Clinical presentation, risk factors, and outcomes of immune effector cell-associated neurotoxicity syndrome following CAR-T cell therapy: A Systematic Review

Abstract

Importance:

Chimeric Antigen Receptor (CAR)-T cell therapy is a novel therapy for patients with relapsed or refractory hematological malignancies. Most CAR-T cell therapy recipients will experience clinical features of the immune effector cell-associated neurotoxicity syndrome (ICANS), a potentially life-threatening condition.

Objective:

To describe the clinical, biological, and radiological findings associated with ICANS among adults with hematological malignancies treated with CAR-T cell therapy and the acute or long-term outcomes of ICANS.

Evidence Review:

A literature search of Ovid Medline, EMBASE, PubMed, Scopus, Web of Science Core Collection, Cochrane Library, and Google Scholar was conducted from each database’s inception through February 1, 2022, using search terms reflecting CAR-T and ICANS. We included studies that enrolled adults (age 18 years or older) who received CAR-T cell therapy as management for hematological malignancies and reported on the clinical presentation, predictors, or the acute or long-term outcomes of ICANS. Two reviewers independently extracted data following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis reporting guidelines. Quality was assessed using Joanna Briggs Institute critical appraisal tool for cohort studies.

Findings:

Of 2928 studies screened, 23 observational studies (10 prospective, 11 retrospective, one mixed-design, and one cross-sectional) met the eligibility criteria and included 1666 participants in total. The most common hematological malignancies were diffuse large B-cell lymphoma, acute lymphocytic leukemia, non-Hodgkin lymphoma, and chronic lymphocytic leukemia. ICANS onset was most often associated with the presence and severity of cytokine release syndrome, c-reactive protein (CRP), and ferritin levels. Aphasia was the most common ICANS-related symptom reported, although the neurological manifestations of ICANS were highly variable. Neuroimaging studies (MRI or CT) were often normal in cases of ICANS; however, electroencephalograms often showed generalized background slowing, abnormal rhythmic and periodic discharge patterns. The pooled mean (± Standard Deviation) onset of ICANS was 6.4 ± 3.2 days, with a pooled mean duration of 8.3 ±10.5 days. Two (9%) of the 23 studies reported 5 ICANS related deaths among 233 participants. A subset of patients experienced persistent neurocognitive complaints ≥ 1-year post-CAR-T cell therapy.

Conclusion and Relevance:

The clinical presentation, onset, severity, long-term sequelae, and grading system of ICANS are variable. Future studies should consider using a consensus grading/reporting scale that would permit cross-trial comparisons of the safety profile of various CAR-T products and enable interventions to be developed to mitigate or manage these neurotoxicities.

Funding:

SJG is supported by National Cancer Institute Grant No. (5-K12-CA120780-13; PI: William Kim) and National Institute on Aging Grant No, (1 R03 AG074030-01; PI: Shakira Grant). SG is supported by the UAB Walter B. Frommeyer Fellowship in Investigative Medicine.

Registration:

PROSPERO CRD42020207864

Introduction

Chimeric Antigen Receptor (CAR)-T cell therapy represents a novel immunotherapeutic strategy that has changed the treatment landscape for those with heavily pre-treated acute lymphoblastic leukemia,^1,2 large B cell lymphoma,^3,4 mantle cell lymphoma,⁵ and most recently multiple myeloma.⁶

Immune effector cell-associated neurotoxicity syndrome (ICANS), a clinical syndrome characterized by a constellation of neuropsychiatric symptoms, is one major toxicity associated with CAR-T cell therapy that can affect approximately 20–70% of CAR-T cell therapy recipients.⁷ While the ICANS course is often mild in severity and limited in duration,^8–10 some patients will develop severe, life-threatening ICANS,^9,11,12 or have long-term neuropsychiatric effects persisting one year or more following treatment.^13,14 Despite these observations, identifying those at greatest risk for severe ICANS and the subset of patients likely to experience long-term neuropsychiatric effects remain poorly understood.

Our systematic review summarizes the biological, clinical, radiological, and electroencephalography (EEG) features of ICANS and outcomes in adults with hematologic malignancies treated with CAR-T cell therapy.

Methods

This systematic review was conducted according to a published protocol (CRD42020207864) and followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis and Synthesis Without Meta-analysis in Systematic Review reporting guidelines (Supplementary Table 1).^15,16

Search Strategy

A medical librarian (AG) searched the following databases: Cochrane Library, Google Scholar, Ovid Medline, Ovid EMBASE, PubMed, Scopus, and Web of Science Core Collection from the inception of each database to February 1, 2022 (Supplementary Table 2) using controlled vocabulary and keywords with synonyms that reflected CAR-T therapy and ICANS concepts. Reference lists and citations associated with the included studies were also reviewed.

