ASBMT Consensus Grading for Cytokine Release Syndrome and Neurological Toxicity Associated with Immune Effector Cells

Abstract

Chimeric antigen receptor (CAR) T cell therapy is rapidly emerging as one of the most promising therapies for hematological malignancies. Two CAR T products were recently approved in the United States and Europe for the treatment of patients with relapsed or refractory B-cell acute lymphoblastic leukemia up to the age of 25 years and/or adults with large B-cell lymphoma. Many more CAR T products as well as other immunotherapies including various immune cell- and bi-specific antibody-based approaches that function by activation of immune effector cells are in clinical development for both hematologic and solid tumor malignancies. These therapies are associated with unique toxicities of cytokine release syndrome (CRS) and neurological toxicity. The assessment and grading of these toxicities have varied considerably across clinical trials and across institutions, making it difficult to compare safety of different products and hindering the ability to develop optimal strategies for management of these toxicities. Moreover, some aspects of these grading systems can be challenging to implement widely across centers. Therefore, in an effort to harmonize the definitions and grading systems for CRS and neurotoxicity, experts from all aspects of the field met on June 20–21, 2018, at a meeting supported by the American Society for Blood and Marrow Transplantation (ASBMT) in Arlington, VA. Here, we report the consensus of the group and propose new definitions and grading for CRS and neurotoxicity that are objective, easy to use, and ultimately more accurately categorize the severity of these toxicities. The goal is to provide a uniform consensus grading system for CRS and neurotoxicity associated with immune effector cell therapies, for use across clinical trials and in the post-approval clinical setting.

Introduction

Chimeric antigen receptor (CAR) T cell therapies are revolutionizing the management of B-cell leukemias and lymphomas and are quickly being extended to numerous other malignancies. Two CD19 CAR T cell products were recently approved in the United States and Europe (1–4) and more indications are expected in the coming years. Tisagenlecleucel was approved for multiply relapsed or refractory B-cell acute lymphoblastic leukemia (ALL) in patients up to age 25 years as well as for adults with relapsed or refractory large B-cell lymphoma, and axicabtagene ciloleucel was approved for adults with relapsed or refractory large B-cell lymphoma. These therapies are now being tested in earlier lines of treatment, signaling the growing scope of CAR therapy in the management of these diseases. CAR therapies targeting CD22 in B-cell malignancies (5) and B-cell maturation antigen (BCMA) in multiple myeloma (6), highly successful in early trials, are also forthcoming along with combination approaches targeting multiple antigens simultaneously.

Early clinical trials of CD19 CAR T cells quickly uncovered toxicities greater than those seen in other cellular therapies, indicating profound and generalized immune system activation. Some of these toxicities, especially cytokine release syndrome (CRS), were reminiscent of those seen in a study where 6 of 6 healthy, young male volunteers who received a low dose of TGN1412, a superagonist monoclonal antibody to CD28, required critical care for the rapid onset of multiorgan failure (7). Symptoms induced by TGN1412 consisted of fever, rigors, hypotension requiring vasopressor support and other aggressive management, tachycardia, hypoxia, respiratory failure, capillary leak, acute kidney injury, coagulopathy, and even neurological manifestations including poor concentration and delirium. Retrospective analysis revealed marked elevations in C-reactive protein (CRP), interferon γ (IFNγ), tumor necrosis factor α (TNFα), IL-6, IL-10, IL-2, and IL-1β, among other cytokines. Though all volunteers eventually recovered with the use of high-dose methylprednisolone, the IL-2 receptor antagonist, daclizumab, and aggressive supportive care, the rapidity and severity of the toxicities precluded further development of TGN1412.

Similar toxicities were observed in the first patients treated with CD19 CAR T cells. In the first pediatric ALL patient treated (8, 9), it became clear that supraphysiologic cytokine elevation was responsible for the vast majority of symptoms, suggesting that these toxicities were the result of CRS. Investigators struggled with the then-accepted definition of and grading scheme for CRS. In the Common Terminology Criteria for Adverse Events version 3 (CTCAE v3 (10), in effect when many of these studies began, CRS onset was defined as being within 24 hours of initiation of therapy, which is atypical for CRS associated with CAR T cells as well as other immune effector cells therapies. In CTCAE v4.03 (11) (Table 1) the definition did not include fever as a prerequisite for CRS, and the grading was dependent in part on whether the drug infusion was interrupted or not, a feature not applicable to CAR T cells, which are generally infused in a single dose over a concise time (2–30 minutes). Indeed, CTCAE v4.03 was more applicable to toxicity seen with antibody infusions rather than cell infusions. Without a clear and accurate consensus available, CRS grading varied widely between institutions and evolved over time, making toxicity comparisons between products and trials exceedingly difficult. For these reasons, and since immune effector cell-associated CRS can be fatal if not recognized and treated promptly, a CRS grading system that more accurately captures the potentially severe syndrome observed after immune effector cell therapies is needed. Additionally, this CRS grading system should have broad applicability across multiple institutions and/or CAR products and other cellular immunotherapies with minimal effort for implementation.

Table 1:

Published CRS grading systems.

Grading System (Ref)	Grade 1	Grade 2	Grade 3	Grade 4
CTCAE Version 4.03 (11)	Mild reaction; infusion interruption not indicated; intervention not indicated	Therapy or infusion interruption indicated but responds promptly to symptomatic treatment (antihistamines, NSAIDS, narcotics, IV fluids); prophylactic medications indicated for ≤ 24 hours	Prolonged (eg, not rapidly responsive to symptomatic medication and/or brief interruption of infusion); recurrence of symptoms following initial improvement; hospitalization indicated for clinical sequelae (such as renal impairment, pulmonary infiltrate)	Life-threatening consequences; pressor or ventilatory support indicated
CTCAE Version 5.0 (13)	Fever, with or without constitutional symptoms	Hypotension responding to fluids. Hypoxia responding to <40% FiO2	Hypotension managed with one pressor. Hypoxia requiring ≥40% FiO2	Life-threatening consequences; urgent intervention needed
Lee Criteria (14)	Symptoms are not life threatening and require symptomatic treatment only (fever, nausea, fatigue, headache, myalgias, malaise)	Symptoms require and respond to moderate intervention: • Oxygen requirement <40% FiO2 OR • Hypotension responsive to IV fluids or low dose of one vasopressor OR • Grade 2 organ toxicity*	Symptoms require and respond to aggressive intervention: • Oxygen requirement ≥40% FiO2 OR • Hypotension requiring high dose or multiple vasopressors OR • Grade 3 organ toxicity* or grade 4 transaminitis	Life-threatening symptoms: • Requirement for ventilator support OR • Grade 4 organ toxicity* (excluding transaminitis)
Penn Criteria (17)	Mild reaction: Treated with supportive care such as antipyretics, antiemetics	Moderate reaction: Some signs of organ dysfunction (grade 2 creatinine or grade 3 LFTs) related to CRS and not attributable to any other condition. Hospitalization for management of CRS-related symptoms, including neutropenic fever and need for IV 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 and not attributable to any other condition. Hypotension treated with multiple fluid boluses or low-dose vasopressors. Coagulopathy requiring fresh frozen plasma, cryoprecipitate, or fibrinogen concentrate. Hypoxia requiring supplemental oxygen (nasal cannula oxygen, high-flow oxygen, CPAP, or BiPAP).	Life-threatening complications such as hypotension requiring high-dose vasopressors. Hypoxia requiring mechanical ventilation.
MSKCC Criteria (16)	Mild symptoms requiring observation or supportive care only (e.g. antipyretics, antiemetics, pain meds, etc.)	Hypotension requiring any vasopressors <24 hours Hypoxia or dyspnea requiring supplemental oxygen <40%	Hypotension requiring any vasopressors ≥24 hours Hypoxia or dyspnea requiring supplemental oxygen ≥40%	Life-threatening symptoms Hypotension refractory to high dose vasopressors Hypoxia or dyspnea requiring mechanical ventilation
CARTOX Criteria (12)	Temperature ≥ 38°C Grade 1 organ toxicity**	Hypotension responds to IV fluids or low-dose vasopressor Hypoxia requiring FiO2 < 40% Grade 2 organ toxicity**	Hypotension needing high-dose or multiple vasopressors Hypoxia requiring FiO2 ≥40% Grade 3 organ toxicity** or grade 4 transaminitis	Life-threatening hypotension Needing ventilator support Grade 4 organ toxicity** except grade 4 transaminitis

