Tocilizumab for the treatment of chimeric antigen receptor T cell-induced cytokine release syndrome

Abstract

Introduction:

Cancer-directed immunotherapies are transforming the landscape in oncology as new and exciting therapies move from the laboratory to the bedside. Chimeric antigen receptor T (CAR-T) cells are one of these novel therapies, demonstrating impressive efficacy against B-cell malignancies. With the development of new therapies, it is not uncommon to identify new and unanticipated toxicities. CAR-T cells cause unique toxicities not typically found with traditional cytotoxic chemotherapy or small molecule inhibitors.

Areas Covered:

CAR-T cell associated toxicities include cytokine release syndrome (CRS) and CAR-T cell-related encephalopathy syndrome (CRES), alternatively known as immune effector cell-associated neurotoxicity syndrome (ICANS). Prompt identification and management of CRS and CRES is imperative for the prevention of life-threatening complications of these innovative therapies. This literature review describes the seminal trials of CD19 directed immunotherapy and the pathophysiology and management of the toxicities found with CAR-T cells. In addition, the use of the interleukin-6 receptor antibody tocilizumab for CRS is reviewed.

Expert Opinion:

This review describes the recommended management of CRS and CRES and examines the current limitations in management. Alternative therapies for the treatment of CAR-T cell related toxicities are also explored. Furthermore, the review proposes future directions for research.

1. Introduction

Over the past decade, anti-cancer immunotherapeutic approaches have transformed the landscape in oncology as new therapies have moved from the bench to the bedside. In 2017 alone, three novel immunotherapies received full Food and Drug Administration (FDA) and/or European Medicines Agency (EMA) approval for the treatment of acute lymphoblastic leukemia (ALL). In contrast, only three new drugs received FDA approval for BCR-ABL1 negative ALL in the preceding 25 years. Broadly, cancer immunotherapy is the use of the immune system to detect and target malignancy. Multiple types of immunotherapy exist in clinical practice, including naked (unconjugated) and conjugated monoclonal antibodies, bispecific T-cell engaging antibodies (BiTEs), and chimeric antigen receptor T cells. Unconjugated monoclonal antibodies are the most frequently used form of cancer directed immunotherapy in current clinical practice. Unconjugated monoclonal antibodies work independently without attached drug or radioactive material through direct cell lysis and induction of apoptosis, complement-dependent cytotoxicity (CDC), or antibody-dependent cell-mediated cytotoxicity (ADCC) [1]. Examples of unconjugated monoclonal antibodies include alemtuzumab (anti-CD52) and rituximab (anti-CD20) (Table 1). Conjugated monoclonal antibodies, such as inotuzumab and brentuximab, are joined to a chemotherapeutic agent or radioactive particle which delivers the cytotoxic substance to the target antigen. Inotuzumab, for example, is a humanized monoclonal antibody (anti-CD22) bound covalently to calicheamicin dimethyl hydrazide [2]. Bi-specific T-cell engaging antibodies link CD3+ T effector cells with a target antigen resulting in lysis of the target cell. An example of a BiTE is blinatumomab, which links CD3+ T effector cells with CD19+ T cells with subsequent T cell engagement resulting in lysis of the target CD19+ cells [3].

Table 1.

Examples of Immunotherapies in Current Clinical Practice

Immunotherapy Class	Example	Target	Mechanism of Action
Monoclonal antibodies, unconjugated	Rituximab	Anti-CD20	Complement-mediated cytotoxicity and antibody-dependent cell-mediated cytotoxicity
Monoclonal antibodies, conjugated	Inotuzumab	Anti-CD22	Antibody binds to CD22 on tumor cell and induces cell death via cytotoxic calicheamicin-induced apoptosis
Bi-specific T-cell engaging antibodies	Blinatumomab	Links CD3+ T effector cells with CD19+ T cells	Directs unstimulated CD3+ T cells to CD19 expressing tumor cells and induces cytotoxicity
Chimeric Antigen Receptor T-cell	Tisagenlecleucel	CD19 with 4-1BB co-stimulatory domain	Target and induce cytotoxicity in CD19 expressing tumor cells
Chimeric Antigen Receptor Natural Killer Cell	Cord Blood Derived Natural Killer Cells	CD19	Target and induce cytotoxicity in CD19 expressing tumor cells

A promising form of immunotherapy, chimeric antigen receptor T cells (CAR T), uses genetically modified T cells to target cancer cells. Recently, multiple CAR T cell products have demonstrated impressive efficacy against historically difficult to treat malignancies. The basic structure of a CAR T cell usually comprises a tumor-targeting domain derived from a monoclonal antibody fused to a CD3 zeta chain which functions as the intracellular signaling domain. Second generation CARs also contain a co-stimulatory endodomain, either CD28 or 4–1BB. Examples of CD-19 directed CAR T therapies include tisagenlecleucel and axicabtagene ciloleucel, both of which have demonstrated efficacy in a number of early phase clinical trials.

Immunotherapies have unique toxicities not typically found with traditional cytotoxic chemotherapy or small molecule inhibitors. The timing and severity of toxicity differs between CAR constructs, the patient population treated, and disease. Patients treated with BiTEs and CAR T cells develop a number of toxicities that can be life threatening, including cytokine release syndrome (CRS) and CAR-related encephalopathy syndrome (CRES) alternatively known as immune effector cell-associated neurotoxicity syndrome (ICANS) [3,4]. Interleukin 6 (IL-6) is recognized as a key mediator of CRS, leading to the use of the IL-6 receptor inhibitor tocilizumab for the management of CRS. Tocilizumab is a humanized monoclonal antibody against both the soluble and membrane-bound IL-6 receptor. Tocilizumab initially received FDA approval in 2010 for the management of rheumatoid arthritis in adult patients and subsequently expanded to include the treatment of giant cell arteritis, juvenile polyarticular and systemic idiopathic arthritis [29]. The FDA approved the use of tocilizumab in the treatment of severe CRS in 2017, after multiple trials demonstrated its efficacy (Table 2) [4,5]. This review describes the seminal trials of CD19 directed immunotherapy, the risk factors for and pathophysiology of CRS/CRES, and the management of CRS/CRES with particular focus on the use of tocilizumab. In addition, this review examines the current limitations and gaps in the management of CRES.