Study Selection

Studies included were observational or interventional, enrolling at least five adult CAR-T cell recipients with hematological malignancies and reporting on ICANS as adjudicated by the primary papers. Studies reporting clinical, radiographic, EEG findings, biological correlates, or outcomes of ICANS were also included.

Selected citations from the initial search were imported into the Endnote X9^™ database. After removing duplicates, the remaining articles were uploaded into Covidence^™. All studies were selected and reviewed by two reviewers (S.J.G. and S.G.). A third reviewer (D.M) was consulted when consensus could not be reached to resolve discrepancies.

Data Extraction

Data were extracted by (S.J.G and S.G) using a pre-piloted excel document (Supplementary Table 3). Extracted data included study-level and individual participant characteristics, ICANS symptoms and signs as adjudicated by the primary studies, radiographic and EEG findings, biologic correlates, and outcomes of patients with ICANS. Each study’s quality was evaluated using the Joanna Briggs Institute (JBI) Critical Appraisal Tool.¹⁷

Definition of Outcomes

The primary outcome of interest was ICANS as adjudicated by the primary studies and guided by the American Society of Transplantation and Cellular Therapy’s (ASTCT) definition of ICANS as a pathologic process involving the central nervous system, developing after immune therapy and results in the activation or engagement of endogenous or infused T cells or other immune effector cells. Data on the clinical features of ICANS were abstracted for each eligible study and included aphasia, altered level of consciousness, impairment of neurocognitive skills, motor weakness, seizures, and cerebral edema anxiety, delirium, dizziness, dyscalculia, headache, myoclonus, seizure, and tremor.^8,10,18

Statistical Analysis

While our original intent was to perform a quantitative meta-synthesis of the results, considerable heterogeneity in the study design, patient characteristics, and clinical outcomes precluded meta-analysis. Instead, we chose to summarize the results descriptively, presenting pooled data when appropriate. Specifically, we computed pooled mean and pooled standard deviations using weighted averages across individual studies. For studies reporting medians instead of means, we used the Quantile Estimation method as outlined by McGrath et al. to estimate the mean and standard deviation.¹⁹

Results

Our initial literature search yielded 2928 potential studies after removing duplicate publications. Following the further exclusion of studies without ICANS data, and those published in abstract form alone, (Supplementary Table 4), 140 full-text articles were reviewed, yielding 23 studies for the systematic review (Figure 1).

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Flowchart

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Flowchart depicts the identification, screening, and inclusion criteria yielding 19 studies for systemic review. Adapted from: Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021;372: n71. doi: 10.1136/bmj.n71. For more information, visit: http://www.prisma-statement.org/

Study Characteristics

Table 1 summarizes the study characteristics. All studies were observational (10 prospective cohort,^9,10,20–27 11 retrospective cohort,^8,13,28–36 1 mixed prospective and retrospective cohort,¹² and 1 cross-sectional).¹⁴ Most studies (17/23) were published between 2020–2021,^{13,14,21–27,29–34,36} with the remainder published between 2017–2019.^{8–10,12,20,28} One study evaluated post-marketing adverse events (n=804), therefore did not report on the unique number of patients;³³ the 18 remaining studies enrolled 1,666 participants.^{8–10,12–14,20–32,34–36} Most studies (19/23) only enrolled adult participants (aged 18–85 years).^{8,9,12–14,21–27,30,31,35,36} Four studies included mixed cohorts of pediatric and young adults (age range 2–30.5 years).^20,23,28,34 The pooled mean across 18 (78%) of 23 studies was 53.5 years ± Standard Deviation 10.5.^{8,9,12–14,20–28,30,31,35,36} CD19 was the most common CAR-T target evaluated across 23 studies.^{8–10,13,14,21–25,28–34} One study evaluated a CD22 CAR-T product,²⁰ and for one study the CAR-T cell target was not reported.¹² The most prevalent hematological malignancies were diffuse large B-Cell lymphoma (DLBCL) (8 studies),^{20–22,25,30,31,33,34} acute lymphocytic leukemia (ALL) (7 studies),^{10,13,14,20,26–29,34–36} non-Hodgkin’s lymphoma (NHL) (4 studies),^13,14,29,32 and chronic lymphocytic leukemia (CLL) (4 studies).^8,13,14,29 Nine (39%) of 23 studies reported the CAR-T product for study participants.^{21,22,24,28,30–34} In three (13.0%) of 23 studies, reported on the use of axicabatagene ciloleucel,^22,24,31 two (8.7%) reported use of tisagenlecleucel,^28,32 and four (17.4%) included patients who received either both axicabatagene ciloleucel or tisagenlecleuel.^21,30,33,34

Table 1.