Ref: Reference; LFTs: Liver function tests; CPAP: Continuous positive airway pressure; BiPAP: Bilevel positive airway pressure

^*As per CTCAE Version 4.03

^**Cardiac (tachycardia, arrhythmias, heart block, low ejection fraction), Respiratory (tachypnea, pleural effusion, pulmonary edema), GI (nausea, vomiting, diarrhea), Hepatic (increased serum ALT, AST, or bilirubin levels), Renal (acute kidney injury, increased serum creatinine, decreased urine output), Dermatological (rash), Coagulopathy (disseminated intravascular coagulation)

Besides CRS, another common toxicity observed after CAR T cell therapy is neurotoxicity (12). Immune effector cell-associated neurotoxicity syndrome (ICANS) may manifest as delirium, encephalopathy, aphasia, lethargy, difficulty concentrating, agitation, tremor, seizures and, rarely, cerebral edema. In addition, headache is very commonly seen and may not represent neurotoxicity per se. Previously considered in aggregate with CRS, neurotoxicity is now treated as a separate entity due to its distinct timing and response to intervention. Neurologic symptoms may occur during or more commonly following CRS symptoms (but rarely before CRS), are variable between patients, and have an unclear pathophysiology, distinct from CRS. One challenge has been the identification of the symptoms most relevant to neurotoxicity. Investigators used multiple different terms for similar symptomatology resulting in considerable variation in grading across trials and also across various institutions within the same trial. For example, a patient experiencing encephalopathy after CAR T cell therapy may be reported as having any or a combination of the following vague, overlapping CTCAE adverse event terms: confusion, delirium, encephalopathy, cognitive disturbance, concentration impairment, somnolence, and/or depressed level of consciousness (13). Ascertainment changed over time and even within trials as the issue of neurotoxicity became more apparent. Moreover, CTCAE grading of neurological toxicities based on ability to perform instrumental and self-care activities of daily living are not always applicable to children or for hospitalized adult CAR T patients who may be bedridden for related co-morbidities. The multiple adverse event terms used to grade neurological toxicities are also not practical for application at the bedside for rapid and dynamic assessment of patients, so discerning which term(s) are appropriate is difficult and often highly subjective. Therefore, there is a need for an objective, reproducible, and easy to use practical tool that can be utilized by all health care providers and possibly caregivers for recognition and assessment of immune effector cell-associated neurological toxicities in the inpatient or outpatient setting.

Initial Attempts to Better Define and Grade CRS

Though early clinical trials modified the CTCAE v4.03 grading of CRS, further refinement was achieved when a multi-institutional group of pediatric oncologists leading CAR T cell trials across the United States published what is now commonly referred to as the Lee criteria (14). This work re-defined the clinical signs and symptoms associated with CRS (Table 1). Of note, neurologic toxicities such as confusion, delirium, aphasia, etc. were included but are now generally accepted to be a separate syndrome (though cytokines have been implicated in the pathophysiology of this syndrome) owing to the differential time of presentation compared to the other signs of CRS and lack of knowledge surrounding its etiology and pathophysiology. The new constellation of symptoms incorporated the experience across CAR T studies in hematologic malignancies and included, for the first time, fever as a hallmark of CRS.

Lee and colleagues then redefined the grading criteria for CRS revolving around hypoxia requiring oxygen supplementation, hypotension, and other end-organ toxicities (Table 1) (14). In contradistinction to conventional CTCAE grading schemes, hypotension responsive to low-dose vasopressors was considered a Grade 2 CRS. Early experience demonstrated that reliance on intravenous fluids (IVF) alone to manage persistent hypotension was inferior to early vasopressor use owing to significant capillary leak and subsequent pulmonary edema and effusions after IVF management leading to a cascade of events that can quickly result in life-threatening toxicity. Further, patients who are easily managed with minimal vasopressors are decidedly distinct in terms of CRS severity from those who require high dose or multiple vasopressors, a key difference accounted for by the grading criteria.

Likewise, the grading schema distinguished between those patients who require minimal oxygen supplementation and those requiring aggressive supplementation or continuous positive airway pressure support (CPAP). An FiO2 of 40% was arbitrarily chosen as the dividing line between Grades 2 and 3 CRS. However, this aspect of the definition was problematic, for the reason that delivery of oxygen to patients (14) will vary significantly from hospital to hospital, from patient to patient, and from shift to shift. Similar to patients requiring low-dose versus aggressive vasopressor support, patients requiring minimal oxygen supplementation are distinct in terms of severity from those who require more aggressive intervention, ranging from CPAP to intubation.

Other Grading Schemes for CRS

The Lee criteria have been widely adopted by many CAR T cell groups, especially as it was the first to link specific grading to a suggested treatment algorithm. The group at Memorial Sloan Kettering Cancer Center (MSKCC) identified objective factors that distinguished severe versus non-severe CRS in their early clinical trials; however, this relies on the availability of serum cytokine levels in patients in real-time (15). Recognizing that assays for serum cytokines are not readily available at most centers, thereby limiting the utility of this approach, MSKCC redefined CRS grading used in their clinical trials (Table 1) (16). Hypotension requiring less than 24 hours of vasopressor use was deemed Grade 2 while 24 hours or more defines Grade 3 and hypotension not corrected with high-dose vasopressor within 3 hours as Grade 4. Hypoxia also contributes to CRS grading with required FiO2 of 40% serving as the demarcating line between Grades 2 and 3. Intubation triggers Grade 4, but there is no mention of other methods of delivering positive pressure such as CPAP.

The University of Pennsylvania published a grading scale that has been used in their CD19 CAR T cell trials (Penn criteria; Table 1) (17). In contrast to the Lee criteria, the Penn criteria assign the same grade 3 CRS to patients requiring any amount of IVF for hypotension as to those requiring low dose vasopressors, and to patients requiring minimal oxygen supplementation and those requiring more aggressive support including CPAP. Owing to these differences and the inclusion of neutropenic fever as a trigger for Grade 2 CRS, the Penn criteria tend to assign a higher grade of CRS compared to the Lee criteria making comparison of clinical trial safety data across centers difficult.

Most recently, a multi-institutional group of investigators on several industry-sponsored CAR T cell trials published a manuscript on CAR-toxicity (CARTOX) grading and management of CRS and CAR-associated neurotoxicity (12). The CARTOX CRS grading differs slightly from Lee criteria by including grade 1 organ toxicity to be considered under grade 1 CRS and defined fever, hypotension, and hypoxia for grading of CRS in adults (Table 1). In addition, a separate system was proposed for grading of neurotoxicity. Differences between the Penn, MSKCC, CARTOX, and Lee approaches in the management of CRS are outside the scope of the present discussion.

Efforts to Harmonize Immune Effector Cell-Associated CRS and Neurotoxicity Definitions and Grading

Recognizing the disparity in published grading schemas and the need for harmonization of definitions and grading systems for immune effector cell-associated CRS and neurotoxicities seen after immune effector cell therapies including CAR T therapy, 49 experts from all aspects of the field met in Arlington, VA on June 20–21, 2018, at a meeting supported by the American Society for Blood and Marrow Transplantation (ASBMT). Attendees included leaders from major academic centers involved in CAR T cell therapy research as well as representatives from industry, the Center for International Blood and Marrow Transplant Research (CIBMTR), the American Society of Hematology (ASH), and the National Cancer Institute (NCI). Key presentations regarding immune effector cell-associated CRS and neurotoxicity were followed by focused discussion of salient points. The writing group was then tasked with generating language encompassing a new consensus that is both easily applied at the bedside and easily verifiable during chart reviews. A full, iterative drafting and vetting process was undertaken. In addition, these guidelines were presented at the CIBMTR CT Registry Forum on October 25, 2018 for discussion and comment. The participation at this second meeting included a broad group of multiple stakeholders including investigators, industry, payors, and NIH and other governmental agencies. Here, we report the consensus and rationale of the group related to grading and reporting of toxicities.

