Immune Effector Cell-Associated Neurotoxicity Syndrome

Immune effector cell-associated neurotoxicity syndrome (ICANS) is a distinct, potentially life-threatening neurotoxicity unique to immune effector cell therapies, chimeric antigen receptor (CAR) T cells, bispecific antibodies, and T-cell receptor (TCR) T cells, characterized by encephalopathy, focal neurologic deficits, and seizures that typically emerge days after infusion ^[2]D5.

Also Called / Synonyms

• CAR-T-related encephalopathy syndrome (CRES), historical term
• Immune effector cell-associated encephalopathy
• Neurotoxicity after cellular immunotherapy

Classification ICANS is graded by the American Society for Transplantation and Cellular Therapy (ASTCT) consensus criteria, which integrate the 10-point ICE score (orientation, naming, following commands, writing, attention) with level of consciousness, seizure burden, motor findings, and raised intracranial pressure ^[3]B3b. The same framework applies across CAR-T products (axicabtagene ciloleucel, tisagenlecleucel, lisocabtagene maraleucel, brexucabtagene autoleucel) and bispecific engagers such as tarlatamab ^[13]D5.

Clinical Significance Grade ≥3 ICANS independently increases non-relapse mortality (HR 2.46, 95% CI 1.00-6.04) and worsens overall survival (HR 1.79, 95% CI 1.14-2.81) after CD19 CAR-T therapy, whereas CRS and early hematologic toxicity do not retain independent prognostic significance ^[3]B3b.

Pearl: ICANS is not simply "CRS of the brain", it has distinct pathophysiology, grading, and prognostic weight; a baseline plasma neurofilament light chain >87 pg/mL predicts development with 96% AUC ^[6]B3b.

Core Pathogenic Cascade

ICANS arises from a sequence that begins with CAR T-cell activation and culminates in blood-brain barrier (BBB) disruption, neuroinflammation, and neuronal injury. The process is distinct from cytokine release syndrome (CRS), although the two syndromes frequently overlap temporally.

1. CAR T-cell activation and cytokine surge, Following infusion, CAR T cells engage target antigen (CD19, BCMA, or others), proliferate, and release a burst of proinflammatory cytokines including IFN-γ, IL-6, IL-2, IL-8, IL-10, IL-15, and TNF-α ^[15]C4. Peak CAR T-cell expansion (Cmax) and early exposure (AUC₀-₂₈) correlate with higher-grade CRS, and by extension with the cytokine milieu that primes the neurovascular unit ^[15]C4.
2. Endothelial activation and BBB breakdown, Cytokines (particularly IFN-γ and TNF-α) activate cerebral endothelial cells, upregulating adhesion molecules (ICAM-1, VCAM-1) and increasing vascular permeability. Serum biomarkers of endothelial stress (mEASIX, lactate dehydrogenase) rise in parallel ^[27]B3b.
3. CAR⁺ extracellular vesicle (CAR⁺EV) trafficking, Activated CAR T cells shed CAR⁺EVs within 1 hour of target engagement. Plasma CAR⁺EV concentrations >132.8/µL at hour +1 or >224.5/µL at day +1 predict ICANS with high sensitivity and specificity, preceding clinical onset by ~4 days ^[22]B2b. In vitro, CAR⁺EVs induce release of neuronal injury markers (ENO2⁺ nanoparticles) from iPSC-derived neural cells, implicating them as direct mediators of neurotoxicity ^[22]B2b.
4. Neuroinflammation and microglial activation, Cytokines and CAR⁺EVs penetrate the compromised BBB, activating microglia and astrocytes. This triggers a secondary intrathecal cytokine cascade (IL-1β, IL-6, CXCL10, CCL2) that amplifies local inflammation ^[16]D5.
5. Neuronal and astroglial injury, Serum neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP), markers of axonal and astrocytic damage, are elevated at baseline in patients who later develop ICANS, indicating pre-existing neuronal susceptibility ^[27]B3b. Day-7 NfL and GFAP elevations correlate with coagulopathy (prolonged clotting time, DIC markers) and systemic inflammation (CRP), linking neurotoxicity to endothelial stress ^[27]B3b.

Distinct Phenotypic Variants

Central (thalamic/hippocampal) variant, Symmetric involvement of thalami, hippocampi, brainstem, and variably the geniculate bodies or corpus callosum on MRI. Associated with severe ICANS, acute clinical deterioration, and diffuse centrum semiovale edema with intracranial hypertension . This pattern largely resolves on follow-up imaging ^[21]C4.

Stroke-like (cortico-subcortical) variant, Focal DWI-restricted cortico-subcortical white matter lesions appearing in subacute phases, often in both ICANS and asymptomatic patients. Lesions persist as non-enhancing FLAIR abnormalities ^[21]C4.