Characteristics of the 19 Studies Included in the Systematic Review

CHARACTERISTICS	NO. OF STUDIES (N=19)	REFERENCES
Year Published
2017–2019	6	Gust 2017,9 Gofshteyn 201828, Santomasso 201810, Shalabi 201820, Karschnia 20198, Rubin 201912
2020–2022	17	Belin 202021, Cordeiro 202013, Gajra 202033, Maziara 202032, Ruark 202014, Rubin 202024, Sokolov 202029, Strati 202022, Wudhikarn 202030, Gust 202123, Holtzman 202131, Brown 202134, Maillet 202125, Yuen 202135, Greenbaum 202126, Beuchat 202236, Wudhikarn 202127
Sample Size
< 50	7	Shalabi 201820, Karschnia 20198, Ruark 202014, Sokolov 202029, Holtzman 202131, Brown 202134, Yuen 202135
50–99	8	Gofshteyn 201828, Santomasso 201810, Belin 202021, Cordeiro 202013, Wudhikarn 202030, Maillet 202125, Beuchat 202236, Wudhikarn 202127
100–199	6	Gust 2017,9 Rubin 201912, Maziara 202032, Strati 202022, Gust 202123, Greenbaum 202126
≥200	2	Rubin 202024, Gajra 202033
Age
Pediatric and young adult cohort (mean, SD)*	4	Gofshteyn 2018 (11.5y, 4.3)28, Shalabi 2018 (17.9y, 6.9)20, Gust 2021 (cEEG-14.3y, 7.5; rEEG- 10.7y, 8.2)23, Brown 202134
Adult cohort (mean, SD)*	15	Gust 2017,9 Santomasso 201810, Karschnia 2019 (56y, 16.3)8, Rubin 2019 (54.6y, 12.3)12, Belin 2020 (51.7y, 12.8)21, Cordeiro 2020 (51.7y, 11.5)13, Gajra 202033, Maziara 202032, Ruark 2020 (51.3y, 12.3)14, Rubin 2020 (60y,12.5)24, Sokolov 202029, Strati 2020 (54.3y, 14.4)22, Wudhikarn 2020 (55.7y, 15.5)30, Holtzman 2021(53.7y, 12.2)31, Maillet 2021(58y,14)25, Yuen 202135, Greenbaum 202126, Beuchat 2022 (60y, 11)36, Wudhikarn 202127
Study Design
prospective cohort	8	Gust 2017,9 Santomasso 201810,Shalabi 201820, Belin 202021, Rubin 202024, Strati 202022, Gust 202123, Maillet 202125, Greenbaum 202126, Wudhikarn 202127
retrospective cohort	9	Gofshteyn 201828, Karschnia 20198, Cordeiro 202013, Gajra 202033, Maziara 202032, Sokolov 202029, Wudhikarn 202030, Holtzman 202131, Brown 202134, Yuen 202135, Beuchat 202236
mixed prospective and retrospective cohort	1	Rubin 201912
cross-sectional	1	Ruark 202014
Cancer Type
Diffuse large B-Cell lymphoma (DLBCL)	11	Shalabi 201820, Belin 202021, Gajra 202033, Strati 202022, Wudhikarn 202030, Holtzman 202131, Brown 202134, Maillet 202125, Yuen 202135, Greenbaum 202126, Wudhikarn 202127
Acute lymphocytic leukemia (ALL)	8	Gofshteyn 201828, Santomasso 201810,Shalabi 201820, Cordeiro 202013, Ruark 202014, Sokolov 202029, Brown 202134, Beuchat 202236
Non-Hodgkins lymphoma (NHL)	4	Cordeiro 202013, Maziara 202032, Ruark 202014, Sokolov 202029
Chronic lymphocytic leukemia (CLL)	4	Karschnia 20198, Cordeiro 202013, Ruark 202014, Sokolov 202029
Lymphoma (not classified)	4	Gust 2017,9 Rubin 201912, Rubin 202024, Beuchat 202236
Leukemia (not classified)	2	Rubin 201912, Rubin 202024
Follicular Lymphoma	2	Belin 202021, Sokolov 202029
Hepatocellular Carcinoma	1	Karschnia 20198
Multiple Myeloma	1	Beuchat 202236
Sarcoma	1	Rubin 201912
CAR-T Cell Target
CD19	17	Gust 2017,9 Gofshteyn 201828, Santomasso 201810, Karschnia 20198, Belin 202021, Cordeiro 202013, Gajra 202033, Maziara 202032, Ruark 202014, Rubin 202024, Sokolov 202029, Strati 202022, Wudhikarn 202030, Gust 202123, Holtzman 202131, Brown 202134, Maillet 202125, Yuen 202135, Greenbaum 202126, Beuchat 202236, Wudhikarn 202127
CD22	1	Shalabi 201820
BCMA	1	Yuen 202135
Not indicated	1	Rubin 201912
CAR-T Product
Axicabtagene ciloleucel	3	Rubin 202024, Strati 202022, Holtzman 202131
Tisagenlecleucel	2	Gofshteyn 201828, Maziara 202032
Axicabtagene ciloleucel and Tisagenlecleucel	4	Gajra 202033, Belin 202021, Wudhikarn 202030, Brown 202134