Table 2.

CAR T Cell Constructs in Clinical Trials for B-cell Acute Lymphoblastic Leukemia, Incidence of Cytokine Release Syndrome, and Administration of Tocilizumab

Institution	Product Name/Sponsor	Clinical Trial Identifier	CAR T construct	Domain	Patient Population (Age Range in years)	Indication	Response	Incidence of CRS	Subjects Receiving Tocilizumab
CHOP/PENN [6]	CTL019/Novartis	NCT01626495, NCT01029366	CD19	4-1BB	5–60, median 14	r/r CD19+ B-ALL	CR 27/30	30/30	9/30 (4 received 2nd dose)
MSKCC [7]	JCAR015/JUNO	NCT01044069	CD19	CD28	23–74, median 44	r/r CD19+ B-ALL	CR 44/53	45/53	19/45
FHCRC [8]	FHCRC	NCT01865617	CD19	4-1BB	20–70, median 54	r/r CD19+ B-cell malignancies (ALL, CLL, NHL)	CR 31/33 (ALL specific subjects)	92/133 (all subjects)	21/133
Multi-center [5]	CTL019/Novartis	NCT02435849	CD19	4-1BB	3–23, median 11	r/r B-ALL	CR/Cri 61/75	58/75	28/58
NCI [10]	NCI	NCT02315612	CD22	4-1BB	7–30, median 19	r/r B-ALL	CR 12/21	16/21	1/21

Abbreviations: CHOP Children’s Hospital of Philadelphia; PENN University of Pennsylvania; MSKCC Memorial Sloane Kettering Cancer Center; FHCRC Fred Hutchinson Cancer Research Center, NCI National Cancer Institute; CAR chimeric antigen receptor; CRS cytokine release syndrome; B-ALL B-cell acute lymphocytic leukemia; r/r relapsed/refractory; CLL chronic lymphocytic leukemia; NHL non-hodgkin’s lymphoma; CR complete response; Cri complete response with incomplete hematologic recovery; N/R not reported

2. Efficacy of CAR T in B-ALL

Dramatic anti-tumor effect has been observed with CAR T cell therapy, specifically CD19-targeted CAR T cells in patients with ALL, chronic lymphocytic leukemia (CLL), and non-Hodgkin’s lymphomas (NHL). Multiple generations of CAR T cells now exist. First generation CARs failed to elicit a robust cytokine response and T cell expansion, necessitating the addition of a co-stimulatory domain in the second generation of CARs. The addition of the co-stimulatory endodomain improved expansion and persistence of CAR T cells [38]. Tisagenlecleucel (Kymriah, formerly CTL019) is a second generation CD19-targeted CAR developed by Novartis in collaboration with the Children’s Hospital of Philadelphia (CHOP) and University of Pennsylvania (Penn) which contains a 4–1BB co-stimulatory domain. In a seminal multi-center single cohort phase 2 study of 75 children with relapsed or refractory (r/r) B-ALL, the overall remission rate with tisagenlecleucel administration was 81% [5]. Moreover, the rate of overall survival (OS) at 12 months was 76%. Late relapses with tisagenlecleucel were associated with CD19-negative variants or loss of CAR T cell persistence [6]. An update to the published data was presented at the American Society for Hematology annual meeting reporting an OS at 18 months of 70%[39].

Davila and colleagues demonstrated similarly impressive results in a phase 1 clinical trial conducted at Memorial Sloan Kettering Cancer Center. Treatment of 16 adult patients with r/r B-ALL with 19–28z CAR T cell infusion demonstrated higher complete response rate (88%) than expected with salvage chemotherapy alone [7]. Subsequently, a total of 53 adult patients with r/r B-cell ALL who received 19–28z CAR T cells demonstrated CR in 83% of patients with a median overall survival of 12.9 months[8]. Similar studies have demonstrated remarkable efficacy with CAR T cells targeting CD22. 12 of 21 patients receiving CD22 with 41BB co-stimulatory domain CAR T cell construct achieved CR with 11 of 15 patients achieving CR who received >/= 1×10⁶ CD22 CAR per kilogram. Of note, 9 of 10 patients with CR had previously received CD19 directed immunotherapy suggesting that prior CD19 therapy does not decrease likelihood of response to CD22 CAR T therapy (Table 2) [10].

Third-generation CAR T cells consist of variable combinations of signaling domains such as 4–1BB, OX40 (CD134), ICOS, and CD27 with the goal to enhance T cell expansion and persistence thus increasing cytotoxic effect [40]. Allogeneic, “off the shelf” CAR T cell products that would not require individual collection and manufacture of autologous CAR T cell products are the next innovative step in immunotherapy. A current example of this is FT819, a TCR-less CD19 CAR T construct which in preclinical studies was shown to be effective against CD19 positive cancer cells [11]

In addition to CAR T cells, other T-cell engaging therapies have shown significant clinical responses in early phase clinical trials for r/r leukemia. In a phase 1/2 trial with blinatumomab, CR was achieved in 39% of pediatric patients with r/r B-cell ALL within the first two treatment cycles, with a median overall survival of 7.5 months [12]. A multi-institution phase 3 clinical trial demonstrated similarly promising results in the adult population with r/r ALL, with a higher rate of EFS than that with chemotherapy at 6 months (31% vs 12%, respectively) and a longer median duration of remission (7.3 vs. 4.6 months) [13].

