulatory reaction also drives a major cellular response at the early stage of CRS outbreak in Cilta-cel therapy.

The stepwise procedure of Cilta-cel-induced CRS depicted a kinetic change in the axis of signaling-cytokine-presentation. On one hand, inflammatory signaling pathways are shown to be activated in 3~5 days prior to cytokine storm, illuminating that close monitoring and essential supporting care are needed as early as day 3~5 to reduce the risk of unmanageable toxicity caused by delayed intervention. On the other hand, the hyper-activation of IL-2/STAT5, IL-6/JAK-STAT3, TNFα signaling indicates the potential targets for drug administration. Tocilizumab is the most common agent to help control CRS²⁵. TNFR blocker, Etanercept, has been applied in some cases³. Other agents approved for controlling GVHD, such as JAK2 inhibitor (Ruxolitinib), IL-2 receptor antagonist (Basiliximab), can also be considered used against the over-reacted inflammation in the CAR-T therapy^22,23. Meanwhile, the optimal time point of targeted agent usage is expected not to be later than day 10, at which, both cytokines and activated signaling are reaching peak levels.

Although re-expansion of CAR-T could occur under certain circumstances^35,36,37, multi-factor-induced CAR-T re-expansion hasn’t been elucidated yet. In this study, we report a rare case who suffered two sequential CRS hits within one month post CAR-T infusion. To our knowledge, among 219 r/r MM cases receiving Cilta-cel therapy so far^3,9,10,38, this was the only one who developed a lethal CRS with an evidenced clonal expansion of CAR-T. We assume that several factors may result in this unexpected issue: first, late immune reaction could be induced by CAR-T and antigen-exposing cells interaction again at localized lesion³⁹; second, virus reactivation may occur in the setting of immune incompetency after CAR-T treatment^40,41; third, the somatic TET2 mutation is most likely the key intrinsic cause of CAR-T’s clonal expansion and subsequent fatal inflammatory storm in RJ-31, which was not relevant to Cilta-cel manufacturing.

With respect to the somatic TET2^mut found in this study, RJ-31 aged 61 was speculated to have the clonal hematopoiesis of indeterminate potential (CHIP), which occurs in apparently healthy elderly people in particular, with DNMT3A, TET2 and ASXL1 loss-of-function mutations being the most commonly detectable ones^42,43. A small clone with TET2 mutation could conceal in RJ-31’s T lymphocytes before their collection for CAR-T preparation, constituting a risk of wild outgrowth of CAR-T cells derived from the autologous lymphocytes under certain circumstance, such as viral infection. Nevertheless, the other two patients (RJ-25 and RJ-27) were not detected with somatic TET2 mutation, as well as other gene mutations associated with CHIP including DNMT3A and ASXL1, so the risk of TCR clonality formation was largely minimized. Therefore, for the RJ-31 case, the genetic defect was considered the overriding determinant to drive the clonal proliferation of CAR-T, and the viral infection was probably an antigen stimulus to trigger adaptive immunity with CAR-T cells being a part.

The phenomenon in RJ-31 was reminiscent of a case (Patient-10) reported in the anti-CD19 CAR-T therapy (referred to as the CD19 CAR-T case, hereafter), in which, the CAR-T infused for the second time bore TET2 sequence abnormalities on both alleles, enabling clonal proliferation of CD8⁺ central memory CAR-T, and rapidly achieving tumor clearance accompanied by high-grade CRS⁴⁴. Though the similarities in TET2 mutagenesis, CD8⁺ graft clonal outgrowth and severe toxicity, the disparity of the consequent outcomes between RJ-31 and the CD19 CAR-T case probably lies in the following reasons: first, the distinction in CAR-T immunophenotypes. The cytotoxic effector phenotype in RJ-31 was of higher cytokine-producing capacity, compared with the central memory phenotype found in the CD19 CAR-T case; second, the difference in collaborative effects. Albeit TET2^mut should play a prominent role in CAR-T’s aggressive growth in RJ-31, the influence from other two factors, viral activation and antigen stimulation, could not be neglected. Relatively, the CD19 CAR-T case was not reported to have other triggers besides TET2 abnormalities; third, the difference in the CRS occurring time. The next CRS struck consecutively when RJ-31 hadn’t completely recovered from the first storm, whereas the CD19 CAR-T case didn’t experience sequential CRS; fourth, the distinction in cytokine profiles. The CD19 CAR-T case was observed having common inflammatory molecules elevation; however, RJ-31’s cytokine signature was considerably broadened and more toxic. Despite the objective observations above, such genetic events which happened to either RJ-31 or the CD19 CAR-T case are occasional. Screening of CHIP-associated genes, particularly TET2, DNMT3A and ASXL1, can be an optional measure in CAR-T manufacturing in some elderly populations presenting CHIP bias, such as immune cytopenia, marrow dysplasia, and nutritional deficiency.

Additionally, the life-threatening CRS in RJ-31 evoked our awareness of proper and timely administration of immunoglobulins or specific anti-microbe agent, minimizing the possibility of pathogenic infection at the period when immunoglobulin declines and circulating CAR-T remain in the effector status. Our work also conveyed an implication for the use of a bi-specific CAR-T simultaneously targeting BCMA antigen and a commonly affected virus, such as CMV. Such a combination targeted approach may help empower immunity and lower the risk of severe virus infection. Furthermore, CAR-T constitutively carrying suicide genes are in great demand to cope with life-threatening side effects.