^*Mean and SD was calculated using formula by McGrath et al using minimum, maximum and median when mean was not provided.^{8,12–14,21–23,28,30,31} Mean used when provided.^20,24,25

Risk-of-Bias Assessment

The quality rating for each study was independently assessed using the JBI checklist for cohort studies. The overall quality rating of all studies but one was considered good. The Quality Rating Scheme and results are summarized in Supplementary Table 5.

Studies evaluating the incidence of ICANS

Ninteen (83%) of 23 studies reported the incidence of ICANS which ranged from 37.5–77%.^{9,10,12,14,20–22,24–28,30–36} The remaining studies included cohorts of participants with ICANS only,^8,29 or data was not included to allow estimation of ICANS incidence.^13,23 ICANS was most often graded using the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) in 17 (74%) of 23 studies.^{8–10,12,14,21,22,24–27,29,31,32,34,36} Seventeen (74%) of 23 studies reported on the number of grade 3–4 ICANS events.^{8–10,12,14,21,22,24–27,29,31,32,34–36} Two studies reported five grade 5 (death directly related to ICANS) events.^9,12 Eleven (48%) of 23 studies reported the time to onset of ICANS following infusion of the CAR-T product. Across these studies the pooled mean onset of ICANS was 6.4 days ± 3.2.^{8,10,21,22,24,26,27,30,31,34,35} Seven (30%) of 23 studies also reported on the duration of ICANS symptoms. Across these studies the pooled mean duration of ICANS was 8.3 days ± 10.5.^{8,9,21,22,25,27,34}

Studies evaluating risk factors for ICANS

Thirteen (57%) of 23 studies^{8–10,12,21,22,24,26–28,30,33,34} reported factors associated with ICANS development (Figure 2). Six (50%) of 12 studies reported that the presence and severity of cytokine release syndrome (CRS) predicted the onset of ICANS.^{10,24,27,28,30,33} In two other studies, earlier time to onset of CRS (median range of 1–2 days)^12,24 after CAR T-cell infusion was also associated with ICANS development. Other identified factors included elevated C-reactive protein levels in six (50%) of 12 studies,^{8,12,21,22,24,27} and elevated serum ferritin levels in six (50%) of 12 studies.^{8,21,22,24,27,34} Older adult patient age was also associated with ICANS onset in three (27%) of 11 studies.^10,24,33 One study by Greenbaum et al.²⁶ found the Endothelial Activation and Stress Index (EASIX) score (lactate dehydrogenase [LDH; U/L] × creatinine [mg/dL]/platelets [PLTs; 109 cells/L]) in combination with serum ferritin and c-reactive protein levels to be predictive for ICANS development.

FIGURE 2.

Risk Factors Associated with ICANS Onset Across 13 studies

Demographic, disease-related, blood-based, and clinical risk factors associated with ICANS onset across 11 studies. The most frequently noted risk factors were C-reactive protein levels, presence and severity of cytokine release syndrome, and high ferritin levels. Abbreviations: cytokine release syndrome (CRS), interleukin (IL)

Studies evaluating the clinical presentation of ICANS

Seventeen (74%) of 23 studies reported on the clinical presentation of ICANS (Figure 3).^{8–10,12,20–22,25–31,33–36} Aphasia was the most common symptom of ICANS reported in 15 (88%) of 17 studies reporting on clinical features of ICANS.^{8–10,12,20–22,25,27–31,33,34} Generalized or absence seizures were reported as a clinical feature of ICANS in 13 (87%) of the 17 studies..^{8,10,12,20,21,25,27–31,33,34,36} Additional symptoms of ICANS included encephalopathy^{8,10,12,21,28,33} in six (35%) of 17 studies and headache^{9,10,12,21,27,30,33,35} in eight (47%) of 17 studies. Additionally, 10 (59%) of 17 studies reported on disorders of consciousness^{8–10,12,20,21,25,27,31,34} and nine studies reported confusional states^{8,21,27,30,31} or disorientation.^10,20,21,35 Only three studies reported stroke as a clinical manifestation of ICANS.^12,31,33 Overall, 19 stroke cases were reported, among 254 participants enrolled in two studies.^12,31 Of these, three were ischemic.^12,31 For the remaining 16 cases identified in a retrospective database analysis of the Food and Drug Administration’s adverse events reporting systems, frequencies on the type of stroke were not reported.³³

FIGURE 3.