3. Chimeric Antigen Receptor T Cell Associated Toxicities

3.1. Cytokine Release Syndrome (CRS)

The toxicity profile of immunotherapies differs from conventional cytotoxic chemotherapy. One of the most clinically significant and potentially life-threatening toxicities seen is CRS. CRS occurs due to the high level of immune activation of lymphocytes, macrophages, or myeloid cells with subsequent massive release of inflammatory cytokines. CRS occurs in association with CAR T cell therapy, bispecific T-cell engaging antibodies, and monoclonal antibodies (such as alemtuzumab and anti-PD-1/PD-L1 antibodies). However, CRS is most commonly seen with the T-cell engaging therapies [14,15]. The onset and severity of CRS is dependent on the immunologic agent and the degree of immune cell activation. For example, rituximab (anti-CD20) has been shown to induce CRS within hours of infusion, whereas CRS with T cell therapies generally occurs in days to weeks after infusion [16]. Increased disease burden is historically considered a risk factor for the development of CRS based on the observation that pediatric patients with B-ALL with higher baseline tumor burden have greater CAR-T cell expansion and higher severity of CRS [17]. A single center review of 133 adult patients with B-ALL who received CD19 CAR T cells identified independent predictors of CRS in addition to baseline tumor burden. Specifically, CRS was observed at a higher rate in patients who underwent lymphodepletion using cyclophosphamide and fludarabine, demonstrated thrombocytopenia prior to lymphodepletion, or received higher CAR T-cell dose [18]. These predictors have not been confirmed in malignancies other than B-ALL or with other CAR-T cell products in patients with B-ALL.

The clinical signs and symptoms of CRS involve multiple organ systems, with life-threatening complications including fluid-refractory hypotension and cardiac dysfunction, respiratory failure, coagulopathy, renal and liver failure. Fever is usually the first sign of CRS and develops prior to additional signs and symptoms [18]. Low grade CRS presents with a flu-like illness and patients often complain of fatigue, myalgia, and arthralgia. While the most common toxicity of CAR T cells is CRS, they have also been associated with anaphylaxis and infusion reactions [41]. Based on clinical experience with tisagenlecleucel, multiple groups have developed grading scales to distinguish mild, moderate, severe, and life-threatening CRS (Table 3). These scales apply to both early and delayed-onset CRS associated with T cell therapies and are focused on dividing patients into groups to help with management. While subtle differences exist between scales, they ultimately divide CRS into life-threatening and non-life threatening which directs therapeutic approach.

Table 3.

Cytokine Release Syndrome Grading Scales

Scale/Institution	Grade 1	Grade 2	Grade 3	Grade 4
CHOP/PENN	Mild reaction: treated with supportive care	Moderate reaction: some signs of organ dysfunction related to CRS and not attributable to any other condition. Hospitalization for management of CRS-related symptoms, including fevers with associated neutropenia, need for intravenous therapies (not including fluid resuscitation for hypotension)	More severe reaction: hospitalization required for management of symptoms related to organ dysfunction, including grade 4 LFTs or grade 3 creatinine related to CRS; includes hypotension treated with intravenous fluids or low-dose vasopressors, coagulopathy requiring FFP/cryoprecipitate, and hypoxia requiring supplemental oxygen	Life-threatening complications such as hypotension requiring high dose vasopressors, hypoxia requiring mechanical ventilation,
Lee Criteria	Non-life threatening, (fever, nausea, fatigue, headache, myalgia), symptomatic treatment only	Symptoms require and response to moderate intervention; supplemental oxygen requirement <40% or hypotension responsive to fluids or low dose of one vasopressor or Grade 2 organ toxicity (CTCAE v4.0 grading)	Symptoms require and response to aggressive intervention; supplemental oxygen requirement >40% or hypotension requiring high dose/multiple vasopressors or Grade 3 organ toxicity or Grade 4 transaminitis	Life-threatening symptoms requiring ventilator support or Grade 4 organ toxicity (excluding transaminitis)
CTCAE version 4.0	Mild reaction; infusion interruption not indicated; intervention not indicated	Therapy or infusion interruption indicated but responds promptly to symptomatic treatment (antihistamines, NSAID, fluids, narcotics) or prophylactic meds less than 24 hours	Prolonged reaction (not rapidly responsive to symptomatic medication and/or brief interruption of infusion); recurrence of symptoms following initial improvement; hospitalization indicated for clinical sequelae	Life-threatening consequences; pressor or ventilator support indicated
American Society for Blood and Marrow Transplantation	Temperature >/= 38.0 Celsius	Temperature >/= 38.0 Celsius, hypotension not requiring vasopressors, and/or hypoxia requiring low-flow nasal cannula or blow-by	Temperature >/= 38.0 Celsius, with either hypotension requiring one vasopressor with or without vasopressin, and/or hypoxia requiring high-flow nasal cannula, facemask, non-rebreather mask, or Venturi mask	Temperature >/= 38.0 Celsius with either hypotension requiring multiple vasopressors (excluding vasopressin), and/or hypoxia requiring positive pressure

Abbreviations: CHOP Children’s Hospital of Philadelphia, PENN University of Pennsylvania, CRS cytokine release syndrome, LFT liver function test, FFP fresh frozen plasma, CTCAE Common Terminology Criteria for Adverse Events, NSAID nonsteroidal anti-inflammatory drug

In CAR T cell related CRS, marked elevation of IL-6 is thought to be one main driver of symptoms. IL-6 is secreted by monocytes and macrophages and is induced with inflammation and infection. IL-6 stimulates the differentiation of T helper 17 cells and is a driver of the innate and acquired immunologic response. IL-6 and interferon-gamma (IFN-γ increase after tisagenlecleucel infusion in association with peak tisagenlecleucel (CTL019) transgene levels and remain elevated in CRS[18,19]. Some studies have demonstrated IL-6 rises significantly within the first few days prior to development of CRS and may predict patients who will develop severe CRS [18]. However, other studies using different CD19-directed CAR T cells have not found early rise of IL6 can predict severe CRS. While IL6 can be markedly elevated in the first few days prior to critical illness, this was inconsistent and marked elevations were also seen in patients who did not develop severe CRS [19].