In conclusion, our study intensively evaluated the physiological process and the timing of CAR-T therapy-related CRS development, deepening our understanding of systemic toxicities in cellular immunotherapy. Although this study was carried out in the r/r MM patients with Cilta-cel treatment, we believe that the key findings is not restricted to this typical disease type and engineered product, instead, the conclusions can be shared by some of other CAR-T treatments. Of course, CAR-T-related CRS studies with a larger sample size or distinct tumors are highly encouraged to consolidate our discoveries in the future.

Methods

Patients and clinical assessment

The Legend-2 (phase I) and CARTIFAN-1 (phase II) studies were two pivotal trials of Cilta-cel in China. This study included all participants with available specimens in our clinical center, Ruijin Hospital affiliated with Shanghai Jiao Tong University School of Medicine (referred to as RJ hereafter). The participating sites of the phase I trial comprised four centers with one in the West China and the other three in the East (NCT03090659, ChiCTR-ONH-17012285). The East area (RJ, First Affiliated Hospital of Nanjing Medical University and Changzheng Hospital) with RJ as the leading site administrated Cilta-cel treatment in 17 patients, of whom, 16 patients’ sequential serum samples were obtained while in one case the sample was unavailable. The phase II trial was conducted in eight centers (NCT03758417). Each was in charge of its own patient samples collection. As of manuscript preparation, RJ had treated 10 phase II patients whose serum samples were entirely investigated. Therefore, 26 r/r MM patients were included in this study without any selection bias. All the evaluated patients underwent cyclophosphamide-based (with or without fludarabine) lymphodepletion therapy followed by Cilta-cel infusion at a median dose of 0.595×10⁶ CAR-T cells/kg. The written informed consent to sample collection and data mining was obtained from all the patients. The study was approved by the RJ ethics committee.

Patient clinical response was assessed by the criteria of treatment response according to International Myeloma Working Group (IMWG) consensus recommendations⁴⁵. AEs were graded according to National Cancer Institute Common Terminology Criteria for Adverse Events (NCI-CTCAE v. 4.03). Determination of CRS was based on the criteria proposed by Neelapu et al.⁴⁶. Five grades were set for CRS stratification, in which, Grade 1–2 was defined as mild toxicity and Grade 3 to 5 was referred to as severe event. The term of coagulation dysfunction used in this study was a collective description of the AEs relevant to platelet count, activated partial thromboplastin time or fibrogen. The term of liver damage used in this study was a collective description of the AEs relevant to AST, ALT, fibrogen and Bilirubin. The grading of coagulation dysfunction and liver damage depended on any of the above-mentioned AEs that met the end of the highest AE stage.

CAR-T functionality assessment

The killing capacity of CAR-T was evaluated at the quality control step of manufacturing. CAR-T were co-incubated for 24 h with luciferase labeled-RPMI-8226 cells at an E:T ratio of 20:1 (phase I) or 5:1 (phase II). Controls were the untransfected T cells from the same donor. After 24 h incubation in 37°C, 5% CO₂ cell culture incubator, the cells in each well were added with 25 μl ONE-Glo Firefly luciferase assay reagent mix (Promega, E6120) and incubated at room temperature for 1 min. Relative light unit (RLU) of luciferase activities from the remaining living target cells were read in a microplate reader (Tecan Spark 10 M) and were used for cell viability calculation.

In phase I with an E:T ratio of 20:1, the mean apoptotic cell fraction of the mild CRS group was 98% (range: 87–100%), and that of the severe group was 95% (range: 83–100%), without significant difference. For the 10 investigated patients in phase II with an E:T ratio of 5:1, the average killing capacity of CAR-T was 61% and 56% in mild (one patient) and severe (9 patients) groups, respectively, and wasn’t reflective of an evident apoptotic disparity.

Flow cytometry

Fresh PBMCs were harvested, and then incubated with antibodies for 30 min at 4 °C for staining. CAR-T immunophenotyping was detected using fluorochrome-conjugated antibodies against human CD45 (BD Biosciences, Cat: 555483; Clone: HI30; Lot: 9337233), CD3 (BD Biosciences, Cat:555335; Clone: UCHT1; Lot: 7200698), CD4 (BD Biosciences, Cat:557871; Clone: RPA-T4; Lot: 7348612), CD8 (BD Biosciences, Cat:564116; Clone: SK1; Lot: 8137958), CD45RA (BD Biosciences, Cat:563733; Clone: HI100; Lot: 8073653), CCR7 (BD Biosciences, Cat:562555; Clone: 150503; Lot: 7355865) and FITC-labeled human BCMA protein (ACRO Biosystems, Cat: BCA-HF254; Lot: FL894-20BHF1-VK). All the data were collected using an LSRII flow cytometer (Becton Dickinson, Franklin Lakes, NJ, USA) and were analyzed with FlowJo software V10 (TreeStar, Ashland, OR, USA). Gating strategy can be found in Supplementary Fig. 11.