Clinical Features of ICANS Across 17 studies

Clinical features for ICANS noted across 14 studies were highly variable but most often included aphasia, seizure, headache, hallucinations, and encephalopathy.

Studies reporting on the role of neuroimaging and EEG in the diagnostic evaluation of ICANS

Fourteen (61%) of 23 studies reported on the role of neuroimaging and EEG to evaluate ICANS (Figure 4).^{8–10,12,21–23,27–29,31,34,35} All 14 studies used magnetic resonance imaging (MRI),^{8–10,12,21–23,28,29,31,34} and 12 (86%) of 14 studies used EEG.^{8–10,12,21–23,29,31,34–36} Additional neuroimaging included computerized tomography (CT),^{8,12,21,28,35} CT angiography,¹² and Positron Emission Tomography-Computed Tomography (PET-CT).¹² Eleven (78%) of 14 studies reported MRI or CT of the brain as either normal or unchanged from pre-CAR T-cell therapy infusion.^{8–10,12,21–23,29,31,34,36} Among these 14 studies, 9 (64%) of 14 reported abnormal MRI findings involving the pons,^8–10,23,31 thalamus,^9,10,22,23 white matter,^22,23,31,34 corpus callosum,^8,10 leptomeninges,^9,22 and temporal lobe.^12,22 Ten (71%) of 14 reported the abnormalities on MRI of the brain as T2/FLAIR hyperintensities.^{8–10,22,23,29,31,34–36} Six (43%) of 14 studies reported MRI or CT brain findings such as stroke,^12,22,31 intracranial hemorrhage,^8,9,12,31,34, and diffuse cerebral edema.^9,12 Across 8 (80%) of 12 studies, findings on EEG included generalized background slowing, abnormal rhythmic and periodic discharge patterns.^{8–10,12,21–23,29,31,34–36} Five (42%) of 12 studies reported seizure-related activity on EEG.^{10,12,22,23,29}

FIGURE 4.

Neuroimaging and EEG Finding of ICANS Onset Across 14 studies

Neuroimaging findings with MRI and CT across 11 studies were most often normal, but when abnormalities were noted, they were cited most frequently in the pons. EEG findings across 11 studies most often included generalized slowing, periodic discharges, and epileptic activity. Abbreviations: magnetic resonance imaging (MRI), electroencephalography (EEG).

Studies evaluating biological correlates of ICANS

Only one study evaluated cerebrospinal fluid (CSF) markers and their association with ICANS development.⁹ Six (26%) of 23 studies reported correlative analyses of blood-based markers (Figure 5).^{9,10,20,28,31,34} The most commonly identified blood-based markers associated with ICANS development were Interleukin 6,^9,10,20,34 Interleukin 2,^10,20,28 Interleukin 15,^10,20,28 and Interferon-gamma (IFN-gamma).^9,10,34 Additional markers reported included Granulocyte-macrophage colony-stimulating factor (GM-CSF),^10,20 Interleukin 10,^10,20 Monocyte Chemoattractant Protein-1 (MCP-1),^9,10 and Tumor Necrosis Factor-alpha (TNF-alpha).^9,20

FIGURE 5.

Blood-based Biologic Markers associated with ICANS Across 6 studies

Various blood-based markers were associated ICANS across 6 studies. IL15, IL2, IL6 were not most frequently cited markers. Abbreviations: angiopoietin (ANG), granulocyte colony stimulating factor (G-CSF), granulocyte macrophage colony stimulated factor (GM-CSF), hepatocyte growth factor (HGF), interferon gamma (INFγ), interleukin (IL), monocyte chemoattractant protein-1 (MCP-1), tumor necrosis factor-alpha (TNF-alpha), von Willebrand factor (VWF).

Studies reporting on the management of ICANS

Nine (39%) of 23 studies reported on the management of ICANS, which often included supportive care, steroids (dexamethasone or methylprednisone) either as monotherapy or in combination with the anti-IL-6 receptor monoclonal antibody, tocilizumab.^{8–10,21,22,27,31,34,36} Of these studies, only one reported using the anti-IL-6 receptor monoclonal antibody, siltuximab, and in the same study, the IL-1 receptor antagonist anakinra was used in the management of ICANS.³⁶ Among the studies reporting tocilizumab use, most patients had a preceding diagnosis of CRS. In four (44%) of the nine studies reporting ICANS management, prophylactic anti-seizure medication was reported. The most commonly used anti-seizure medication was levetiracetam.