In a cohort of patients at CHOP receiving tisagenlecleucel who developed grade 4 or 5 (severe) CRS, 43 distinct cytokines, chemokines, and soluble receptors (collectively called cytokines) were measured. Patients with severe CRS demonstrated elevated levels of specific cytokines as compared with those who did not develop severe CRS. Notable cytokines differentially elevated in severe CRS included IFN-γ, IL-6, IL-8, s-IL-1R, sIL-2R, soluble glycoprotein 130 (sgp130), sIL-6R, macrophage inflammatory protein (MIP1α), monocyte chemoattractant protein 1 (MCP1), and granulocyte-macrophage colony-stimulating factor (GM-CSF). Interestingly, only IFN-γ and sgp130 were significantly elevated within the first 72 hours after infusion in patients who would later develop severe CRS as compared with those who did not [19]. Based on an analysis of cytokines sent prior to the development of symptoms in a combined cohort of adult and pediatric patients with severe CRS (Grade 4–5) versus not severe (grade 1–3) CRS, elevated levels of IFN-γ, sgp130, and sIL-1R were able to predict the development of severe CRS with a sensitivity of 86% and specificity of 89%. Focusing on pediatric patients only, a forward-selected logistic regression model was able to predict with even better accuracy; a combination of IL-13, IFNγ, and MIP1α had a specificity of 96% and sensitivity of 100%. Disease burden alone has an excellent negative predictive value but poor positive predictive value, e.g. patients with low disease burden are highly unlikely to develop severe CRS but patients with high disease burden have an almost equal chance of developing or not developing severe CRS [19]. Combining the measurement of a single cytokine (IL10) in the first 72 hours after infusion with disease burden was highly sensitive (86%) and specific (100%) for predicting severe CRS.

Hay and colleagues evaluated biomarkers for early identification of patients at risk of developing severe CRS. In a cohort of 133 patients, 93 patients developed CRS with 10 of those developing grade 4 CRS. Patients who developed grade 4 CRS demonstrated higher concentration of IL-6, IL-8, IL-10, IL-15, INFγ, MCP-1, tumor necrosis factor receptor p55, and MIP-1β within 36 hours after CAR T infusion [17]. An algorithm using classification-tree modeling for early identification of patients at high risk of severe CRS was created based on this data; patients with fever greater than 38.9 degrees Celsius within 36 hours of CAR T infusion with a serum MCP-1 greater than 1343.5pg/mL was predictive of grade 4 CRS with a sensitivity of 100% and specificity of 95% [17]. Additional studies are needed to determine if these predictive models are specific for CD-19 directed CARs in B-ALL or whether or not they also help predict severity of CRS with different CAR targets or CAR structures and with different diseases.

Significant parallels exist between the clinical and laboratory parameters of CRS and hemophagocytic lymphohistiocytosis/macrophage activating syndrome (HLH/MAS). HLH is a pathophysiologic process of immune dysregulation resulting in excessive inflammation and abnormal activation of macrophages and T cells. Patients with hemophagocytic syndromes may be conceptually divided into three groups: patients with homozygous genetic mutations in genes involving cytolytic granule exocytosis who are predisposed to develop life-threatening HLH in the setting of minimal stressors; patients with underlying systemic inflammatory disease or immunodeficiency, where patients may have heterozygous mutations in HLH-causative genes and develop HLH in the setting of significant stressors such as cancer, infection, or rheumatic disease; and, patients with secondary HLH as a consequence of massive inflammation such as sepsis or T-cell engaging therapies who may not have any genetic predisposition. Patients with both CRS and HLH demonstrate liver dysfunction and transaminitis, hypofibrinogenemia and coagulopathy, and hyperferritinemia, in addition to elevation of similar cytokines including IL-6, INFγ, and IL-10 [19]. Laboratory analysis of patients treated with tisagenlecleucel with severe CRS all had a peak ferritin >10,000mg/dL, demonstrating hyperferritinemia to a value in which is considered sensitive for the diagnosis of HLH [4,19,43]. Similar to those with HLH, patients with CRS also develop hypofibrinogenemia (defined as <150mg/dL) with a strong association between severe CRS and low fibrinogen observed in a pediatric cohort receiving tisagenlecleucel. Patients with severe CRS also demonstrate elevations in INFγ and IL-10 that mirrors elevations of these key cytokines in HLH. Xu and colleagues previously demonstrated that an INFγ of greater than 75pg/mL and IL-10 of greater than 60pg/mL has a 98.9% sensitivity for the diagnosis of HLH in pediatric patients with or without malignancy [44]. Additional laboratory derangements observed in CRS are non-specific elevation of lactate dehydrogenase, creatinine, and blood urea nitrogen (BUN) likely from end organ dysfunction from shock and tissue hypoxemia and are not helpful in the prediction of severity of CRS [19].