Is Cytokine Release Syndrome the Appropriate Term for Immune Effector Cell-Associated Toxicity?

We first discussed whether cytokine release syndrome is the most appropriate term to assign to the constellation of symptoms occurring after CAR T cell and other immune effector cell therapies. The pathophysiology of the syndrome is unclear because no animal models of CRS existed until recently (18, 19). In patients, most CD19 CAR T cell clinical trials to date have found marked inflammatory cytokine elevations in association with the onset of symptomatology and degree of severity. In addition, rapid clinical stabilization is frequently seen with the use of the IL-6 receptor antagonist, tocilizumab, strongly implicating cytokines, especially IL-6 (8, 20–22) in the pathophysiology of the syndrome. In the absence of data suggesting an alternative mechanism, we conclude that the most appropriate term for these immune effector cell-associated symptoms and signs is cytokine release syndrome. We recognize that as CAR T and other immune effector cell therapies are successfully adapted to treat both hematologic malignancies and solid tumors, additional or alternative mechanisms of toxicity may be found.

Definition of Cytokine Release Syndrome

The CTCAE v4.03 defines CRS as “a disorder characterized by nausea, headache, tachycardia, hypotension, rash, and shortness of breath; it is caused by the release of cytokines from the cells (11).” Though inclusive of many of the features of immune effector cell-associated CRS, this definition does not include fever, the hallmark of immune effector cell-associated CRS. The CTCAE v5.0 refined the definition as: “a disorder characterized by fever, tachypnea, headache, tachycardia, hypotension, rash, and/or hypoxia caused by the release of cytokines (13).” Though the list of associated symptoms is more in line with what is seen clinically during immune effector cell-associated CRS, this definition limits the cause to cytokines alone and is not contextually defined. For example, in bacterial sepsis, high levels of many cytokines are produced, and symptoms such as fever and hypotension overlap with CRS, but there is no infection and the overall clinical picture is distinctly different from immune effector cell-associated CRS.

It is also important to note that CRS is observed not just with CAR T and other immune effector cell therapies. In addition to the TGN1412 experience, it has been described in many patients treated with blinatumomab, a bi-specific T cell engaging molecule consisting of two covalently linked single chain antibody fragments targeting CD3 on T cells and CD19 on normal and malignant B cells (23, 24). Preclinical studies suggest that CRS could be observed with CAR NK cell therapy as well (25). Since the same constellation of symptoms has been observed after treatment with multiple agents each working in different ways to activate T and/or other immune effector cells, CRS as we have described it appears to be an immune effector cell-associated phenomenon. Therefore, we define CRS as “a supraphysiologic response following any immune therapy that results in the activation or engagement of endogenous or infused T cells and/or other immune effector cells. Symptoms can be progressive, must include fever at the onset and may include hypotension, capillary leak (hypoxia) and end organ dysfunction.” CRS should be applied to any immune effector cell engaging therapy, not just CAR T cells. Cytokine profiles with other therapies may not be the same, and this may have therapeutic implications. As new, effective immunotherapies centered around cell types other than T cells are developed, the definition may need to be altered.

Symptoms Defining Cytokine Release Syndrome Must be Attributed to Immune Effector Cell Engagement

The common symptoms of CRS are not unique to CRS. Indeed, practitioners must be cautious and exclude other causes of fever, hypotension, hemodynamic instability, and/or respiratory distress such as an overwhelming infection. Bacteremia and other infections have been reported concurrent with and even mistaken for CRS. A reasonable temporal relationship to the cell therapy must be present. Though immune effector cell-associated CRS may have a delayed onset, it rarely presents beyond 14 days after initiation of therapy. Patients exhibiting symptoms consistent with CRS presenting outside this window should be carefully evaluated for other causes.

Toxicities Excluded from the Definition of CRS

As stated before, neurotoxicity is a frequent complication of CAR T cell and other T-cell engaging therapies. Unlike the classic symptoms of CRS, immune effector cell-associated neurotoxicities do not usually respond to tocilizumab, which is not surprising given the observation that intravenous tocilizumab administration does not generate significant levels of the drug in the cerebrospinal fluid (26). Coupled with the paucity of mechanistic data, the lack of known CD19 expression in the CNS, and since often neurotoxicities develop well after the classic symptoms of CRS have resolved, we conclude that immune effector cell-associated neurotoxicities should be excluded from the definition of CRS. Despite this, CRS can impact neurotoxicity and complicate its assessment. High fever and drug therapy can cause delirium. A sedated and/or intubated patient may not be assessable for neurotoxicity. As more insight is gained, this statement may need to be reevaluated. In the meantime, we recommend the use of a separate grading scale for immune effector cell-associated neurotoxicities as described below.

Hemophagocytic lymphohistiocytosis or macrophage activation syndrome (HLH/MAS) overlaps substantially with CRS, illustrated by ferritin elevations seen in many CAR T cell patients during CRS (2, 12, 20, 22, 27). There are many shared features between CRS and classic acquired HLH/MAS, and the two entities are likely not distinct, reflective of the activation of the reticuloendothelial system initiated by T cell-mediated inflammation. Most patients with moderate to severe CRS have laboratory results that satisfy the classic criteria for HLH/MAS but may or may not have hepatosplenomegaly, lymphadenopathy, or overt evidence of hemophagocytosis. In addition, refractory HLH/MAS has only been described in rare cases of immune effector cell-associated CRS (2, 28), while in the vast majority of patients the symptoms (and characteristic elevated cytokines) suggestive of HLH/MAS resolve with CRS resolution (22). Given this overlap, and the absence of a need to directly treat HLH/MAS in most cases, we conclude that HLH/MAS should be excluded from the definition of CRS: patients may fulfill some of the criteria for HLH/MAS after CAR T cell infusion, but this is part of the CRS. Because of the inability to separate CRS from HLH/MAS, and since grading of HLH/MAS is not available as a separate CTCAE term, the group did not see a need to grade this entity separately.

Laboratory Parameters are Not Included in the Definition or Grading of CRS

Significant alterations in many laboratory parameters clearly occur with CRS. Cytokine aberrations have been well described, but such data are not routinely available in most academic centers in a time frame that is useful to assign grade and determine management of a patient experiencing CRS. C-reactive protein (CRP) is a widely available and relatively inexpensive laboratory test and initially appeared to be a useful biomarker of CRS. However, CRP is not specific for CRS, several scales of measurement exist, and our experience suggests that changes in CRP lag behind clinical changes by at least 12 hours. For these reasons, while CRP is often used to follow inflammation, we excluded the use of laboratory parameters from the definition and grading of CRS and favor a system that is based on clinical observation. We do, however, encourage the continued measurement of cytokines, CRP, ferritin, and other parameters so that additional data may be generated for future study.

Implications of a Grading System Based on Practitioner Intervention

Hospitals have varying capacities and policies for providing escalating care to their patients experiencing serious complications. Our proposed schema separates grade of CRS, in large part, on the degree or type of intervention administered to a patient, for example, IV fluid versus vasopressor use for managing hypotension or admission to the intensive care unit (ICU). There are several instances, (i.e. % oxygen supplementation, see below), where strict definitions are confounded by large variability in clinical practice. In the end, many definitions we use reflect the treatment decisions made by the clinical team at the bedside. There was some concern that some practitioners might alter their usual practice and rely longer on IV fluids for hypotension and/or delay transfer to the ICU in an effort to prevent upgrading a patient. Such practice is not the intent of the grading system and strongly discouraged, since prolonged fluid resuscitation without pressor use is associated with worse outcome and since early and aggressive supportive care, early use of vasopressors, and timely anti-cytokine therapy is paramount to mitigating life-threatening CRS.

Consensus on CRS Grading (Table 2)

Table 2.

ASBMT CRS Consensus Grading

CRS Parameter	Grade 1	Grade 2	Grade 3	Grade 4
Fever †	Temperature ≥38°C	Temperature ≥38°C	Temperature ≥38°C	Temperature ≥38°C
With either:
Hypotension	None	Not requiring vasopressors	Requiring one vasopressor with or without vasopressin	Requiring multiple vasopressors (excluding vasopressin)
And/or ‡
Hypoxia	None	Requiring low-flow nasal cannula^ or blow-by	Requiring high-flow nasal cannula^, facemask, non-rebreather mask, or Venturi mask	Requiring positive pressure (eg: CPAP, BiPAP, intubation and mechanical ventilation)

CPAP: Continuous positive airway pressure; BiPAP: Bilevel positive airway pressure

^†Fever is defined as temperature ≥38°C not attributable to any other cause. In patients who have CRS then receive antipyretics or anti-cytokine therapy such as tocilizumab or steroids, fever is no longer required to grade subsequent CRS severity. In this case, CRS grading is driven by hypotension and/or hypoxia.