Serum cytokine profiling

Magnetic Luminex Assay kits were purchased from R&D Systems (catalog number: FCSTM18-45 and LXSAHM-16). 61 interleukins, interferons, chemokines, TNFs and soluble receptors/ligands were detected and collectively called "cytokines" throughout the context. In detail, they were CCL2, CCL3, CCL4, CCL5, CCL11, CCL19, CCL20, CX3CL1, CXCL1, CXCL2, CXCL10, IL-1α, IL-1β, IL-1ra, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-8, IL-10, IL-12 p70, IL-13, IL-15, IL-17A, IL-17E, IL-33, EGF, FGF basic, VEGF, PDGF-AA, PDGF-AB/BB, TGF-α, G-CSF, GM-CSF, sPD-L1, sCD40 Ligand, sFLT-3Ligand, IFN-α, IFN-β, IFN-γ, TNF-α, TRAIL, Granzyme B, sIL-2Rα, CXCL9, gp130, sIL-1R II, sIL-6Rα, sTNF RI, sTNF RII, RAGE, angiopoietin-1, angiopoietin-2, IL-18, CCL22, sBCMA, sVEGF R1, sVEGF R2, sVEGF R3. Samples cryopreserved at −80 °C were thawed and processed according to the manufacturers’ protocols. Assay plates containing patient serum were placed in the Luminex X-200 instrument which provided a platform for the tests. Data processing was carried out by Milliplex analyst software based on five-parameter logistic curve-fit. All data were subject to strict quality control adhering to the following criteria: the standard curve for each analysis had a R² value > 0.99, and the coefficient of variation of standard curve and sample replicates were below 20%.

CONNECTOR analysis

Since the CC4 and CC5 exhibit the most robust correlations with CRS severity, we adopted the computational methodology CONNECTOR to stratify patients based on their affiliation with CC4 and CC5. Utilizing the mean log₂FC as the observation, we have performed 1,000 iterations using the ClusterAnalysis function within CONNECTOR. This analysis effectively partitions the 26 patients into two distinct groups: Group1 (comprising 8 patients: CZ-02, CZ-03, JS-06, RJ-01, RJ-02, RJ-04, RJ-22, RJ-30) and Group2 (comprising 18 patients: CZ-01, JS-02, JS-03, JS-04, JS-05, JS-07, JS-08, JS-09, RJ-03, RJ-05, RJ-23, RJ-24, RJ-25, RJ-26, RJ-27, RJ-29, RJ-31, RJ-32).

Nucleic acid extraction and RNA sequencing (RNA-seq)

Total RNA and genomic DNA were isolated from PBMCs, whole blood or liver tissue using QIAamp^R DNA Mini Kit (Qiagen) or RNeasy Micro Kit (Qiagen). The quality of RNA/DNA was firstly verified by Agilent Bioanalyzer 2100.

RNA-seq libraries were prepared according to protocol and then sequenced on NovaSeq 6000 platform (Illumina). Raw FASTQ files were aligned to human reference genome GRCh38 (release 37). The human reference genome and its annotation file were downloaded from GENCODE database (https://www.gencodegenes.org/). We used salmon (v1.3.1)⁴⁷ to generate the count and transcripts per kilobase of exon model per million mapped reads (TPM) matrix. The transcript counts were then merged using DESeq2⁴⁸ and transformed as fragments per kilobase million (FPKM) to evaluate the gene expression level by normalizing the length of genes using the TPM matrix. To ensure the robustness of our results and identify DEGs, three well-established and recommended methods limma (v3.44.3), edgeR (v3.30.3), DESeq2 (v1.28.1) were applied. The analysis workflow for DESeq2, limma and edgeR, which were constructed based on the linear model, followed the guidelines provided in the respective software packages. Significant DEGs were identified using adjusted p < 0.05. And significant DEGs between any two time points were included for further soft clustering analysis. As demonstrated in Supplementary Fig. 3a, the majority of DEGs were consistently identified by all three methods, indicating a high level of confidence in our DEGs selection process. Details of analysis workflow were described in Supplementary Fig. 3a legend and “Code availability”.

Soft clustering method

The approach of soft clustering was implemented using the fuzzy c-means algorithm by the Mfuzz R software package (v2.48.0), aiming to assign the genes into different clusters in a gradual manner according to their expression values⁴⁹. After standardization, genes were assigned a unique cluster by membership score (α). The gene with α score greater than 0.6 was considered as a core gene and was used for subsequent functional enrichment analysis. Details of analysis workflow were described in Supplementary Fig. 3b legend and “Code availability”.

Gene ontology (GO) analysis and gene set enrichment analysis (GSEA)

The functional enrichment analysis of GO was performed with clusterprofiler R package (v3.16.1). And GSEA was conducted based on molecular signatures database v7.4 and GSEABase R package (v1.50.1). Gene sets used in GSEA were downloaded from the Molecular Signatures Database (MSigDB, v7.4) of the Broad Institute.

The genes used for GSEA were selected from the FPKM profile. Genes with low expression (expressed in less than 30% of the samples) were excluded. During the GSEA analysis at various observation time points, genes were pre-ranked based on their log₂FC, which was calculated by comparing their expression levels with the baseline. Pathway direction is the median log₂FC of significant transcripts (core enrichment genes) relative to the baseline. In this visualization, an adjusted p < 0.05 with pathway direction > 0 indicated activation, while an adjusted p < 0.05 with pathway direction <0 indicated repression.

T cell receptor repertoire analysis

The inverse Simpson index (1/D)⁵⁰ and clonality index⁵¹ were calculated as reported to quantitatively define the diversity of T cell clones.

The amino acid sequences of CDR3 localized on TRB and TRA were in a perfect match with the TCR-antigen database developed by Immuquad^@ biotech, respectively. The screening method was carried out based on the algorithm of Levenshtein distance⁵² ≤ 3, along with V and J genes exactly matched with the ones in the database.

Virus sequences alignment and preprocessing

We have download 13,778 virus genomes from National Center for Biotechnology Information (NCBI, https://ftp.ncbi.nlm.nih.gov/) and built an in-house viral database. To reduce the impact of genomic homology between human and virus, a human-virus genome was built and we used two methods STAR (v2.7.9a)⁵³ and salmon (v1.3.1)⁴⁷ to calculate the virus transcripts from RNA-seq data.