The CAR T construct may determine the nature of CRS, specifically the time to presentation and severity. CRS was rarely reported with the use of first-generation CAR constructs. The addition of a costimulatory signaling domain in second generation constructs resulted in increased incidence of CRS. CAR constructs with the CD28 costimulatory domain demonstrate rapid but time-limited T cell expansion and appear to be associated with higher risk of CRS whereas CARs with the 4–1BB domain demonstrate longer persistence and later presentation of CRS/CRES [20,45,47]. In patients receiving axicabtagene ciloleucel (anti-CD19 with CD28 domain), the median time to onset of CRS was 2 days post infusion, with median time to resolution of 8 days. Median onset of neurologic symptoms was day 5 post treatment [20]. In the phase 2 multi-center study of patients receiving tisagenlecleucel, median onset of CRS was 3 days post infusion with a median duration of symptoms of 8 days. 35 of 75 patients required admission to the ICU for management of CRS with a median duration of stay of 7 days [5]. A review of 39 patients receiving tisagenlecleucel in a phase 1/2a trial at CHOP who developed severe CRS demonstrated a pattern of cardiogenic dysfunction which occurred within days of fever onset with subsequent renal dysfunction, coagulopathy, and respiratory failure, followed by delayed presentation of hepatic dysfunction and neurotoxicity. The median time from infusion to ICU admission was 5.6 days and median length of stay was 7.8 days. Patients had a median of 15 days of organ dysfunction with the onset of multi-organ dysfunction within 5 days after CAR T infusion. Patients with peak ferritin values greater than 11,200 pmol/L by day 5 post infusion demonstrated longer time to resolution of organ dysfunction. The most common organ dysfunction was hepatic and renal with hepatic dysfunction presenting later than cardiovascular dysfunction. 35% of patients developed cardiovascular dysfunction with a median development of 5 days after infusion. In addition, 33% of patients developed encephalopathy and returned to neurologic baseline by day 18 post-infusion [4,5].

3.2. Neurotoxicity (CRES, ICANS)

Significant insights have been made into the biology and management of CRS, yet major limitations exist in the understanding of the pathophysiology and management of CRES. CRES has been observed with multiple CD19 and CD22 constructs, BiTEs, and in preclinical models using anti-GD2 CAR T cells for neuroblastoma [21,22]. CRES can present initially with more subtle neurologic signs and symptoms, including tremor, mild aphasia, apraxia, dysgraphia, and lethargy. Headache is often a non-specific early symptom. Santomasso and colleagues identified expressive aphasia as a specific early marker of severe neuro-toxicity as it was the first symptom in 19 of 22 patients whom developed severe neurotoxicity (NT) [23]. Symptoms may progress to severe aphasia, delirium including hallucinations, encephalopathy, and rarely seizures, coma, and fatal cerebral edema over the course of hours to days. CRES may occur in concert with other symptoms of CRS or may arise after these symptoms have resolved. The long-term effects of CRES are unknown.

CRES is now considered a distinct syndrome from CRS due to the timing of development in relation to CRS, lack of response to CRS management, and differences in biology. Lee and colleagues proposed a grading tool for encephalopathy symptoms named the Immune Effector Cell-Associated Encephalopathy (ICE) score for objective evaluation of encephalopathy, with a score of 1–10 dividing patients into grade 1–4 encephalopathy [24]. This is further integrated into the ASBMT ICANS consensus grading for adults. A consensus grading of ICANS has also been proposed for children incorporating the Cornell Assessment of Pediatric Delirium based on limitations of use of ICE assessment for children less than 12 years of age due to cognitive abilities.

The mechanism of CRES is poorly understood and NT is hypothesized to occur due to direct effect from CAR T cells or cytokine release in the setting of CRS. There is currently no consensus on the effective management of CRES other than administration of corticosteroids in cases with cerebral edema, which may threaten the persistence of CAR T cells [7,13]. In 133 adult patients with ALL, CLL or NHL treated with a CD19–41BB CAR T construct, 40% had grade 1 or higher neurologic adverse events [25]. The most common side effects were delirium with preserved alertness (35/53 patients) and headache (29/53 patients). 7 patients developed grade 4 NT (CTCAE grading scale) with 4 deaths. Factors associated with the development of severe NT included early or severe CRS, early and higher peak CD4 and CD8 CAR T cell counts, capillary leak syndrome, and coagulopathy. In addition, an early peak of IL-6 was associated with higher risk of severe NT. In a pediatric cohort, 45% of 51 children treated with CD19–41BB CAR T developed symptoms of NT. There were 4 patients with seizure and no deaths [5].

In 53 adults treated with CD19–28z CAR T, 62% of patients had grade 1 or higher neurologic adverse events. Expressive aphasia was the most common symptom and also the earliest manifestation of severe CRES. Seizures were observed in 16 of 22 patients with severe CRES. There were no deaths [23]. In the adult patients receiving CD19–28z who developed severe NT, higher serum concentrations of IL-1, IL-2, IL-3, IL-5, IL-6, IL-10, IL-15, interferon gamma-induced protein 10 (IP10), IFNγ, MCP1, and GM-CSF were observed on day 3 post infusion. Within the CSF IL-1α, IL-6, IL-10, G-CSF, TNFα, IFNγ, IFNα, were elevated, with IL-8, IP10, and MCP1 observed to be markedly elevated in the CSF of patients with severe NT relative to the serum [23]. Neither CSF cell counts nor quantity of CAR T cells within the CSF correlated with NT grade. CSF protein levels and CSF/serum albumin quotient did correlate with grade of NT, suggesting increased blood-CSF barrier permeability and dysfunction. Neuro-excitatory NMDA receptor agonists glutamate and quinolinic acid (QA) were also found to be elevated in the CSF during symptoms of NT compared to low levels in a patient without NT. Increased QA may potentiate inflammation via the induction of expression of TNFα, IL-6, MCP1 by astrocytes. QA is known to alter the integrity of the blood brain barrier, potentially allowing further increase of serum inflammatory cytokines in the CSF. Subsequently identified biomarkers predictive of the development of severe NT were low baseline platelet count (<50k/uL), high serum IL-15 and IL-10, and low epidermal growth factor (EGF) on day 3 post infusion [23].