^‡CRS grade is determined by the more severe event: hypotension or hypoxia not attributable to any other cause. For example, a patient with temperature of 39.5°C, hypotension requiring one vasopressor and hypoxia requiring low-flow nasal cannula is classified as having Grade 3 CRS.

^#Organ toxicities associated with CRS may be graded according to CTCAE v5.0 but they do not influence CRS grading.

^{^}Low-flow nasal cannula is defined as oxygen delivered at ≤ 6 liters/minute. Low flow also includes blow-by oxygen delivery, sometimes used in pediatrics. High-flow nasal cannula is defined as oxygen delivered at > 6 liters/minute..

Grade 1 CRS.

We define Grade 1 CRS as fever (≥38.0 °C). Fever may occur with or without constitutional symptoms. The associated constitutional symptoms may be reported as per CTCAE v5.0 but do not influence the CRS grade. All CRS grading schemes proposed to date mostly agree on what constitutes grade 1 CRS. However, not all systems include fever as a requirement. The constitutional symptoms of CRS such as myalgia, arthralgia and malaise are by themselves non-specific. However, when coincident with fever in the expected timeframe, the etiology of CRS is more likely.

Grade 2 CRS.

We define Grade 2 CRS as fever (≥38.0 °C) with hypotension not requiring vasopressors and/or hypoxia requiring the use of oxygen delivered by low-flow nasal cannula (≤ 6 liters/minute) or blow-by. Lee and colleagues attempted to separate grading for patients who require minimal vasopressor support from those that require intensive vasopressor use (also a feature of Penn grading) and those requiring minimal oxygen supplementation from those requiring more aggressive assistance (14). This was done, in part, out of concern that intervening with anti-cytokine therapy such as tocilizumab, as well as early, prolonged or high dose corticosteroids would abrogate the anti-tumor response. Though prospective clinical trials evaluating the timing of intervention are lacking, retrospective analyses suggest that this is not the case, at least when such therapies are implemented after CRS is well under way (2, 21). How the use of preemptive or prophylactic tocilizumab or corticosteroids affect the anti-tumor response or alter the natural history of other immune effector cell-associated toxicities, such as neurotoxicity, remains an open question and one worth exploring further in well-controlled studies.

The trend across many groups has been to move towards employing anti-cytokine therapy earlier in the development of severe CRS rather than later. For example, many investigators will administer tocilizumab with any vasopressor requirement, even low-dose, or with a significant oxygen requirement, representing a shift in the treatment algorithm initially proposed by Lee and colleagues (12, 14), as well as much of the early Penn experience. In general, we agree with this approach as it initiates CRS management earlier, allowing for earlier resolution, while still preserving efficacy. We also differentiate this practice shift from the prophylactic or pre-emptive use of tocilizumab, which remains experimental. However, the goal in this paper is to define a grading system, and the group clearly recognized the reality of and need for variations in practice in initiating and escalating CRS treatment.

Despite this shift towards earlier intervention, we recognize there is a distinct difference between patients requiring low-dose vasopressor or minimal oxygen supplementation and those who require more aggressive interventions. We sought to capture this difference in our CRS grading scheme as the utilization of resources to support the former are significantly less than those required to support the latter. Our scheme is also aligned with the general concept in the CTCAE that toxicities requiring specific intervention, (e.g. anti-cytokine therapy), meet criteria for Grade 3 at least. However, it is important to recognize that fever may not always be present concurrently with hypotension or hypoxia as it may be masked by interventions such as antipyretics, anti-cytokine therapy, and/or corticosteroids while hypotension and hypoxia may take longer to resolve.

Grade 3 CRS.

We define Grade 3 CRS as fever (≥38.0 °C) with hypotension requiring one vasopressor with or without vasopressin and/or hypoxia requiring high-flow nasal cannula (> 6 liters/minute), facemask, non-rebreather mask, or venturi mask not attributable to any other cause. Several key features of these criteria deserve discussion.

The Lee and Penn criteria relied on established definitions of low- versus high-dose vasopressor use in defining lower vs. higher grade CRS (14, 17). While these definitions are well accepted in the critical care literature, they are cumbersome in practice when assigning or auditing CRS grade. The MSKCC criteria used duration of any vasopressor dose for less than or greater than 24 hours as differentiating between grades 2 and 3 CRS (16). However, this arbitrary time point may not accurately distinguish patients requiring minimal versus significant critical care support. As a result, and owing to real differences in severity between patients requiring one versus two or more vasopressors, we use this distinction (1 vs. 2 or more vasopressors) in our proposed grading system.

Many critical care practitioners administer vasopressin simultaneously with any dose of norepinephrine to capitalize on its vasoconstrictive effects in an effort to mitigate capillary leak and minimize norepinephrine dose requirements. Use of vasopressin in this setting is not in response to escalating toxicity, so our grading scheme is agnostic to its use. There was also discussion regarding the inotrope milrinone, which is often used to aid in contractility and does not escalate grade of CRS.

While prior versions of CRS grading relied on capturing the fraction of inspired oxygen (FiO2) required to maintain normoxia, this data point can fluctuate hour-to-hour making interpretation and auditing data difficult. To remedy this problem, we elected to separate grade of CRS due to hypoxia by the device used to deliver oxygen. Simple, low-flow nasal cannula (≤ 6 liters/minute), for example, is considered Grade 2 while high-flow devices are Grade 3. This distinction serves as a surrogate for the severity of oxygenation deficit.

What constitutes hypoxia, or more accurately what oxygen saturation is sufficiently low or what clinical signs are sufficient to warrant supplemental oxygen, varies widely between centers, nursing practice, and according to patient age. Normalizing all centers to a single set of criteria is an exceedingly difficult task. For similar reasons, we cannot dictate criteria for which supplemental oxygen is no longer needed in all situations. Therefore, we allow practitioner discretion and recommend grading to be determined by the minimal oxygen delivery device that is required to correct the perceived deficit(s).

Grade 4 CRS.

We define Grade 4 CRS as fever (≥38.0 °C) with hypotension requiring multiple vasopressors (excluding vasopressin) and/or hypoxia requiring positive pressure (eg: CPAP, BiPAP, intubation and mechanical ventilation) not attributable to any other cause. Irrespective of total cumulative dose, the institution of multiple vasopressors constitutes Grade 4 CRS. An exception for vasopressin is again made based on reasoning above. Outside of vasopressin, adding a second agent is a strong indication that the patient remains hemodynamically unstable after the first intervention. Such a scenario would be consistent with grade 4 CRS.

As CRS progresses, capillary leak often leads to pulmonary edema and impairment of ventilation in addition to oxygenation. These patients tend to respond to positive pressure ventilation, which may be accomplished in several ways up to and including intubation and mechanical ventilation. Any employment of positive-pressure ventilation constitutes a Grade 4 CRS.

Intubation may be indicated in patients who have a degree of neurotoxicity where there is concern for their ability to maintain a patent airway. This may occur either in the setting of CRS or after CRS has resolved. The severity of the neurotoxicity driving the decision for intubation will be captured by the grading of that neurotoxicity and should not be captured again as a Grade 4 CRS when the other criteria for such are not met. In other words, intubation of a patient without hypoxia for the possible neurologic compromise of a patent airway alone or for a procedure is not, by definition, Grade 4 CRS. By extension, a patient having seizures where a compromised airway affects oxygenation and where intubation reverses such deficits is not considered to have grade 4 CRS since the seizure rather than CRS is the cause of the hypoxia. Further, a patient who remains intubated for a neurologic cause is not considered to have CRS when the other signs of CRS have resolved.

Grade 5 CRS.

By convention, Grade 5 CRS is defined as death due to CRS where another cause is not the principle factor leading to this outcome.