Mutations calling and screening

To identify mutations from RNA-seq data, raw reads were aligned to human reference using STAR (v2.7.9a)⁵³, which was downloaded from UCSC Genome Browser (http://hgdownload.soe.ucsc.edu/). SAMtools (v1.8-47) was used to generate name-sorted and indexed BAM files and the PICARD tool (https://broadinstitute.github.io/picard/) was used to mark duplicate reads in BAM files. The mutation calling steps followed the Genome Analysis Toolkit (GATK, v4.1.8)⁵⁴ forum recommended best-practice pipeline. The steps were described as followed: (i) two software, GATK HaplotypeCaller and VarScan2 (v2.4.4), were used to generate mutations including single-nucleotide variations (SNVs) and small insertions/deletions (INDELs); (ii) mutations were merged together and annotated by ANNOVAR⁵⁵ using specific databases such as snp138, avsnp147, avsnp150, COSMIC, SIFT and PolyPhen. (iii) RNAmut was used for verification mutation sites from RNA-seq data; (iv) the filtration conditions were ≥ 10x depth in the variant sites, ≥ 3% variant allele frequency (VAF) and > 3 individual mutant reads, not a SNP and annotated by COSMIC database or predicted as damaging by ANNOVAR.

Lentiviral vector integration site analysis

Integration site analysis was implemented based on standard ligation target amplification PCR and next-generation sequencing. Briefly, 500 ng of genomic DNA was sheared to an intermediate length of 500 ± 50 bp using the Covaris M220 Gene Fragmentation Instrument. After purification, the sheared DNA was extended by biotinylated 3′-long terminal repeat (3′-LTR)-specific primer (Primer 1) and captured by streptavidin-coated beads (Dynabeads^TMM-280; invitrogen). The captured genome was seeded to a ligation cassette including a sequencing barcode (UMI) according to the kit (Fast-Link Ligation Kit; Epicentre). Amplification of integration site was conducted via a procedure of two nested PCR. The first exponential PCR primers were referred to as Primer 2 and Primer 3. After magnetic capturing, the product underwent the second exponential PCR using Primer 4 (PE 2.0 UMI short). Primer sequences are summarized in the SupplemetaryTable 1.

The purified product proceeded to the subsequent Illumina Miseq next-generation sequencing platform for paired-end reads with parameter settings referenced in the literature^56,57. Raw sequencing data were analyzed with Genome Integration Site Analysis Pipeline⁵⁸ based on human reference genome GRCh38.

Single-cell RNA sequencing (scRNA-seq)

CAR-positive cell populations from RJ-31’s PBMCs on day 11 and day 28 were obtained by flow cytometry cell sorting (FACSAria^TM III BD Biosciences) with FITC-labeled human BCMA protein (ACRO Biosystems). The target cells underwent single cell isolation to form Gel Bead in Emulsion by 10x Genomics Chrominum^TM. And then scRNA-seq libraries were performed according to the Chromium Single Cell 5′ Library Construction Kit and Chromium Single Cell Human TCR Amplification Kit (10x Genomics). After quality verification, the libraries were sequenced on NovaSeq 6000 platform (Illumina).

The raw sequencing reads in FASTQ format of single-cell RNA sequencing (scRNA-seq) data were aligned to the human GRCh38 reference (2020-A version) and the gene expression matrices were generated using CellRanger (10X Genomics, default settings, Version 6.0.2). Both CellRanger software and the reference were downloaded from 10X Genomics website (https://www.10xgenomics.com/). And for V(D)J T-cell analysis, the single-cell T cell receptor (TCR) sequencing (scTCR-seq) data were aligned to human GRCh38 (ensembl-5.0.0) using the vdj module in CellRanger. The gene expression matrices generated by CellRanger were then imported and merged using the Seurat R package⁵⁹ for downstream data analysis. Cells that expressed less than 800 genes or over 10% mitochondrial RNA were filtered out. Cells which matched with TCR were identified as T cells and for further study. Top 3,000 highly variable genes were identified to perform principal component analysis (PCA) and ComBat method in R package SVA⁶⁰ was used to correct batch effects. UMAP (uniform manifold approximation and projection) was chosen to further cell visualization. Marker genes of each cluster were calculated using ‘FindAllMarkers’ function in Seurat. Cell types were identified and annotated by the following criteria: (i) annotation generated in SingleR⁶¹ was used as reference, (ii) manually annotated using the top expressed genes in each cluster, (iii) clone type identified by scTCR-seq.

To compare the toxicity capacity of cells, we calculated the average expression level of toxicity-related genes for each cell to quantify cytotoxicity, called cytotoxic score. Based on our own scRNA-seq data, we computed the Pearson correlation coefficients of all genes with GZMB and GNLY separately. And then intersection of the top 20 ranked genes significantly associated with GZMB and GNLY were defined as Gzmb_Gnly_related gene set. Meanwhile, we extracted the reported toxicity-related gene set⁶² as a supplement.

Quantitative real-time polymerase chain reaction (qPCR)

To measure the copy number of the transgene in the Cilta-cel, qPCR of targeted fragment was subsequently performed as reported previously³.