Acute myelopathy, Rare but distinct syndrome of quadriparesis/paraparesis with spinal cord and/or brain white matter abnormalities. Autopsy and CSF studies show chronic white matter destruction without inflammatory infiltrates or elevated proinflammatory cytokines (IFN-γ, CCL17, CCL23, CXCL10 are lower than in typical ICANS), suggesting a non-inflammatory mechanism ^[14]C4[17]D5.

Modulating Factors

• CAR design, CD28-costimulated CARs historically associate with higher ICANS rates than 4-1BB designs. The calibrated-signaling 1XX CAR (inactivating distal ITAMs) retains CD28 effector function while reducing severe ICANS to 7% (grade ≥3) across doses up to 200×10⁶ cells ^[15]C4.
• Antigen target, BCMA-directed CAR T therapy for multiple myeloma shows no ICANS in several small series ^{[23]B2b[24]B2b[30]C4}, whereas CD19-directed products carry 30-40% any-grade ICANS incidence ^[21]C4.
• Baseline neuronal vulnerability, Pre-infusion serum GFAP and NfL levels independently predict ICANS risk and steroid requirement ^[27]B3b.
• Clonal hematopoiesis (CHIP), Present in 34% of pre-CAR T patients, but does not significantly alter CRS/ICANS incidence or severity ^[18]D5.

Mermaid Flowchart

mermaid flowchart TD A[CAR T-cell infusion] --> B[Antigen engagement & T-cell activation] B --> C[Cytokine surge (IFN-γ, IL-6, IL-2, IL-8, IL-10, IL-15, TNF-α)] C --> D[Endothelial activation & BBB disruption] B --> E[CAR⁺EV release (within 1 hr)] E --> F[CAR⁺EVs cross compromised BBB] D --> F F --> G[Microglial/astrocyte activation] G --> H[Intrathecal cytokine cascade (IL-1β, IL-6, CXCL10, CCL2)] H --> I[Neuronal & astroglial injury (↑NfL, ↑GFAP)] I --> J[Clinical ICANS phenotypes] J --> K1[Central variant: thalamic/hippocampal edema] J --> K2[Stroke-like variant: cortico-subcortical lesions] J --> K3[Acute myelopathy: non-inflammatory white matter injury]

Pearl: CAR⁺EVs appear within 1 hour of infusion and predict ICANS days before symptoms; measuring them at hour +1 (>132.8/µL) or day +1 (>224.5/µL) offers a actionable early-warning window ^[22]B2b.

Building on the pathophysiological framework of endothelial activation and cytokine-mediated blood-brain barrier disruption, the epidemiology of ICANS is defined by product type, disease context, and host factors that modulate peak cytokine exposure and CAR-T expansion. Incidence varies widely across approved constructs and indications, reflecting differences in CAR affinity, costimulatory domain, and target antigen density.

Incidence by Product and Indication

• Axicabtagene ciloleucel (axi-cel) for large B-cell lymphoma: ICANS (any grade) in 55% of 280 patients ^[43]C4; grade ≥3 ICANS in 7.1% of 127 obe-cel recipients ^[47]C4 and 3% of var-cel recipients ^[34]B2b.
• Tisagenlecleucel (tisa-cel) and lisocabtagene maraleucel (liso-cel): lower ICANS rates than axi-cel; in a real-world cohort of 624 LBCL patients, axi-cel had significantly higher ICANS (G1+ aOR 2.10, P=0.048) than liso-cel ^[50]B3b.
• Brexucabtagene autoleucel (brexu-cel) for mantle cell lymphoma : ICANS odds ratio 1.66 (95% CI 1.06-2.60) with prior autologous HCT ^[41]B3b.
• BCMA-directed CAR-T (ide-cel, cilta-cel) for multiple myeloma : classic ICANS rates similar between products, but delayed neurotoxicity markedly higher with cilta-cel (OR 20.07, 95% CI 4.46-90.20) ^[38]C4. Non-ICANS neurologic toxicities (NINTs) pooled incidence 0.81% (95% CI 0.37-1.77%), driven by cilta-cel (4.6% vs 0.5% for ide-cel) ^[35]B2a.
• Bispecific antibodies (epcoritamab, blinatumomab): ICANS 6% (grade 1-2) with epcoritamab in follicular lymphoma ^[45]C4; 17% with subcutaneous blinatumomab in B-ALL ^[46]C4.
• Novel constructs: prizlon-cel (CD19/CD20) ICANS 6.3% (no grade ≥3) ^[44]C4; obe-cel (intermediate-affinity CAR) grade ≥3 ICANS 7.1% ^[47]C4.