CRS Severity is Determined by Hypotension and Hypoxia

The clinical manifestations of CRS are varied and frequently involve multiple organ systems. Arrhythmia, cardiomyopathy, prolonged QTc, heart block, renal failure, pleural effusions, transaminitis, and coagulopathy are but a few significant complications of CRS. While important to document all adverse events experienced by CAR T cell patients, we determined that such significant events are uncommon in the absence of either significant hypotension, hypoxia or both. Moreover, these organ dysfunctions are usually managed symptomatically as per standard guidelines and do not influence the decision to use CRS-specific interventions such as anti-cytokine therapy and corticosteroids. Hence, hypotension and hypoxia are the principle determinants of our consensus grading scale. For these reasons and to simplify reporting, references to other specific organ toxicities have been removed from CRS grading. However, organ toxicities associated with CRS may be graded according to CTCAE (currently v5.0) and reported as required.

Definition of Fever, Hypotension, and Hypoxia as it Relates to CRS Grading

Fever is defined in the CTCAE v5.0 as “a disorder characterized by elevation of the body’s temperature above the upper limit of normal” and a temperature ≥38.0 ^oC is considered grade 1 fever (13). We propose to use this same definition to define fever associated with CRS.

The CARTOX criteria defined hypotension as a systolic blood pressure (SBP) less than 90 mmHg in adults while other CRS grading scales did not specifically define hypotension (11–14, 17). However, a SBP of 80–90 mmHg is considered normal in many children. We also noted that some base the definition of hypotension on SBP while others consider just the mean arterial pressure (MAP). Both are acceptable and can be used to determine CRS grade. Therefore, hypotension should be determined on a case-by-case basis accounting for age and the patient’s individual baseline. Indeed, hypotension is defined in CTCAE v5.0 as “a disorder characterized by a blood pressure that is below the normal expected for an individual in a given environment” (13). For practical purposes of CRS grading, an individual requiring IVF boluses or vasopressors to maintain normal blood pressure may be considered to have hypotension.

Hypoxia is another term that is not consistently defined. The CTCAE v5.0 defines hypoxia as, “a disorder characterized by a decrease in the level of oxygen in the body,” but does not specifically define what decrease is considered to be abnormal or even how to measure it (13). In fact, most physicians cannot agree. Many consider hypoxia as an oxygen saturation (SaO2) less than 94% or even 88%, while others base it on other measurements such as the partial pressure of arterial oxygen. For CRS grading, an individual requiring supplemental oxygen to correct a deficit in oxygenation is considered to have hypoxia. Oxygen provided as a comfort measure only should not be used to inform CRS grade.

Definition of CRS Resolution

While most centers are comfortable defining the onset and grade of CRS at its presentation, there is less clarity regarding when CRS is considered to be resolved. This is in large part due to anti-cytokine therapies dramatically and effectively treating fever. Temperatures normalize often within a few hours of administering tocilizumab while the other components of CRS take longer to resolve. Our definition and grading of CRS requires fever. Anti-cytokine therapies are only indicated for patients with CRS, that is for patients that have fever and meet the definition of CRS. Once such therapies are employed, the patient is considered to still have CRS, even in the absence of fever, until all signs and symptoms leading to the diagnosis of CRS have resolved. Likewise, CRS can be downgraded in an afebrile patient treated with anti-cytokine therapy as their hemodynamic status and/or hypoxia improves. Typically, a patient with severe CRS in whom fever, oxygen, and pressor requirements have resolved may be assumed to have resolved CRS unless there are alternative causes for the fever, hypoxia, and/or hypotension. Any neurotoxicity occurring concurrent with or subsequent to the period of CRS does not inform the grade of CRS but is instead captured separately in the neurotoxicity scale.

Symptoms of Immune Effector Cell-Associated Neurotoxicity Syndrome (ICANS)

Symptoms of ICANS (Table 3) have come into better focus through early clinical trials and to date have generally been graded using CTCAE. Although symptoms can be more diverse than for CRS, many patients with neurotoxicity have stereotyped evolution of a specific set of symptoms. The earliest manifestations of ICANS are tremor, dysgraphia, mild difficulty with expressive speech especially naming objects, impaired attention, apraxia, and mild lethargy. Headache is a non-specific symptom, frequently occurring during fever or after chemotherapy in patients without other neurologic dysfunction. Thus, headache alone is not a useful marker of ICANS. Expressive aphasia, on the other hand, appears to be a very specific symptom of ICANS. In one phase I clinical trial, expressive aphasia was observed to be the most characteristic feature, developing in 19 of 22 patients who went on to develop severe neurotoxicity (29). Expressive aphasia, starting as impaired naming of objects, paraphasic errors, hesitant speech, and verbal perseveration, may progress to global aphasia, characterized by expressive and receptive difficulty. Patients with global aphasia may appear wide awake, but are mute and unable to follow commands (akinetic). Many patients have myoclonus or tremor and increased tone. There may be depressed level of consciousness with mild lethargy progressing to obtundation, stupor, or even coma. Mild symptoms may wax and wane with fever initially only to recur a few days later after CRS has resolved. The tempo of progression to severe neurotoxicity may be hours or days. Subclinical electrographic or clinical seizures then may develop, accompanied in some cases by motor weakness. When seizures occur, it is often after the development of severe (global) aphasia. In rare cases, diffuse cerebral edema develops, in some cases after seizures have occurred, but more often cerebral edema may have fulminant onset and few antecedent clinical warning signs, suggesting that it may have a distinct pathophysiology from more reversible neurotoxicity. There appears to be variability in the presentation of neurotoxicity with different CAR products. Nevertheless, there has been progress toward understanding and defining relevant neurotoxicity signs and symptoms in the progression toward severe toxicity that might trigger intervention in the acute setting.

Table 3:

Neurologic and psychiatric adverse reactions reported with approved CAR T products.

tisagenlecleucel (Kymriah)

axicabtagene ciloleucel (Yescarta)

headache--includes headache and migraine encephalopathy--includes: encephalopathy, cognitive disorder, confusional state, depressed level of consciousness, disturbance in attention, lethargy, mental status changes, somnolence, and automatism delirium--includes: delirium, agitation, hallucination, hallucination visual, irritability, restlessness anxiety sleep disorder--includes: sleep disorder, insomnia, and nightmare

encephalopathy--includes: encephalopathy, cognitive disorder, confusional state, depressed level of consciousness, disturbance in attention, hypersomnia, leukoencephalopathy, memory impairment, mental status changes, paranoia, somnolence, stupor headache tremor dizziness--includes: dizziness, presyncope, syncope aphasia: includes aphasia, dysphasia motor dysfunction delirium--includes: agitation, delirium, delusion, disorientation, hallucination, hyperactivity, irritability, restlessness motor dysfunction--includes: muscle spasms, muscular weakness ataxia seizure dyscalculia myoclonus

Definition of Immune Effector Cell-Associated Neurotoxicity

Neurologic symptoms may be observed in association with pathologic processes including hepatic failure, severe hypertension, eclampsia, infection, electrolyte abnormalities, and immunosuppressive and cytotoxic drug therapies. ICANS may have features that overlap with other encephalopathies but has the more specific characteristic of an awake patient who is mute and does not respond verbally or physically to an examiner. ICANS may have a unique pathophysiology compared to other encephalopathies. In a recent report, Gust and colleagues suggested a role for endothelial activation and blood-brain barrier disruption in the pathophysiology of neurotoxicity (30). Another report found elevated levels of the excitatory NMDA receptor agonists glutamate and quinolinic acid in cerebrospinal fluid from patients with neurotoxicity (26). Several reports suggest a role for pro-inflammatory cytokines and myeloid cells besides activated T cells (2, 18, 29, 31–33). The term CAR-related encephalopathy syndrome (CRES) has been proposed to describe neurotoxicity associated with CAR T cell therapy (12). We acknowledge that encephalopathy is a dominant feature of the neurologic changes that occur, however we prefer the term ICANS in order to be inclusive of other symptoms as well as to acknowledge other cellular immunotherapies and therapeutics such as bispecific antibodies that may have similar neurologic side effects. We define ICANS as “a disorder characterized by a pathologic process involving the central nervous system following any immune therapy that results in the activation or engagement of endogenous or infused T cells and/or other immune effector cells. Symptoms or signs can be progressive and may include aphasia, altered level of consciousness, impairment of cognitive skills, motor weakness, seizures, and cerebral edema.” Similar to CRS, ICANS should be applied to any immune effector cell engaging therapy, not just CAR T cells.