These findings support previous observations that elevated levels of serum IL-15 are associated with the development of severe NT, including cerebral edema [46]. In adult patients with ALL treated with CD19–28z, 22% developed CRS and severe neurotoxicity was observed in 56% of patients in the trial, including fatal cerebral edema. Patients with fatal neurotoxicity demonstrated an early peak CAR T cell expansion and a rapid rise in IL-15 [26]. NT may be more severe in patients with a rapid rise of cytokine levels, which supports the observation of earlier development of NT in patients receiving CD19–28z constructs as compared to CD19–41BB [23].

CRES is postulated to arise in the setting of significant inflammation which disrupts the blood-brain barrier (BBB) with early central nervous system (CNS) endothelial cell (EC) activation [25]. CNS EC provide structure and functional support of the BBB and this activation may result in increased permeability allowed penetration of high levels of inflammatory cytokines into the CNS. Gust and colleagues [25] previously reported the association between severe NT and coagulopathy and significant vascular leak and hypothesized that EC activation may be an important component of CRES. In patients received CD19 CAR constructs, they observed that patients with severe NT had higher levels of an EC regulator, angiopoietin-2 (ANG2). Inflammation drives the release of ANG2 and von Willebrand factor (VWF) from CNS EC Weibel-Palade bodies. ANG2 displaces angiotensin-1 (ANG1) and disrupts ANG1 signaling via receptor tyrosine kinase TIE2. This leads to activation of the CNS EC and enhanced BBB permeability with the subsequent elevation in inflammatory cytokines further activating CNS EC resulting in a positive feedback loop [25,27]. As described by Mackall and colleagues, this pathophysiology mimics that of thrombocytopenic thrombotic purpura, which has historically been treated with plasmapheresis. This shared pathophysiology may suggest a potential role for plasmapheresis for CRES however this has yet to be evaluated.

4. Management of Chimeric Antigen T Cell Receptor Toxicities

The severity and timing of CRS varies widely with CAR T cell construct and targeted disease. CRS, when severe, requires prompt identification and intensive medical management. Prompt response can ameliorate most adverse outcomes; however, the selected form of anti-CRS therapy may affect the disease outcome itself. For low grade CRS, the management is primarily supportive, with intravenous fluids and anti-pyretic medications. If more severe CRS develops, such as fluid refractory hypotension or respiratory failure, anti-IL6 directed therapy is indicated and preferred as initial therapy over corticosteroids [18]. Corticosteroids are highly effective in the treatment of CRS and suppress inflammatory response in both CRS and CRES. However, corticosteroids inhibit T cell function and may impair the persistence and anti-tumor activity of CAR T cells [6]. High dose corticosteroids should be reserved for those with life-threatening CRS/CRES that is unresponsive to IL-6 targeted therapy (Figure 1).

Figure 1.

Management of Chimeric Antigen Receptor T-Cell Cytokine Release Syndrome; Cytokine release syndrome CRS, Kilogram kg, Milligram mg

Anti-cytokine therapy targeting IL-6 is an effective therapeutic approach to ameliorate symptoms of CRS. IL-6 acts via three modes of signaling: classic signaling, trans-signaling and trans-presentation [28]. The IL-6 receptor itself exists in two forms, membrane bound and soluble. Membrane bound IL-6R is found on leukocytes, hepatocytes, and epithelial cells. In classic intracellular signaling, IL-6 binds to membrane bound IL-6R and interacts with the membrane protein glycoprotein 130 (gp130) which induces dimerization and results in intracellular signaling. Cells that do not express IL-6R can be stimulated by IL-6 via trans-signaling induced by the soluble IL-6R/IL-6 complex. This complex may bind to gp130 which is expressed on all cells and initiate intracellular signaling. Trans-presentation occurs when dendritic cells present IL-6 via their membrane-bound IL-6R to T cells which subsequently leads to intracellular signaling by gp130 [28,48].

Tocilizumab is a humanized monoclonal antibody against both the soluble and membrane bound IL-6 receptor and competitively inhibits the binding of IL-6 to the receptor. It has been extensively and safely used in adults and children for the treatment of rheumatologic disease. Intravenous dosing is based on body weight, whereas a subcutaneous version is available and is fixed dosing. Tocilizumab has a non-linear pharmacokinetic profile and undergoes biphasic elimination from the circulation. A population pharmacokinetic study in patients with SJIA demonstrated association of increased drug clearance with increased body surface area. In adult patients receiving 3 serial infusions of 8mg/kg, half-life was approximately 240 hours [29]. In comparison with the population PK model, in patients with CRS the observed mean maximum concentration of tocilizumab after the first dose suggested faster clearance of tocilizumab suggesting increased clearance of drug during the state of extreme inflammation.

The use of tocilizumab to treat CRS was pioneered by physicians at CHOP and UPENN and was approved by the FDA in 2017 for the treatment of CAR T associated CRS in patients two years of age and older [29]. Unlike corticosteroids, tocilizumab does not appear to affect expansion of CAR T cells or long-term efficacy [5,6,7]. Tocilizumab has no significant severe adverse side effects based on data obtained from a single study center with healthy controls and prior use in patients with rheumatologic disease. In the healthy patient population, tocilizumab was only shown to result in mild reduction in neutrophil count [29]. The main adverse effect of IL-6 blockade based on use in patients with rheumatologic disease appears to be increased susceptibility to bacterial infections. However, tocilizumab did not appear to diminish humoral responses to pneumococcus vaccination with short-term administration. In addition, in a review of 60 patients with CAR T associated CRS there were no reported adverse events related to administration of tocilizumab [29].