And the transcriptional levels of the gene encoding herpesvirus7 (HHV-7) and ETS1 were measured by RT-qPCR. 500 ng of RNA was reverse-transcribed into cDNA using HiScript III RT SuperMix kit (Vazyme), and ChamQ SYBR qPCR Master Mix kit (Vazyme) was applied for qPCR of the genetic segment of interest. The procedure was run and analyzed on ViiA^TM 7 system (Life technologies). Gene transcriptional level was assessed using cycle threshold and normalized to a housekeeping gene GAPDH. The sequences of the primers were as follows: GAPDH forward 5′-GTCTCCTCTGACTTCAACAGCG-3′ and reverse 5′-ACCACCCTGTTGCTGTAGCCA-3′, HHV-7 encoded gene forward 5′-AGAGGTTGCTTCTTCGAGTCA-3′ and reverse 5′-GCGCTTGTCAAAATCCTCAAAT-3′, ETS1 encoded gene forward 5′-GGAATGTGCAGATGTCCCAC-3′ and reverse 5′-CATTCACAGCCCACATCACC-3′.

Sanger sequencing

The PCR was performed using Phanta Max Super-Fidelity DNA Polymerase kit (Vazyme, P505-d1). The sequences of the primers of TET2 were forward 5′-GTCTCAGCCGATGGATCTGTA-3′ and reverse 5′- ATCTGTTGTAAGGCCCTGTGA-3′.

Targeted DNA sequencing

Targeted DNA sequencing was performed on the gene coding sequence. Library enrichment was carried using NadPrep DNA Preparation Kit (for Illumina) and then sequenced on NovaSeq 6000 platform (Illumina). Variant calling from targeted DNA sequencing was performed as reported previously⁶³.

Droplet digital PCR (ddPCR)

To identify the low-frequency mutation in TET2, ddPCR was performed in RJ-31’s genome DNA at different time points. To make the assay more convinced, water and three healthy controls were included as blank and negative controls, respectively. ddPCR was accomplished on a Bio-Rad QX200 system using ddPCR supermix for Probe (No dUTP) (Bio-Rad) and analyzed on QuantaSoft (Version 1.7.4). The primers and probe sequences targeting TET2 were forward, 5′-GTCCAAGGAGGCTTACACAAAT-3′; reverse, 5′-ACCTCATCGTTGTCCTCTGC-3′; wild probe, 5′VIC-CACACCCTGGACTAG-3′MGB; mutate probe 5′FAM-CACACCCTAGACTAGT-3′MGB. To ensure the quality of the experiment, three replicates were completed for each sample, and each reaction was guaranteed to have a droplet count of more than 10,000. \({{{{{\rm{VAF}}}}}}=\frac{{{{{{\rm{Concentrate}}}}}}\; {{{{{\rm{of}}}}}}\; {{{{{\rm{mutate}}}}}}}{{{{{{\rm{Concentrate}}}}}}\; {{{{{\rm{of}}}}}}\; {{{{{\rm{mutate}}}}}}\; {{{{{\rm{and}}}}}}\; {{{{{\rm{wild}}}}}}}\)

Statistical analysis

Data analyses were performed using R version 4.0.3 or software GraphPad Prism version 8.0.2. Statistical tests were described in the figure legends.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Raw sequencing data generated during this study have been deposited in the Genome Sequence Archive in National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Science (https://ngdc.cncb.ac.cn/gsa-human), with accession number ‘GSA-Human: HRA005381’ [https://ngdc.cncb.ac.cn/gsa-human/browse/HRA005381]. These data are under controlled use conditions set by human privacy regulations, only available upon reasonable request. Access can be obtained by approval via the Data Access Committee of the GSA-human database (https://ngdc.cncb.ac.cn/gsa-human/document/GSA-Human_Request_Guide_for_Users_us.pdf). Original data for graphs is provided in the Source Data file. Source data are provided with this paper.

Code availability

Code used in the study is available on GituHub (https://nrctm-bioinfo.github.io/MM_CART_project/index.html) and has been assigned a (https://doi.org/10.5281/zenodo.10200875).