Symptoms and Signs Excluded from the Definition of Immune Effector Cell-Associated Neurotoxicity

Although other neurological symptoms and/or signs such as headache, tremor, myoclonus, asterixis, and hallucinations may occur, and possibly be attributable to immune effector cell engaging therapies, we have excluded them from the definition of neurotoxicity as they are less specific, are usually managed symptomatically and do not trigger specific interventions such as corticosteroids to abrogate T cell and other immune cell activation. Weakness and balance problems may occur due to deconditioning and loss of muscle mass from immobility and are frequently seen in the transplant and intensive chemotherapy setting and are excluded from the definition of ICANS. Intracranial hemorrhage with or without associated edema may occur due to coagulopathies in these patients and is also excluded. We recommend that practitioners capture and report such associated events as per CTCAE v5.0 (13).

Consensus on ICANS Grading for Adults

Though early clinical trials used CTCAE v4.03 for grading neurotoxicity, further refinement was achieved when a multi-institutional group of oncologists leading CAR T cell trials across the United States published the CARTOX criteria for adults on grading of neurotoxicity. The CARTOX system grades neurotoxicity by assessing multiple neurological domains that span the constellation of signs and symptoms associated with neurotoxicity (Table 4). An important development was a 10-point screening tool called the CARTOX-10, which incorporated key elements of the Mini-Mental State Examination (MMSE) to evaluate the alterations in speech, orientation, handwriting, and concentration which are highly suggestive of the encephalopathy observed in patients with ICANS (Table 5). This screening tool was designed to overcome the subjectivity in grading many overlapping encephalopathy terms such as encephalopathy, delirium, aphasia, confusion, etc. It moved away from defining the grade of encephalopathy according to impairment of ability to perform activities of daily living, which can be difficult to assess in hospitalized patients.

Table 4:

Published neurotoxicity grading systems.

Grading System (Ref)	Adverse Event Term / Neurotoxicity Domain	Grade 1	Grade 2	Grade 3	Grade 4
CTCAE v5.0 (13)1	Encephalopathy	Mild symptoms	Moderate symptoms; limiting instrumental ADL	Severe symptoms; limiting self care ADL	Life threatening consequences, urgent intervention indicated
	Seizure	Brief partial seizure and no loss of consciousness	Brief generalized seizure	New onset seizures (partial or generalized); multiple seizures despite medical intervention	Life threatening consequences
	Dysphasia	Awareness of receptive or expressive characteristics; not impairing ability to communicate	Moderate receptive or expressive characteristics; impairing ability to communicate spontaneously	Severe receptive or expressive characteristics; impairing ability to read, write, communicate intelligibly
	Tremor	Mild symptoms	Moderate symptoms; limiting instrumental ADL	Severe symptoms; limiting self care ADL
	Headache	Mild pain	Moderate pain; limiting instrumental ADL	Severe pain; limiting self care ADL
	Confusion	Mild disorientation	Moderate disorientation; limiting instrumental ADL	Severe disorientation; limiting self care ADL	Life threatening consequences; urgent intervention indicated
	Depressed level of consciousness	Decreased level of alertness	Sedation; slow response to stimuli; limiting instrumental ADL	Difficult to arouse	Life threatening consequences; coma; urgent intervention indicated
	Cerebral edema			New onset; worsening from baseline	Life-threatening consequences; urgent intervention indicated
CARTOX Criteria (12)	Neurological Assessment Score (CARTOX-10)	7–9 (Mild impairment)	3–6 (Moderate impairment)	0–2 (Severe impairment)	Patient in critical condition, and/or obtunded and cannot perform assessment of tasks
	Raised intracranial pressure	N/A	N/A	Stage 1–2 papilledema2 or CSF opening pressure <20 mmHg	Stage 3–5 papilledema2, or CSF opening pressure ≥20mmHg, or cerebral edema
	Seizures or motor weakness	N/A	N/A	Partial seizure or non-convulsive seizures on EEG with response to benzodiazepine	Generalized seizures or convulsive or non-convulsive status epilepticus, or new motor weakness

¹CTCAE: Under CRS Listing: “Also consider neurologic toxicities such as psychiatric disorders: Hallucinations or Confusion; Nervous system disorders: seizure, dysphasia, tremor, headache.”

²Papilledema grading is performed according to the Modified Frisén scale (35).

ADL: Activities of daily living; CSF: cerebrospinal fluid; EEG: electroencephalogram.

Table 5.

Encephalopathy assessment tools for grading of Immune effector Cell-Associated Neurotoxicity Syndrome (ICANS).

CARTOX-10 (12)	Immune effector Cell-associated Encephalopathy (ICE)
• Orientation: Orientation to year, month, city, hospital, President/Prime Minister of country of residence: 5 points • Naming: Name 3 objects (e.g., point to clock, pen, button): 3 points • Writing: Ability to write a standard sentence (e.g., Our national bird is the bald eagle): 1 point • Attention: Count backwards from 100 by ten: 1 point	• Orientation: Orientation to year, month, city, hospital: 4 points • Naming: Name 3 objects (e.g., point to clock, pen, button): 3 points • Following commands: (e.g., Show me 2 fingers or Close your eyes and stick out your tongue): 1 point • Writing: Ability to write a standard sentence (e.g., Our national bird is the bald eagle): 1 point • Attention: Count backwards from 100 by ten: 1 point

CARTOX-10 (left column) has been updated to the ICE tool (right column). ICE adds a command-following assessment in place of one of the CARTOX-10 orientation questions. The scoring system remains the same.

Score 10: No impairment

Score 7–9: Grade 1 ICANS

Score 3–6: Grade 2 ICANS

Score 0–2: Grade 3 ICANS

Score 0 due to patient unarousable and unable to perform ICE assessment: Grade 4 ICANS

The CARTOX grading system for neurotoxicity also included evaluation of other domains including level of consciousness, motor symptoms, seizures, and signs of raised intracranial pressure (ICP) (Table 4). For evaluation of raised ICP and determination of neurotoxicity grade, the guidelines suggested using elevated cerebrospinal fluid opening pressure and papilledema grade by Frisén scale (Figure 4). Unfortunately, these measurements are cumbersome, potentially inaccurate, and difficult to extend to routine practice. For example, lumbar puncture can be difficult to perform in critically ill patients and, when it is done, opening pressure may vary with age, body habitus, positioning, systemic blood pressure, mechanical ventilation, and pharmacologic sedation (34). In the case of papilledema grading, hospitals have differing capacities for rapid grading of papilledema leading to variable grading (Frisén grade 2 vs. 3) depending on the individual performing the exam, use of fundus photography, etc.

For our consensus grading scheme we propose use of a slightly modified version of CARTOX-10 screening tool, termed here Immune effector Cell-associated Encephalopathy (ICE) Score, to provide objectivity for the grading of multiple overlapping encephalopathy terms currently included on the approved CAR T products (Table 3). The updated encephalopathy screening tool (Table 5) includes an element to assess receptive aphasia seen in these patients. The total number of points, ease of administration, and categorization of scores remain the same as the original CARTOX-10 (12). It is important to note that the 10-point ICE screening tool is helpful to assess patients for encephalopathy. However, the grading of ICANS requires assessment of the 10-point ICE score as well as evaluation of other neurological domains such as level of consciousness, motor symptoms, seizures, and signs of raised ICP/cerebral edema, which may occur with or without encephalopathy.

In contrast to CTCAE v4.03 where a generalized seizure was considered grade 2, our consensus guidelines are more aligned with CTCAE v5.0, which considers a new seizure of any type as grade 3 and any life-threatening seizure as grade 4. Compared to the original CARTOX CRES grading and CTCAE v5.0, the new consensus grading has been simplified so that a single seizure clinical or subclinical electrographic of any type is grade 3 and prolonged or repetitive clinical or subclinical electrographic seizures without return to baseline in between are grade 4 (Table 6). Patients may have EEG changes such as generalized or frontal slowing or frontal intermittent rhythmic delta activity (FIRDA) and this should not be considered seizure.