Importantly, tocilizumab demonstrates impressive efficacy in the management of CAR T related severe CRS. In the phase 1/2a study of patients receiving tisagenlecleucel, tocilizumab was administered to 13 subjects with cardiovascular dysfunction, with median dose given 5 days after T cell infusion. Fever and tachycardia rapidly resolved within a median of 4 hours after treatment however catecholamine-dependent shock demonstrated a delayed responsive with resolution over a median of 4 days after tocilizumab treatment. Respiratory failure was observed in 8 patients with 6 requiring invasive ventilation. Neurotoxicity developed in 33% of patients and lasted a median of 7 days with all patients returning to neurologic baseline by day 18 post infusion. 12 of the 16 patients requiring ICU admission met the criteria for HLH/MAS with fever, cytopenias, hyperferritinemia, hypofibrinogenemia with a median peak ferritin of 135,300pmol/L [4]. Given these parallels between CRS and HLH, the use of tocilizumab in the treatment of HLH is currently being investigated. While there have been no randomized control trials examining the use of tocilizumab for CRS, the growing literature of rapid improvement of vital signs and clinical status provides evidence for its continued use in the setting of life-threatening CRS [5,6,7]. In a comparison of response to tocilizumab therapy between tisagenlecleucel and KTE-019 trials, both groups demonstrated favorable response to tocilizumab with no adverse outcomes attributed to drug [29]. In patients receiving tisagenlecleucel with severe CRS, 31 of 45 patients who received tocilizumab achieved resolution of symptoms, defined as resolution of fever and without vasopressor support, within 14 days of first dose (Table 2). In the cohort of patients receiving KTE-019, 8 of 15 patients demonstrated resolution with a single dose [29]. The data is compelling enough that tocilizumab was FDA approved for the use of CRS and requires that centers treating patients with tisagenlecleucel have tocilizumab immediately available for management of CRS.

In addition to CAR T induced CRS, tocilizumab was effective in the treatment of severe CRS in patients receiving the PD-1 inhibitor, nivolumab, in conjunction with corticosteroids [30,31]. Tocilizumab has also been used to ameliorate CRS after treatment with blinatumomab [3]. Finally, tocilizumab has been used to treat CRS after haploidentical stem cell transplant. In these circumstances’ patients developed clinical symptoms similar to CAR T cell induced CRS. Cytokine profiling with T-cell replete haploidentical hematopoietic cell transplantation has demonstrated hypercytokinemia with significant elevation of IL-6, in addition to increase in IL-2, IL-4, IL-10, IL-15 and TNFα [32,33].

Additional agents inhibiting IL-6 and IL-6 mediated signaling may represent therapeutic interventions for the management of CRS. One of these anti-cytokine therapies is siltuximab, an antibody directed against one of the three distinct binding sites of IL-6. IL-6 has 3 distinct binding sites: site I, site II, and site III. Siltuximab is an antibody directed against site I and blocks IL-6R binding [28]. To evaluate the effectiveness of siltuximab in the treatment of CRS, prospective clinical trials are needed, as there is a relative paucity of data regarding its use in CRS. Olamkicept is a soluble form of glycoprotein 130 (gp130Fc) that targets the soluble IL-6/IL-6R complex and blocks trans-signaling. Phase 1 clinical trials conducted in 2014 did not demonstrate major adverse events and it is currently in phase II clinical trials (TJ301) for treatment of inflammatory bowel disease [28,49]. However, targeting gp130 with neutralizing antibodies does not appear to be an option as evidence suggests blockade of gp130 is too toxic [28].

Another potential target of interest in IL-6 signaling involves blocking the Janus kinases/signal transducer and activator of transcription proteins (Jak/Stat) signaling pathway which is activated by multiple cytokines including IL-6 and INFγ. A number of Jak/Stat inhibitors are FDA approved and others are in different stages of clinical development. Examples include tofacitinib, an inhibitor of JAK 3 and JAK1, is FDA approved for the treatment of rheumatoid arthritis and Crohn’s disease, and ruxolitinib, an inhibitor of JAK1 and JAK2, that is approved for the treatment of myelofibrosis and polycythemia vera [28]. Preclinical data demonstrates ruxolitinib may be effective to treat CRS as was shown in AML xenograft mouse model treated with anti-CD123 CAR T construct [34]. Importantly, ruxolitinib did not appear to impact the efficacy of CAR T cells compared to controls. Historically, ruxolitinib was thought to have limited blood-brain barrier penetration however preclinical data investigating the treatment of HIV-associated neurocognitive disorders demonstrated blood brain barrier penetration with reduction of HIV-induced inflammation and gliosis [35]. Thus, ruxolitinib may represent a management option for CRES given preclinical efficacy for CRS and potential enhanced blood brain barrier penetration during inflammatory states.

While targeting IL-6 with tocilizumab has been effective in the management of CRS, it does not attenuate symptoms of neurotoxicity and is currently not FDA approved for the treatment of CRES. This may be in part due the inability of tocilizumab to cross the blood-brain barrier. As discussed, signs of neurotoxicity often develop during CRS and subside with resolution however a delayed presentation of neurotoxicity has been observed. In adult patients receiving CD19–28z CAR T cell infusion, severe neurotoxicity occurred in 22 of 53 patients either concurrent to or after the resolution of CRS with the median time to presentation with severe NT of 9 days. CRS was responsive to administration of corticosteroids and/or tocilizumab, but neurotoxicity was less responsive to these interventions [23]. In the multi-center ZUMA-1 trial with an anti-CD19 CAR construct with CD28 co-stimulatory domain, patients experienced increased rates of severe NT with early tocilizumab, which Neelapu and colleagues attributed to a transient rise in IL-6 levels following tocilizumab and theoretical passive diffusion of IL-6 into the CNS potentiating symptoms of NT [36].