References

Maude, S. L. et al. Tisagenlecleucel in children and young adults with B-cell lymphoblastic leukemia. N. Engl. J. Med. 378, 439–448 (2018).
Article CAS PubMed Central Google Scholar
Neelapu, S. S. et al. Axicabtagene ciloleucel CAR T-cell therapy in refractory large B-cell lymphoma. N. Engl. J. Med. 377, 2531–2544 (2017).
Article CAS PubMed PubMed Central Google Scholar
Xu, J. et al. Exploratory trial of a biepitopic CAR T-targeting B cell maturation antigen in relapsed/refractory multiple myeloma. Proc. Natl Acad. Sci. USA 116, 9543–9955 (2019).
Article CAS PubMed Central ADS Google Scholar
Teachey, D. T. et al. Identification of predictive biomarkers for cytokine release syndrome after chimeric antigen receptor T-cell therapy for acute lymphoblastic leukemia. Cancer Discov. 6, 664–679 (2016).
Article CAS PubMed PubMed Central Google Scholar
Norelli, M. et al. Monocyte-derived IL-1 and IL-6 are differentially required for cytokine-release syndrome and neurotoxicity due to CAR T cells. Nat. Med. 24, 739–748 (2018).
Article CAS PubMed Google Scholar
Giavridis, T. et al. CAR T cell-induced cytokine release syndrome is mediated by macrophages and abated by IL-1 blockade. Nat. Med. 24, 731–738 (2018).
Article CAS PubMed PubMed Central Google Scholar
Sterner, R. M. et al. GM-CSF inhibition reduces cytokine release syndrome and neuroinflammation but enhances CAR-T cell function in xenografts. Blood 133, 697–709 (2019).
Article CAS PubMed PubMed Central Google Scholar
Zhao, W. H. et al. Four-year follow-up of LCAR-B38M in relapsed or refractory multiple myeloma: a phase 1, single-arm, open-label, multicenter study in China (LEGEND-2). J. Hematol. Oncol. 15, 86 (2022).
Article CAS PubMed PubMed Central Google Scholar
Mi, J. Q., et al. Phase II, open-label study of ciltacabtagene autoleucel, an anti-B-cell maturation antigen chimeric antigen receptor-T-cell therapy, in Chinese patients with relapsed/refractory multiple myeloma (CARTIFAN-1). J. Clin. Oncol. 41,Jco2200690 (2022).
Berdeja, J. G. et al. Ciltacabtagene autoleucel, a B-cell maturation antigen-directed chimeric antigen receptor T-cell therapy in patients with relapsed or refractory multiple myeloma (CARTITUDE-1): a phase 1b/2 open-label study. Lancet 398, 314–324 (2021).
Article CAS PubMed Google Scholar
Ri, M. et al. Ciltacabtagene autoleucel in patients with relapsed/refractory multiple myeloma: CARTITUDE-1 (phase 2) Japanese cohort. Cancer Sci. 113, 4267–4276 (2022).
Article CAS PubMed PubMed Central Google Scholar
Maecker, H. T., McCoy, J. P. & Nussenblatt, R. Standardizing immunophenotyping for the Human Immunology Project. Nat. Rev. Immunol. 12, 191–200 (2012).
Article CAS PubMed Central Google Scholar
Hay, K. A. et al. Kinetics and biomarkers of severe cytokine release syndrome after CD19 chimeric antigen receptor-modified T-cell therapy. Blood 130, 2295–2306 (2017).
Article CAS PubMed PubMed Central Google Scholar
Pernice, S. et al. CONNECTOR, fitting and clustering of longitudinal data to reveal a new risk stratification system. Bioinformatics 39, btad201 (2023).
Article CAS PubMed PubMed Central Google Scholar
Casulleras, M., Zhang, I. W., López-Vicario, C. & Clària, J. Leukocytes, systemic inflammation and immunopathology in acute-on-chronic liver failure. Cells 9, 2632 (2020).
Article CAS PubMed PubMed Central Google Scholar
Tillie, R. et al. A switch from cell-associated to soluble PDGF-B protects against atherosclerosis, despite driving extramedullary hematopoiesis. Cells 10, 1746 (2021).
Article CAS PubMed PubMed Central Google Scholar
Thomas, M. & Augustin, H. G. The role of the Angiopoietins in vascular morphogenesis. Angiogenesis 12, 125–137 (2009).
Article CAS PubMed Google Scholar
Boisramé-Helms, J., Kremer, H., Schini-Kerth, V. & Meziani, F. Endothelial dysfunction in sepsis. Curr. Vasc. Pharmacol. 11, 150–160 (2013).
Google Scholar
Gavriilaki, E., Sakellari, I., Gavriilaki, M. & Anagnostopoulos, A. A new era in endothelial injury syndromes: toxicity of CAR-T cells and the role of immunity. Int. J. Mol. Sci. 21, 3886 (2020).
Article CAS PubMed PubMed Central Google Scholar
Hay, K. A. et al. Factors associated with durable EFS in adult B-cell ALL patients achieving MRD-negative CR after CD19 CAR T-cell therapy. Blood 133, 1652–1663 (2019).
Article CAS PubMed PubMed Central Google Scholar
Page, A. V. & Liles, W. C. Biomarkers of endothelial activation/dysfunction in infectious diseases. Virulence 4, 507–516 (2013).
Article PubMed Central Google Scholar
Pan, J. et al. Ruxolitinib mitigates steroid-refractory CRS during CAR T therapy. J. Cell. Mol. Med. 25, 1089–1099 (2021).
Article CAS PubMed Google Scholar
Wei, S. et al. Adjuvant ruxolitinib therapy relieves steroid-refractory cytokine-release syndrome without impairing chimeric antigen receptor-modified T-cell function. Immunotherapy 12, 1047–1052 (2020).
Article CAS PubMed Google Scholar
Chen, Y. et al. Therapeutic potential of TNFα and IL1β blockade for CRS/ICANS in CAR-T therapy via ameliorating endothelial activation. Front. Immunol. 12, 623610 (2021).
Article CAS PubMed PubMed Central Google Scholar
Fajgenbaum, D. C. & June, C. H. Cytokine storm. N. Engl. J. Med. 383, 2255–2273 (2020).
Article CAS PubMed PubMed Central Google Scholar
Wang, X. et al. Quantitative characterization of T-cell repertoire alteration in Chinese patients with B-cell acute lymphocyte leukemia after CAR-T therapy. Bone Marrow Transplant. 54, 2072–2080 (2019).
Article CAS PubMed Google Scholar
Turtle, C. J. et al. CD19 CAR-T cells of defined CD4+:CD8+ composition in adult B cell ALL patients. J. Clin. Investig. 126, 2123–2138 (2016).
Article PubMed Central Google Scholar
Busque, L. et al. Recurrent somatic TET2 mutations in normal elderly individuals with clonal hematopoiesis. Nat. Genet. 44, 1179–1181 (2012).
Article CAS PubMed PubMed Central Google Scholar
Jaiswal, S. et al. Age-related clonal hematopoiesis associated with adverse outcomes. N. Engl. J. Med. 371, 2488–2498 (2014).
Article PubMed Central Google Scholar
Cong, B., Zhang, Q. & Cao, X. The function and regulation of TET2 in innate immunity and inflammation. Protein cell 12, 165–173 (2021).
Article CAS Google Scholar
Siraki, A. G. The many roles of myeloperoxidase: from inflammation and immunity to biomarkers, drug metabolism and drug discovery. Redox Biol. 46, 102109 (2021).