Table 6.

ASBMT Immune effector Cell-Associated Neurotoxicity Syndrome (ICANS) Consensus Grading for Adults

Neurotoxicity Domain	Grade 1	Grade 2	Grade 3	Grade 4
ICE Score ^	7–9	3–6	0–2	0 (patient is unarousable and unable to perform ICE)
Depressed level of consciousness ❖	Awakens spontaneously	Awakens to voice	Awakens only to tactile stimulus	Patient is unarousable or requires vigorous or repetitive tactile stimuli to arouse. Stupor or coma
Seizure	N/A	N/A	Any clinical seizure focal or generalized that resolves rapidly; or Non-convulsive seizures on EEG that resolve with intervention	Life-threatening prolonged seizure (>5 min); or Repetitive clinical or electrical seizures without return to baseline in between.
Motor findings §	N/A	N/A	N/A	Deep focal motor weakness such as hemiparesis or paraparesis
Raised ICP / Cerebral edema	N/A	N/A	Focal/local edema on neuroimaging#	Diffuse cerebral edema on neuroimaging; Decerebrate or decorticate posturing; or Cranial nerve VI palsy; or Papilledema; or Cushing's triad

^‡ICANS grade is determined by the most severe event (ICE score, level of consciousness, seizure, motor findings, raised ICP/cerebral edema) not attributable to any other cause. For example, a patient with an ICE score of 3 who has a generalized seizure is classified as having Grade 3 ICANS.

^{^}A patient with an ICE score of 0 may be classified as having Grade 3 ICANS if the patient is awake with global aphasia. But a patient with an ICE score of 0 may be classified as having Grade 4 ICANS if the patient is unarousable.

^❖Depressed level of consciousness should be attributable to no other cause (e.g. no sedating medication)

^§Tremors and myoclonus associated with immune effector cell therapies may be graded according to CTCAE v5.0 but they do not influence ICANS grading.

^#Intracranial hemorrhage with or without associated edema is not considered a neurotoxicity feature and is excluded from ICANS grading. It may be graded according to CTCAE v5.0.

ICE: Immune effector Cell-associated Encephalopathy; ICP: Intracranial pressure; EEG: electroencephalogram.

We have also modified the criteria for assessment of raised ICP to improve the ease of grading compared to CARTOX CRES grading system by removing CSF opening pressure and the requirement to grade papilledema on the modified Frisén scale (35) (Table 6). This doesn’t negate the importance of making a clinical assessment to determine if elevated ICP is present, but acknowledges that other signs and symptoms, including simply the presence or absence of papilledema taken in conjunction with depressed level of consciousness, can be used for making this assessment. We have highlighted the importance of evaluating level of consciousness by making it a more detailed part of the grading.

In the grading system, the final ICANS grade is determined by the most severe event among the different domains.

Grade 1 ICANS.

We define Grade 1 ICANS as a score of 7–9 on the ICE assessment (Table 6). A patient with grade 1 ICANS may have a delay in responses or disorientation to time or place, mild inattention with difficulty in counting numbers backwards, or impairment of handwriting. There may be drowsiness but patients awaken spontaneously, and when prompted, the patient should be able to complete most of the ICE assessment. Grade 1 ICANS may be seen during CRS waxing and waning with febrile episodes.

Grade 2 ICANS.

We define Grade 2 ICANS as a score of 3–6 on the ICE assessment (Table 6). Patients with grade 2 ICANS often have some expressive aphasia limiting the ability to communicate spontaneously. Patients may also be noted to have difficulty writing a standard sentence due to poor handwriting and apraxia. They have difficulty naming objects due to expressive aphasia and/or following commands due to receptive aphasia and poor concentration. In our experience, expressive aphasia is the most specific first sign of severe neurotoxicity and early signs during grade 2 include paraphasic errors (the production of unintended syllables and words during attempts to speak) and verbal perseveration with patients repeating the same words over and over. Patients with grade 2 ICANS are able to communicate their needs but it is effortful. Patients may have depressed level of consciousness but are arousable to voice and the responses may be slowed.

Grade 3 ICANS.

We define Grade 3 ICANS as a score of 0–2 on the ICE assessment (Table 6). Patients with grade 3 ICANS have severe global aphasia and are not speaking or following commands even when wide awake and therefore may be unable to complete any of the ICE questions. Alternatively, they may have excessive drowsiness and need tactile stimulus to attend to examiner. Any clinical seizure whether simple partial, complex partial or generalized, and any electrographic seizures would also meet criteria for grade 3 ICANS (Table 6). This acknowledges that seizure may be the peak of an excitatory neurotoxicity process that first manifests clinically as progressive aphasia and then peaks with onset of a clinical or electrographic seizure. If neuroimaging shows new focal or local edema this would also be categorized as grade 3 ICANS (Table 6). However, intracranial hemorrhage due to coagulopathy or other causes with or without associated edema is not considered a neurotoxicity feature and is excluded from ICANS grading.

Grade 4 ICANS.

We define Grade 4 ICANS as patients who have a score of 0 on the ICE assessment due to being unarousable and unable to perform the ICE assessment. Stupor and coma may be seen; the stuporous patient only responds by grimacing or drawing away from vigorous or repetitive tactile stimuli and the comatose patient is unarousable and/or unresponsive (Table 6). This depressed level of consciousness should be attributable to no other cause (e.g. no sedating medication), which is often a complicating factor in sick patients with CRS. Some patients may require intubation for airway protection. In addition, any patient having prolonged or repetitive clinical or subclinical electrographic seizures without return to baseline in between, or deep focal motor weakness such as hemiparesis or paraparesis would be considered to have Grade 4 ICANS (Table 6). Patients with symptoms and signs of elevated ICP such as projectile vomiting with headache, depressed consciousness, cranial nerve VI palsies, papilledema, Cushing’s triad of bradycardia, hypertension and respiratory depression, decerebrate or decorticate posturing, or diffuse cerebral edema on head imaging would also be considered to have grade 4 ICANS (Table 6).

The new grading is similar to the CARTOX CRES grading guideline in regard to the CARTOX-10 screening assessment in that it would classify any patient too obtunded to perform the assessment as having grade 4 ICANS. In contrast to CARTOX CRES, the updated grading classifies an isolated generalized seizure with return to baseline as Grade 3, and reserves Grade 4 classification for prolonged >5 minutes or repetitive clinical or subclinical (EEG only) seizures without return to baseline in between. This is consistent with life-threatening seizures as defined by CTCAE v5.0.

Grade 4 patients typically need to be intubated for airway control and seizure management. A patient may be intubated for grade 4 ICANS, but this should not be captured again as grade 4 CRS when other signs of severe CRS have resolved.

Grade 5 ICANS.

By convention, Grade 5 ICANS is defined as death due to ICANS where another cause is not the principle factor leading to this outcome.

Consensus on ICANS Grading for Children

While the 10-point ICE assessment is useful for screening adults for encephalopathy, its use in children may be limited to those who are ≥12 years with sufficient level of cognitive abilities to perform the ICE assessment. For children <12 years, the Cornell Assessment of Pediatric Delirium (CAPD) (36, 37) (Table 7) is recommended to assist in the overall grading of ICANS as recently proposed by Mahadeo et al (38). A CAPD score of ≥9 is suggestive of delirium and should be considered as grade 3 ICANS. The CAPD score may also be used in patients >12 years with baseline developmental delay as it has been validated up to age 21. Other domains evaluated to grade ICANS in children are similar to those used in adults and include level of consciousness, motor symptoms, seizures, and signs of raised ICP (Table 8).

Table 7.

Encephalopathy Assessment for Children <12 years using Cornell Assessment of Pediatric Delirium (CAPD) (36, 37). Adapted from Traube et al (Ref 36) and reproduced with permission from Wolters Kluwer.

Answer the following based on interactions with the child over the course of the shift	Never 4	Rarely 3	Sometimes 2	Often 1	Always 0
1. Does the child make eye contact with the caregiver?
2. Are the child’s actions purposeful?
3. Is the child aware of his/her surroundings?
4. Does the child communicate needs and wants?
	Never 0	Rarely 1	Sometimes 2	Often 3	Always 4
5. Is the child restless?
6. Is the child inconsolable?
7. Is the child underactive – very little movement while awake?
8. Does it take the child a long time to respond to interactions?