Recently, a xenotolerant mouse model was created that accurately replicated CRS and severe neurotoxicity and demonstrated that monocytes are primarily responsible for CRS and are the major source of IL-6 and IL-1 [37, 50]. The IL-1 family of cytokines, including IL-1, IL-18, IL-33, and IL-36 have an important role in immune response and generation of inflammation and also have the ability to activate mast cells. Thus, IL-1 represents a desirable target in the pathophysiology of CRS [50]. Mice were treated prophylactically with anakinra (an IL-1 receptor antagonist), tocilizumab, or placebo at the time of CAR T infusion. Neither tocilizumab nor anakinra was shown to affect T cell expansion in mice. Surprisingly, mice treated with vehicle or tocilizumab demonstrated a delayed lethal neurotoxicity. Those treated with anakinra did not demonstrate delayed neurologic disease. This was effectively prevented by anakinra and postmortem analysis demonstrated meningeal thickening of mice receiving tocilizumab but not anakinra. Thus, anakinra prophylaxis resulted in statistically significant improvement in survival of mice and prevent lethal neurotoxicity. Anakinra has also been effective in the treatment of macrophage activation syndrome, which shares significant pathophysiological overlap with CRS. While anakinra has not been used in the management of CRS, it may represent a promising agent for the treatment of CRS and potential prophylaxis for life-threatening neurotoxicity [37].

Additional preclinical data from Sterner and colleagues investigating GM-CSF blockade with lenzilumab in a patient-derived xenograft model of CRS and NT demonstrated prevention of the development of CRS and reduction in severity of NT symptoms, suggesting another potential option for the management of CRES [38]. Importantly, lenzilumab did not impact CAR T cell function in vitro or in vivo in xenograft models. Phase 2 trial of lenzilumab with anti-CD19 CAR T construct is currently planned.

5. Expert Opinion:

Anti-tumor immunotherapeutic approaches offer promising treatment effects on historically difficult to treat malignancies. However, with the broadening use of these novel therapies comes the need for clinicians to recognize and manage associated toxicities. A number of recent seminal papers have made key insights but more work is needed to further elucidate the pathophysiology of CRS and CRES which will undoubtedly improve management. Ongoing efforts to improve the management of CRS include pre-emptive and early use of cytokine blockade with tocilizumab to mitigate toxicity and the preclinical testing of agents targeting alternate cytokines and pathways. Future steps in the management of CRS should include prospective trials that explore alternative therapies to tocilizumab that block IL6 signaling and agents that target different pathways, including inhibitors of IL-1 and Jak/Stat signaling. In addition, there is variation between institutions regarding the grading and management of CRS, thus controlling for clinical status between institutions while exploring the efficacy of newer agents may present a significant challenge.

While there has been recent advancement in the understanding of CRES, more work is needed to further define the pathophysiology and identify effective therapies other than corticosteroids that threaten T cell persistence. The matrix of variables that constitute a cellular therapy such as CAR T cells is large, and each change may impact the severity and nature of toxicity. While the HLH/MAS-like picture may be a final common pathway, the triggers may vary with costimulatory domain, lymphodepletion strategy, ex vivo culture method and even disease type. As the field moves toward solid tumors, new frontiers in toxicity may appear, such as the on-target off-tumor effects inherent in many solid tumor CAR strategies. This is even true in hematologic malignancies as we target acute myeloid leukemia, where targets such as CD33 or CD123 may exist on normal stem cells. There is a high need for new therapies in T cell malignancies, but designing a CAR T cell that kills only malignant T cells and not each other may unveil new toxicities not previously seen with CD19 CARs. As field is in rapid evolution, researchers must avoid the challenges of over-generalizing conclusions on toxicity in immune therapies to stay vigilant for new and unexpected pathways and new treatments. As discussed, targeting IL-6 is important, but perhaps targeting the pathways that trigger IL-6 release will be more effective or will unveil a new toxicity previously masked.

It is of utmost importance that research continue to advance the understanding of the pathophysiology and management of these toxicities as undoubtedly the range of novel toxicities will increase with different CAR-T cell products and as CAR-T cells are expanded in the clinical arena as therapy for non-B cell malignancies. Collaborative multi-institutional studies integrating data collected from different trials and CAR-T cell products will be imperative to generate a complete understanding of the toxicities inherent to modern immunotherapies. Moreover, collaborative studies are essential to allow for consensus on definition of CRS and CRES and improved management of these toxicities allowing for improved patient outcomes.

Article Highlights:

• Tocilizumab is a humanized monoclonal antibody against both the soluble and membrane-bound interleukin 6 receptor and was approved by the FDA in 2017 for the treatment of cytokine release syndrome.

• Tocilizumab demonstrates impressive efficacy in the management of chimeric antigen receptor T cell related cytokine release syndrome.

• Tocilizumab does not attenuate symptoms of neurotoxicity and is currently not FDA approved for the treatment of CAR T cell related encephalopathy syndrome.

• High dose corticosteroids should be reserved for those with life-threatening cytokine release syndrome and CAR T cell related encephalopathy syndrome that is unresponsive to interleukin 6 targeted therapy.

• Future steps in the management of cytokine release syndrome are prospective trials comparing anti-interleukin-6 targeted therapies to tocilizumab in the management of both cytokine release syndrome and CAR T cell related encephalopathy syndrome.