Article CAS PubMed PubMed Central Google Scholar
Barnes, B. J. et al. Targeting potential drivers of COVID-19: Neutrophil extracellular traps. J. Exp. Med. 217, e20200652 (2020).
Article CAS PubMed PubMed Central Google Scholar
Middleton, E. A. et al. Neutrophil extracellular traps contribute to immunothrombosis in COVID-19 acute respiratory distress syndrome. Blood 136, 1169–1179 (2020).
Article CAS PubMed Google Scholar
Socié, G. et al. Prognostic value of apoptotic cells and infiltrating neutrophils in graft-versus-host disease of the gastrointestinal tract in humans: TNF and Fas expression. Blood 103, 50–57 (2004).
Article PubMed Google Scholar
Wang, D. et al. A phase 1 study of a novel fully human BCMA-targeting CAR (CT103A) in patients with relapsed/refractory multiple myeloma. Blood 137, 2890–2901 (2021).
Article CAS PubMed Google Scholar
Bernabei, L. et al. B-cell maturation antigen chimeric antigen receptor T-cell re-expansion in a patient with myeloma following salvage programmed cell death protein 1 inhibitor-based combination therapy. Br. J. Haematol. 193, 851–855 (2021).
Article CAS PubMed Google Scholar
Sun, Z., Xie, C., Liu, H. & Yuan, X. CD19 CAR-T cell therapy induced immunotherapy associated interstitial pneumonitis: a case report. Front. Immunol. 13, 778192 (2022).
Article CAS PubMed PubMed Central Google Scholar
Zhao, W. H. et al. A phase 1, open-label study of LCAR-B38M, a chimeric antigen receptor T cell therapy directed against B cell maturation antigen, in patients with relapsed or refractory multiple myeloma. J. Hematol. Oncol. 11, 141 (2018).
Article MathSciNet CAS PubMed PubMed Central Google Scholar
Tanyi, J. L. et al. Possible compartmental cytokine release syndrome in a patient with recurrent ovarian cancer after treatment with mesothelin-targeted CAR-T cells. J. Immunother. 40, 104–107 (2017).
Article CAS Google Scholar
Handley, G. et al. Human herpesvirus 6 myelitis after chimeric antigen receptor T-cell therapy. Int. J. Infect. Dis. 112, 327–329 (2021).
Article CAS PubMed Google Scholar
Wang, D. et al. Viral infection/reactivation during long-term follow-up in multiple myeloma patients with anti-BCMA CAR therapy. Blood Cancer J. 11, 168 (2021).
Article PubMed PubMed Central Google Scholar
Jaiswal, S. Clonal hematopoiesis and nonhematologic disorders. Blood 136, 1606–1614 (2020).
PubMed PubMed Central Google Scholar
Arends, C. M. et al. Hematopoietic lineage distribution and evolutionary dynamics of clonal hematopoiesis. Leukemia 32, 1908–1919 (2018).
Article CAS PubMed Google Scholar
Fraietta, J. A. et al. Disruption of TET2 promotes the therapeutic efficacy of CD19-targeted T cells. Nature 558, 307–312 (2018).
Article CAS PubMed PubMed Central ADS Google Scholar
Kumar, S. et al. International Myeloma Working Group consensus criteria for response and minimal residual disease assessment in multiple myeloma. Lancet Oncol. 17, e328–e346 (2016).
Article PubMed Google Scholar
Neelapu, S. S. et al. Chimeric antigen receptor T-cell therapy - assessment and management of toxicities. Nat. Rev. Clin. Oncol. 15, 47–62 (2018).
Article CAS PubMed Google Scholar
Patro, R., Duggal, G., Love, M. I., Irizarry, R. A. & Kingsford, C. Salmon provides fast and bias-aware quantification of transcript expression. Nat. Methods 14, 417–419 (2017).
Article CAS PubMed Central Google Scholar
Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).
Article PubMed Central Google Scholar
Kumar, L. & M, E. F. Mfuzz: a software package for soft clustering of microarray data. Bioinformation 2, 5–7 (2007).
Article PubMed Central Google Scholar
Yew, P. Y. et al. Quantitative characterization of T-cell repertoire in allogeneic hematopoietic stem cell transplant recipients. Bone Marrow Transplant. 50, 1227–1234 (2015).
Article CAS PubMed Central Google Scholar
Zhang, L. et al. 3D: diversity, dynamics, differential testing - a proposed pipeline for analysis of next-generation sequencing T cell repertoire data. BMC Bioinform. 18, 129 (2017).
Article Google Scholar
Young, B., Faris, T. & Armogida, L. Levenshtein distance as a measure of accuracy and precision in forensic PCR-MPS methods. Forensic Sci. Int. Genet. 55, 102594 (2021).
Article CAS PubMed Google Scholar
Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).
Article CAS PubMed Google Scholar
McKenna, A. et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res 20, 1297–1303 (2010).
Article CAS PubMed PubMed Central Google Scholar
Wang, K., Li, M. & Hakonarson, H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 38, e164 (2010).
Article PubMed PubMed Central Google Scholar
Gabriel, R. et al. Comprehensive genomic access to vector integration in clinical gene therapy. Nat. Med. 15, 1431–1436 (2009).
Article CAS PubMed Google Scholar
Paruzynski, A. et al. Genome-wide high-throughput integrome analyses by nrLAM-PCR and next-generation sequencing. Nat. Protoc. 5, 1379–1395 (2010).
Article CAS PubMed Google Scholar
Afzal, S., Wilkening, S., von Kalle, C., Schmidt, M. & Fronza, R. GENE-IS: time-efficient and accurate analysis of viral integration events in large-scale gene therapy data. Mol. Ther. Nucleic Acids 6, 133–139 (2017).
Article CAS Google Scholar
Butler, A., Hoffman, P., Smibert, P., Papalexi, E. & Satija, R. Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat. Biotechnol. 36, 411–420 (2018).
Article CAS PubMed Central Google Scholar
Leek, J. T., Johnson, W. E., Parker, H. S., Jaffe, A. E. & Storey, J. D. The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics 28, 882–883 (2012).
Article CAS PubMed PubMed Central Google Scholar
Aran, D. et al. Reference-based analysis of lung single-cell sequencing reveals a transitional profibrotic macrophage. Nat. Immunol. 20, 163–172 (2019).
Article CAS PubMed PubMed Central Google Scholar
Tirosh, I. et al. Dissecting the multicellular ecosystem of metastatic melanoma by single-cell RNA-seq. Science 352, 189–196 (2016).
Article CAS PubMed Central ADS Google Scholar
Cheng, W. Y. et al. Transcriptome-based molecular subtypes and differentiation hierarchies improve the classification framework of acute myeloid leukemia. Proc. Natl Acad. Sci. USA 119, e2211429119 (2022).
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