Studies evaluating long-term neurological outcomes following CAR-T cell therapy

Two related studies reported on the neurological outcomes of patients greater than one year after receiving CAR-T cell therapy.^13,14 In the first study, a retrospective chart review, nine (10%) of 86 patients developed 11 new neurologic findings, including three cerebrovascular accidents and one case of transient ischemic attack, Alzheimer’s dementia, and peripheral neuropathy.¹³ Eight of 86 (9%) patients developed mood disorders such as depression and anxiety, which required interventions.¹³ In the second study, patient-reported measures were used to identify 15 (38%) of 40 patients with at least one cognitive difficulty (difficulties with concentration, memory, word-finding, and problem-solving).¹⁴ Two (9 %) of 23 studies reported neurological outcomes for patients 1 to 12 months after treatment with CAR-T cell therapy.^8,25 One of these studies reported the following neurological complaints: confusion, myoclonus, and extremity weakness affecting two patients.⁸ However, a second study showed no exacerbation or development of new neurological symptoms in 56 patients 6–12 months after receiving CAR-T cell therapy.²⁵

Studies reporting death due to ICANS

Two (9%) of 23 studies enrolling 233 participants reported a total of five deaths due to ICANS.^9,12 In one study, death occurred five days following CAR T-cell infusion in a 21-year-old male with relapsed B-cell acute lymphoblastic leukemia.¹² The ICANS events resulting in death in this patient were acute obtundation, which progressed to coma and brain death. In a study of 133 patients, 4 (3%) died due to ICANS.⁹ In one patient, death was due to multifocal brainstem hemorrhage and edema associated with disseminated intravascular coagulation. In the same study, two patients died from acute cerebral edema and one from cortical laminar necrosis, which resulted in a persistent minimally conscious state until death four months after CAR-T cell infusion.⁹

Discussion

This systematic review included 23 studies. Data presented summarized the clinical presentation, risk factors, neuroimaging and EEG findings, biologic correlates, management, and clinical outcomes of ICANS in adults who received CAR-T cell therapy for hematological malignancies. Our findings indicate that the incidence and clinical presentation of ICANS are highly variable.

ICANS onset was associated with the presence and severity of CRS, older patient age at the time of CAR-T infusion, and elevation in inflammatory markers (c-reactive protein and ferritin). Aphasia, seizures, headache, encephalopathy, and other disorders leading to a decreased level of consciousness, were the most common clinical features of ICANS. ICANS developed early, usually within the first week following the CAR-T infusion, most often resolving within 5–6 days of symptom onset. Notably, the grading of ICANS varied across all studies. In 2019 the American Society for Transplantation and Cellular Therapy proposed a new grading system to harmonize existing scoring systems.¹⁸ The use of this consensus grading scale would permit cross-trial comparisons of the safety profile of various CAR-T products and enable interventions to be developed to mitigate or manage these neurotoxicities.

Across studies that evaluated neuroimaging in cases with clinically confirmed ICANS, neuroimaging was most often normal. However, neuroimaging could be considered when excluding other conditions which may mimic the neurological symptoms of ICANS. EEG was most often abnormal when assessed in those with clinically confirmed ICANS. Despite its utility, EEG testing may be unavailable in some clinical settings. Several pro-inflammatory markers have been associated with ICANS development, including Interleukin (IL) 2, IL-6 and IL-15, interferon-gamma (IFN-gamma), and tumor necrosis factor-alpha (TNF-alpha), but the routine use of these markers in clinical practice is limited. Alternatives include evaluating the trends and peak levels of c-reactive protein and ferritin, which can be used to predict the risk of ICANS development.

Overall, only nine of the 23 included studies reported on the management of ICANS.^{8–10,21,22,27,31,34,36} Treatment most often involved supportive care alone in mild cases. However, most required glucocorticoid treatment and anti-seizure prophylaxis. Among the studies reporting coadministration of tocilizumab and steroids for ICANS, these patients also had concomitant CRS. Based on prior studies, using tocilizumab to manage ICANS in the absence of concurrent CRS is not recommended due to its poor blood-brain barrier penetration and concerns around worsening ICANS related to tocilizumab’s ability to cause elevations in CSF IL-6 levels.^9,37–39

Few studies reported in our systematic review evaluated long-term neurologic outcomes following CAR-T cell therapies. However, two reports from a single institution suggest that neuropsychiatric difficulties could persist beyond one year following CAR-T therapy. In one of these reports of long-term neurological complications among CAR-T cell therapy recipients, patients also reported clinically meaningful depression and anxiety symptoms.¹³ In the other published report, over 37% of patients reported at least one cognitive difficulty, predominantly difficulties with memory.¹⁴ These findings suggest a need for further research to characterize long-term patient-centered outcomes, which would then lend to developing interventions to prevent or mitigate the impact of ICANS. Finally, five fatal ICANS events were reported across two studies. These fatal outcomes support the need for early identification and management of those with symptoms of ICANS to prevent rapid clinical deterioration and death. Furthermore, there is a need to develop risk prediction tools capable of identifying the subset of patients at risk for severe ICANS development and targeted interventions to mitigate these neurologic complications.