Host and Disease Risk Factors

Temporal Patterns

• New-onset CRS and ICANS are exceedingly rare after 2 weeks post-infusion (0% and 0.7%, respectively) ^[42]B3b.
• Median onset of in-hospital CV events (often concurrent with ICANS) day 8 (IQR 5-11) ^[51]B3b.

Etiologic Considerations

• HHV-6B reactivation 6% (95% CI 2.2-12.5%) after CAR-T; encephalitis 0.17% (95% CI 0.02-0.94%), infrequent and not a major ICANS driver ^[37]C4.
• Non-inflammatory pathophysiology suggested for quadriparesis/paraparesis phenotype in children (low CSF proinflammatory cytokines) ^[14]C4.

Pearl: ICANS incidence is product-dependent, highest with axi-cel (∼55% any grade) and cilta-cel (delayed neurotoxicity OR 20 vs ide-cel), and new-onset events after day 14 are vanishingly rare, supporting focused early monitoring.

Presenting Symptoms

ICANS typically emerges within the first 7-10 days after CAR T-cell infusion, with a median onset of 3.5 days in CNS lymphoma (TIAN) and up to 4 weeks for classic ICANS ^[56]D5[55]C4. The core clinical syndrome is an encephalopathy that progresses from mild confusion to obtundation, often accompanied by language dysfunction (aphasia, dysarthria), tremor, headache, and seizures ^{[57]A1a[66]C4}. In pediatric and young adult cohorts, encephalopathy was present in 88% of ICANS cases, seizures in 31%, and focal motor deficits in 7% ^[66]C4. Non-canonical symptoms, headache (33%), vision changes (14%), tremor (14%), are common but do not correlate with specific imaging findings ^[66]C4.

Neurological Examination Findings

Phenotypic Variants

Red Flags Requiring Urgent Action

• Depressed level of consciousness ( GCS ≤ 8) → airway protection, ICU transfer
• Seizures / status epilepticus → emergent EEG, antiseizure medication
• Focal motor deficits progressing to quadriparesis → urgent MRI brain/spine
• Autonomic instability (labile BP, arrhythmia) → continuous monitoring
• FVC < 15 mL/kg or rising pCO₂ → consider intubation

Atypical Presentations

• Isolated cranial neuropathies without encephalopathy (cranial nerve palsy syndrome) ^[53]A1b
• Pure motor syndromes (quadriparesis) with minimal encephalopathy ^[14]C4
• Delayed onset > 4 weeks (late encephalitis, GBS) ^[55]C4[25]C4
• TIAN mimicking ICANS but tethered to pre-existing CNS tumor volume > 3.4 cm³ ^[56]D5
• Neuropsychiatric features (hallucinations, agitation) without focal signs ^[57]A1a

Neuroimaging Correlates

In children and young adults, 36% of ICANS patients had MRI abnormalities, predominantly symmetric T2/FLAIR hyperintensities in the external/extreme capsules (69%), brainstem (central tegmental tracts, inferior olivary nuclei; 40%), and bilateral thalami (26%) ^[66]C4. Higher ICANS grade strongly predicted imaging abnormalities (aOR 3.7, 95% CI 2.0-6.8) ^[66]C4. The "double-smiley sign" (bilateral external capsule + thalamic hyperintensity) and holocord transverse myelitis extending from brainstem are characteristic ^[66]C4. Restricted diffusion in splenium or thalami occurs in a minority. Most abnormalities improve or resolve on follow-up (83% improved, 25% fully resolved) ^[66]C4.

Pearl: ICANS is a clinical diagnosis, neuroimaging supports but does not define it. A normal MRI does not exclude ICANS, especially in low-grade cases where imaging abnormalities are rare (9% in grade 1-2) ^[66]C4.

The clinical presentation of ICANS, encephalopathy, seizures, motor deficits, and cerebral edema, demands a structured diagnostic approach that distinguishes immune effector cell neurotoxicity from infection, metabolic derangement, intracranial hemorrhage, and disease progression. The workup centers on three pillars: the ICE score for bedside quantification, neuroimaging to exclude structural lesions and characterize edema, and cerebrospinal fluid analysis to assess inflammatory burden and rule out alternative etiologies.

ICE Score: Bedside Quantification

The Immune Effector Cell Encephalopathy (ICE) score is the validated, ASTCT-endorsed tool for grading ICANS severity. It evaluates four domains, orientation, naming, following commands, and writing, each scored 0-3, yielding a total of 0-10. A score of 10 indicates normal cognition; 7-9 corresponds to grade 1 ICANS; 3-6 to grade 2; 1-2 to grade 3; and 0 to grade 4. The ICE score is performed at least every 8 hours during the high-risk window (days 0-14 post-infusion) and more frequently if clinical change occurs. Serial scoring tracks trajectory and triggers intervention thresholds: grade ≥2 warrants corticosteroids; grade ≥3 mandates neurology consultation, EEG, and MRI. The ICE score correlates with overall ICANS grade and predicts need for intensive supportive care.