This work was supported by the State Key Laboratory of Medical Genomics, the Double First-Class Project (WF510162602) from the Ministry of Education; the Shanghai Collaborative Innovation Program on Regenerative Medicine and Stem Cell Research (2019CXJQ01); the Overseas Expertise Introduction Project for Discipline Innovation (111 Project;B17029); the National Natural Science Foundation of China (Nos. 82230006, 81970189, 82070227); the Shanghai Clinical Research Center for Hematological disease (19MC1910700); the Shanghai Shenkang Hospital Development Center (SHDC2020CR5002, SHDC2020CR2066B); the Shanghai Major Project for Clinical Medicine (2017ZZ01002); the Innovative Research Team of High-level Local Universities in Shanghai; the Shanghai Pujiang Program (No. 22PJ1409800).

Author information

These authors contributed equally: Shuangshuang Yang, Jie Xu, Yuting Dai.

Authors and Affiliations

Shanghai Institute of Hematology, State Key Laboratory of Medical Genomics, National Research Center for Translational Medicine at Shanghai, Ruijin Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, 200025, China
Shuangshuang Yang, Jie Xu, Yuting Dai, Shiwei Jin, Yan Sun, Jianfeng Li, Chenglin Liu, Xiaolin Ma, Zhu Chen, Jian-Qing Mi & Sai-Juan Chen
Department of Hematology, First affiliated Hospital of Nanjing Medical University, Jiangsu Province Hospital, Nanjing, 210029, China
Lijuan Chen
Department of Hematology, Ren Ji Hospital affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, 200127, China
Jian Hou

Authors

Shuangshuang Yang
View author publications
You can also search for this author in PubMed Google Scholar
Jie Xu
View author publications
You can also search for this author in PubMed Google Scholar
Yuting Dai
View author publications
You can also search for this author in PubMed Google Scholar
Shiwei Jin
View author publications
You can also search for this author in PubMed Google Scholar
Yan Sun
View author publications
You can also search for this author in PubMed Google Scholar
Jianfeng Li
View author publications
You can also search for this author in PubMed Google Scholar
Chenglin Liu
View author publications
You can also search for this author in PubMed Google Scholar
Xiaolin Ma
View author publications
You can also search for this author in PubMed Google Scholar
Zhu Chen
View author publications
You can also search for this author in PubMed Google Scholar
Lijuan Chen
View author publications
You can also search for this author in PubMed Google Scholar
Jian Hou
View author publications
You can also search for this author in PubMed Google Scholar
Jian-Qing Mi
View author publications
You can also search for this author in PubMed Google Scholar
Sai-Juan Chen
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

S.-J.C., Z.C., J.-Q.M. and S.Y. designed the project. J.-Q.M., J.X., S.J., L.C., J.H. performed the clinical trial. S.Y., C.L., X.M., Y.S. performed the experiments. S.Y., Y.D., J.L. provided the bioinformatics analysis. S.-J.C., Z.C., J.-Q.M., S.Y., J.X. supervised the study and wrote the manuscript. All authors have read and approved the final version of this manuscript.

Corresponding authors

Correspondence to Jian-Qing Mi or Sai-Juan Chen.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Francesca Cordero, Nicolaus Kroeger and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Description of Additional Supplementary Files

Supplementary Data 1

Supplementary Data 2

Supplementary Data 3

Reporting Summary

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Yang, S., Xu, J., Dai, Y. et al. Neutrophil activation and clonal CAR-T re-expansion underpinning cytokine release syndrome during ciltacabtagene autoleucel therapy in multiple myeloma. Nat Commun 15, 360 (2024). https://doi.org/10.1038/s41467-023-44648-3

Download citation

Received: 21 February 2023
Accepted: 21 December 2023
Published: 08 January 2024
DOI: https://doi.org/10.1038/s41467-023-44648-3

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.