Strengths and Limitations

Our study has limitations. First, to maintain our focus on hematological malignancies in the adult population, we excluded studies that reported ICANS in those with solid tumors or studies conducted exclusively in pediatric populations. Second, because of the heterogeneity of studies, a meta-analysis could not be performed. Thirdly, some relevant literature may not have been retrieved due to changes in the terminology used to define ICANS, previously known as CAR-T cell-related neurotoxicity or encephalopathy and cytokine release encephalopathy syndrome (CRES). We attempted to address this by conducting citation chasing and citation chaining of the included studies. Finally, all included studies were observational, which introduces the risk of confounding. We attempted to address confounding effects by assessing and reporting the risk of bias.

Despite these limitations, our systematic review has the following strengths. First, to our knowledge, this is the first systematic review to synthesize the literature on the clinical, biological, and radiological findings associated with ICANS among adults with hematological malignancies treated with CAR-T cell therapy. Second, this review reports on the long-term neurological complications of CAR-T cell therapy, which has seldom been the focus of previously published reports. Third, our study enables the identification of gaps in the diagnosis, classification, and long-term follow-up of CAR-T cell therapy recipients. Finally, our findings highlight a need for future studies designed to predict more accurately those at greatest risk for developing high-grade ICANS and long-term neurologic complications of CAR-T cell therapy.

Emerging data

The 2021 annual meetings of the American Society of Clinical Oncology and the American Society of Hematology showcased research abstracts examining the clinical manifestations, predictors, biomarkers, and management of ICANS (Table 2). These studies demonstrated similar incidences and time to onset and resolution of ICANS as those identified in our systematic review. The emergence of serum neurofilament light chain levels as a possible biomarker of ICANS onset and severity is especially promising. In the future, this biomarker could be potentially applied to the clinical setting, thus aiding in the early identification of those at risk for ICANS development. Additionally, the EASIX score has emerged as a possible predictor of ICANS onset and severity in those receiving CAR-T cell therapy. The EASIX score consists of easily and routinely measured labs, allowing easier adoption into clinical practice. Two highlighted abstracts focused on the role of the Interleukin-1 antagonist anakinra in the prevention or management of ICANS, with results suggesting lower and less severe rates of ICANS in those treated with anakinra.^40,41 This preliminary data also suggests a dose-dependent response to anakinra among those with ICANS.⁴¹

Table 2.

Summary of select abstracts presented at the 2021 Annual Meetings of the American Society of Clinical Oncology and the American Society of Hematology