For patients age 1–2 years, the following serve as guidelines to the corresponding questions:

^1.Holds gaze. Prefers primary parent. Looks at speaker.

^2.Reaches and manipulates objects, tries to change position, if mobile may try to get up

^3.Prefers primary parent, upset when separated from preferred caregivers. Comforted by familiar objects (i.e., blanket or stuffed animal)

^4.Uses single words or signs

^5.No sustained calm state

^6.Not soothed by usual comforting actions, for example, singing, holding, talking, and reading

^7.Little if any paly, efforts to sit up, pull up, and if mobile crawl or walk around

^8.Not following simple directions. If verbal, not engaging in simple dialogue with words or jargon

Table 8.

ASBMT Immune effector Cell-Associated Neurotoxicity Syndrome (ICANS) Consensus Grading for Children

Neurotoxicity Domain	Grade 1	Grade 2	Grade 3	Grade 4
ICE Score for children >12 years ^	7–9	3–6	0–2	0 (patient is unarousable and unable to perform ICE)
CAPD score for children ≤12 years	<9	<9	≥9	Unable to perform CAPD
Depressed level of consciousness ❖	Awakens spontaneously	Awakens to voice	Awakens only to tactile stimulus	Patient is unarousable or requires vigorous or repetitive tactile stimuli to arouse. Stupor or coma
Seizure (any age)	N/A	N/A	Any clinical seizure focal or generalized that resolves rapidly; or Non-convulsive seizures on EEG that resolve with intervention	Life-threatening prolonged seizure (>5 min); or Repetitive clinical or electrical seizures without return to baseline in between.
Motor weakness (any age) §	N/A	N/A	N/A	Deep focal motor weakness such as hemiparesis or paraparesis
Raised ICP / Cerebral Edema (any age)			Focal/local edema on neuroimaging#	Decerebrate or decorticate posturing; or Cranial nerve VI palsy; or Papilledema; or Cushing's triad; or Signs of diffuse cerebral edema on neuroimaging

^‡ICANS grade is determined by the most severe event (ICE or CAPD score, level of consciousness, seizure, motor findings, raised ICP/cerebral edema) not attributable to any other cause.

^{^}A patient with an ICE score of 0 may be classified as having Grade 3 ICANS if the patient is awake with global aphasia. But a patient with an ICE score of 0 may be classified as having Grade 4 ICANS if the patient is unarousable.

^❖Depressed level of consciousness should be attributable to no other cause (e.g. no sedating medication)

^§Tremors and myoclonus associated with immune effector cell therapies may be graded according to CTCAE v5.0 but they do not influence ICANS grading.

^#Intracranial hemorrhage with or without associated edema is not considered a neurotoxicity feature and is excluded from ICANS grading. It may be graded according to CTCAE v5.0.

ICE: Immune effector Cell-associated Encephalopathy; CAPD: Cornell Assessment of Pediatric Delirium; ICP: Intracranial pressure; EEG: electroencephalogram.

Data Collection and Reporting for Approved Cell Therapy Products

High unmet medical need and large effect sizes seen in clinical testing of CD19 CAR T cell therapies allowed approval of the two current products based on data collected from <300 total patients on three single arm, Phase II trials. As a result, there is broad agreement that further data collection on toxicity and patient outcomes is paramount. The US FDA has required that pharmaceutical companies that produce commercial CAR T cell products must establish a mechanism to follow recipients of these therapies for 15 years, though this requirement is mainly focused on monitoring for potential long-term genotoxicity. Other health authorities may impose further or more detailed requirements.

This raises important questions on how best to collect appropriate data, especially since these mandates are placed on the companies but carried out by the centers in a largely volunteer effort. Compliance with what could be a significant and unfunded mandate will be enhanced by making sure that data collection is appropriately focused, especially since these data will not be collected in a research setting with research budgets and direct regulatory mandates on the centers. Health authorities will need to harmonize their data requests. In this new field of medicine, companies may have to face the possibility of a gap between what the FDA or European Medicines Agency (EMA) wants them to collect and what centers are actually able to provide. There needs to be a simplified and unified approach to reporting with a single portal for data entry. A registry approach is likely to be the best method of providing as complete data as possible.

The CIBMTR operates a large outcomes database, which for decades has been a valuable resource to the field of transplantation. Most centers with active immune effector cell programs have many years of experience using this database. Recognizing the emergence of the field of cellular immunotherapy, in 2016 the CIBMTR launched a database dedicated to cellular therapy outcomes. This registry tracks long-term follow-up of patients who have received cellular therapies, including CAR T cells and other cellular therapies beyond hematopoietic cell transplants. As part of this process, the CIBMTR Cellular Therapy Task Force developed new reporting forms specific to cellular therapy (39), which were subsequently piloted and refined.

Since the launch in 2016, centers voluntarily have reported data derived from more than 200 recipients of CAR T cells, which are most commonly used for treatment of non-Hodgkin lymphoma and acute lymphoblastic leukemia. Using a single, standardized database to capture information about recipients of immune effector cell therapies can streamline the process and can provide a resource for research. The CIBMTR’s Cellular Therapy Registry infrastructure is well suited to meet this requirement.

Regulators in other regions are considering more requirements to assess the safety and efficacy of immune effector cell therapies. The EMA organized a workshop in February 2018 to identify a minimal set of data elements for commercial CAR T cells (40). In addition to common safety endpoints, the EMA report outlined the capture of grades 3 and 4 organ toxicities (41). Lee et al initially included grades 2–4 nonhematologic organ toxicities in the CRS grading criteria (14). However, organ toxicities are excluded in the new CRS consensus grading scheme proposed here. Requiring reporting of organ toxicities in the post-market setting would add considerable burden for data collection, and it runs the risk of being infeasible. This example highlights the need for harmonization in the data requested.

The CIBMTR Cellular Therapy Registry’s follow-up form captures toxicities after immune effector cell infusion, at 3 months, 6 months, 1 year and yearly thereafter. The outcomes routinely captured in the CIBMTR follow up forms relevant to CAR T-cell toxicities include CRS, neurotoxicities, neutrophil and platelet recovery, hypogammaglobulinemia, severe infections, non-hematologic grade 4 toxicities and death from any cause. For CRS, the registry computes a grade by capturing key information related to CRS, including treatment, i.e. use of vasopressors. This approach can accommodate changes in the grading criteria, as proposed in this consensus statement, or comparisons across grading systems. For ICANS, the forms capture the presence of different manifestations, and whether they resolved. Issues raised on the applicability of CTCAE to assess severity in hospitalized patients apply here as well. Beyond the abovementioned toxicities, centers can report subsequent neoplasms and pregnancies at any time through event-driven forms. These forms aid collection of time-sensitive information or biospecimens, if needed.

We believe that it would be safest and most efficient to use CIBMTR database reporting to meet the mandates placed on the drug companies at a level that is feasible for centers offering immune effector cell therapies outside the research setting. Barring unusual or notable toxicity that any treating physician can choose to report to the FDA on a Medwatch form (as with any other approved therapy), we endorse a system in which the CIBMTR registry would be a single approach that centers can use for reporting for current and future approved cell therapies. The CIBMTR built the Cellular Therapy Registry to serve the community and to help advance the field by making the data available to investigators. Standardized collection of toxicity data in the real-world setting will help identify ways to make these therapies safer.

Conclusion

In conclusion, we have proposed consensus definitions and grading for CRS and ICANS, the two most common toxicities associated with immune effector cell therapies. We acknowledge that as new data becomes available from existing and novel immune effector cell therapies, this grading system may need to be revised in the future. However, we believe that the proposed grading system is objective, easy to use, and more accurately categorizes the severity of these toxicities. We strongly recommend the usage of this consensus grading system for reporting of CRS and neurotoxicity associated with immune effector cell-engaging therapies across all clinical trials as well as in the post-approval standard of care setting. This would allow comparison of the safety of different immune effector cell-engaging therapies and will also likely facilitate the development of optimal strategies for prevention and/or management of these toxicities.