First Author	Title	Study Aim	Study design	Population	Sample Size	CAR-T target and product	Outcome(s)
Clinical manifestations of Immune effector cell-associated neurotoxicity syndrome (ICANS)
Einslee et al.42	Incidence, mitigation, and management of neurologic adverse events in patients with multiple myeloma (MM) treated with ciltacabtagene autoleucel (cilta-cel) in CARTITUDE-2	To describe the clinical features and management of ICANS	Phase II open label	Relapsed multiple myeloma	n=20	BCMA: Cilta-cel (JNJ-68284528)	ICANS developed in n=4 (20%); median time to onset of symptoms: 8 days (range: 7–11), median duration of ICANS: 2 days (range: 1–2). Management of ICANS included: levetiracetam and steroids; all 3 patients with ICANS had concurrent cytokine release syndrome (CRS) and all recovered. One patient developed isolated facial paralysis on Day 29 post infusion, that resolved 51 days post onset of symptoms and with dexamethasone therapy.
Biomarkers and predictors of ICANS development
Gauthier et al.43	CD19 CAR T-cell product type independently impacts CRS and ICANS severity in patients with aggressive NHL	To assess the impact of 3 CD19 CAR T-cell products (axicabtagene ciloleucel, tisagenlecleucel, and JCAR014) on CRS and ICANS severity	retrospective	Relapsed, refractory Non-Hodgkin’s lymphoma	n=136	CD19: Axicabtagene ciloleucel, Tisagenlecleucel BCMA: JCAR014	Factors associated with ICANS severity CAR T-cell product type (Tisagenlecleucel vs axicabatagene ciloleucel OR = 0.14, p <.001; JCAR014 vs axicabatagene ciloleucel, OR = 0.31, p = 0.009) Pre-lymphodepletion LDH (OR, 3.96 per log10 increase; p = 0.04) Age (OR per 10-year increase, 1.32; p =.06)
Butt et al.44	Pre-Infusion Neurofilament Light Chain (NfL) Levels Predict the Development of Immune Effector Cell-Associated Neurotoxicity Syndrome (ICANS) - a Multicenter Retrospective Study	To evaluate pre-infusion levels of plasma neurofilament light chain (NfL), a marker of neurodegeneration, as a predictive biomarker of ICANS	retrospective	not reported	not reported	not reported	n=30 (36%) with ASTCT Gr 1–4 ICANS, higher NfL were found in those who developed ICANS compared to those who did not ([87.6 v 29.4 pg/ml], p = 0.00004); NfL correlated with ICANS development (r = 0.74, p < 0.0001)
Korell et al.45	Easix Predicts Severe Cytokine Release Syndrome (CRS) and Immune Effector Cell-Associated Neuro-Toxicity Syndrome (ICANS) in Patients Receiving CD19-Directed Chimeric Antigen Receptor T (CAR-T) Cell Therapy	To evaluate and validate the Endothelial Activation and Stress Index (EASIX)) as a predictor for CRS and ICANS in patients receiving CD19-directed CAR-T cells	retrospective	Relapsed, refractory large B-cell lymphoma Acute lymphoblastic leukemia Mantle cell lymphoma Chronic lymphocytic leukemia Follicular lymphoma	n=107 (training cohort) n=93 (validation cohort)	CD19: Axicabtagene ciloleucel, Tisagenlecleucel	n=24 patients (22%) developed ICANS grades 1–4. Grade ≥ 3 ICANS occurred in 11 patients (11%; median onset 8 (4–17) days). EASIX values (pre-lymphodepletion, d0, d3, d7) were significantly higher in patients who developed either grade 3–4 CRS, ICANS, or both; EASIX predicted grade 3–4 CRS and ICANS before lymphodepleting therapy (-pre), on day 0 and on day 3 in both cohorts
Management of ICANS
Gazeau et al.41	Safety and Efficacy of Two Anakinra Dose Regimens for Refractory CRS or ICANS after CAR T-Cell Therapy	To describe the safety and efficacy of two anakinra dose regimens to treat refractory CRS and/or ICANS after CAR T-cell therapy	retrospective	Multiple myeloma Mantle cell lymphoma Transformed follicular lymphoma Diffuse large B-cell lymphoma Steroid-refractory ICANS	n=26	not reported	Intervention: Anakinra 100–200mg/day subcutaneously (SC) in 13 patients (pts) or 8mg/kg/day SC or intravenously (IV) in 13 patients. Treatment with anakinra led to improved ICANS/CRS rates in 73% of patients with higher response in those who received the higher dose of anakinra
Frigault et al.40	A Phase II Trial of Anakinra for the Prevention of CAR-T Cell Mediated Neurotoxic	To evaluate whether or not anakinra could be administered prophylactically to prevent severe CRS and neurologic events (NE) in patients receiving axicabtagene ciloleucel	Phase II, open-labelled	Relapsed/refractory large cell lymphoma	N=6	CD19: Axicabtagene ciloleucel	Intervention: Anakinra 200 mg SC starting 4 hours prior to axicabatagene ciloleucel infusion and daily thereafter for a total of 7 days. 2 of 6 patients experienced grade 3 ICANS

Conclusion

Our findings highlight the variability in the biological, clinical, radiological, and EEG findings of ICANS following CAR-T cell therapy and the need for further work to characterize the long-term neurologic outcomes of these patients. As CAR-T cell therapies continue to gain popularity in clinical practice, there is a need to bridge the evidence gaps. Future research studies should focus on early identification and risk prediction modeling for ICANS development. The use of consensus grading scales of ICANS in future studies would provide cross-trial comparisons critical for developing interventions to reduce and manage neurotoxicities related to CAR-T therapies. Additional studies are needed to characterize the long-term neurologic complications following CAR T-cell therapy fully. Studies ought to leverage real-world populations when examining the effects of CAR-T cell therapies.

Suggested future research directions for ICANS related studies:
Utilize consensus grading ICANS scales across studies
Develop and validate risk prediction models for the identification of those most likely to develop high-grade ICANS related complications
Conduct longitudinal studies to assess the long-term outcomes following CAR-T cell therapy with a focus on late-term neurotoxic complications
Evaluate the effects of CAR-T cell therapy in real-world populations and more vulnerable populations such as older adults
Develop interventions to support, manage or prevent the onset of ICANS

Key Points.

1. ICANS typically develops within the first 7–10 days following CAR-T cell infusion and often develops in the setting of cytokine release syndrome

2. Clinical features of ICANS are highly variable but commonly include aphasia, seizures, headache, encephalopathy, and altered level of consciousness

3. Most symptoms of ICANS resolve within 21 days of onset and can be managed with supportive care, steroids, and anti-seizure prophylaxis; tocilizumab (a monoclonal antibody to interleukin-6 receptor) should be reserved for those who have concurrent cytokine release syndrome

4. We systematically demonstrate the association between ICANS and the following: presence and severity of cytokine release syndrome (CRS), c-reactive protein (CRP), and ferritin levels, and the diagnostic utility of magnetic resonance imaging (MRI), computerized tomography (CT) and electroencephalogram (EEG)