Neuroimaging

MRI brain with and without contrast is the imaging modality of choice. It is indicated for new-onset seizures, focal deficits, ICE score decline ≥2 points, or grade ≥3 ICANS. Characteristic findings include T2/FLAIR hyperintensities in the frontal, parietal, and occipital white matter, often symmetric, consistent with vasogenic edema. Diffusion-weighted imaging may show restricted diffusion in severe cases, heralding cytotoxic injury. Contrast enhancement is typically absent, helping distinguish ICANS from leptomeningeal disease or infection. In the pediatric series of quadriparesis/paraparesis, MRI revealed longitudinal extensive transverse myelitis -like spinal cord lesions and supratentorial white matter changes without enhancement, supporting a non-inflammatory mechanism. CT head without contrast is reserved for rapid exclusion of hemorrhage or mass effect when MRI is not immediately available.

Cerebrospinal Fluid Analysis

Lumbar puncture is performed when ICANS grade ≥2, especially with fever, seizures, or meningeal signs, to exclude meningitis, encephalitis, and leptomeningeal malignancy. Typical ICANS CSF shows mild lymphocytic pleocytosis (usually <50 cells/µL), elevated protein (50-150 mg/dL), and normal glucose. CSF cytokine profiling in research settings reveals marked elevation of IL-6, IL-8, IFN-γ, CCL2, CXCL10, and TNF-α, but this is not standard clinical practice. Notably, children with post-CAR-T quadriparesis had lower CSF levels of IFN-γ, CCL17, CCL23, and CXCL10 than those without motor deficits, suggesting a distinct, non-inflammatory pathophysiology for this phenotype. CSF flow cytometry and PCR for common pathogens (HSV, VZV, enterovirus, HHV-6, bacteria, fungi) are sent routinely. Malignant cells are sought by flow cytometry and cytology.

Electroencephalography

Continuous EEG (cEEG) is recommended for grade ≥3 ICANS, any clinical seizure, or unexplained encephalopathy. Non-convulsive status epilepticus (NCSE) occurs in up to 30% of severe ICANS cases and is detectable only on EEG. Typical patterns include generalized slowing, frontal intermittent rhythmic delta activity (FIRDA), and electrographic seizures. EEG guides antiseizure medication escalation and helps differentiate encephalopathy from NCSE. Serial EEGs monitor resolution.

Peripheral Blood Biomarkers

While not diagnostic, serial measurement of CRP, ferritin, IL-6, and CAR-T expansion kinetics (by flow cytometry or qPCR) correlates with ICANS risk and severity. Peak CAR-T expansion (AUC) associates with higher ICANS grades. Pre-lymphodepletion LDH and CRP predict both expansion and toxicity. These biomarkers support risk stratification but do not replace clinical assessment.

Diagnostic Algorithm Summary

1. Baseline: Pre-infusion neurologic exam, ICE score, CBC, CMP, CRP, ferritin, LDH, coagulation studies.
2. Monitoring: ICE score q8h × 14 days; vital signs q4h; neurologic checks q4h.
3. Trigger for workup: ICE <10, new neurologic symptom, seizure, or altered mental status.
4. Immediate: STAT ICE score, glucose, electrolytes, renal/hepatic function, CBC, coagulation, CRP, ferritin, blood cultures.
5. Imaging: MRI brain ± spine within 4 hours if grade ≥2 ICANS with focal signs, seizure, or ICE decline ≥2.
6. LP: If grade ≥2 ICANS with fever, meningeal signs, or diagnostic uncertainty, after imaging excludes mass effect.
7. EEG: cEEG within 1 hour for grade ≥3 ICANS or clinical seizure; routine EEG within 12 hours for grade 2.
8. CAR-T monitoring: Peripheral blood CAR-T quantification (flow cytometry) on days 7, 14, 21, 28 if available.

Pearl: The ICE score is the linchpin of ICANS diagnosis, serial bedside assessment drives every downstream decision (imaging, LP, EEG, steroids). A 2-point drop from baseline mandates MRI within 4 hours; grade ≥3 mandates cEEG within 1 hour. Do not wait for "full" ICANS to act, the ICE score catches it early.

Building on the diagnostic workup, risk stratification in ICANS relies on validated grading systems and emerging biomarker models that predict severity and guide preemptive intervention. The ICE (Immune Effector Cell Encephalopathy) score remains the bedside standard for daily grading, while the ASTCT consensus criteria provide a unified framework for CRS and ICANS severity. Beyond clinical scores, baseline host factors, including inflammatory markers, metabolic profile, and body composition, identify patients at highest risk for early and severe neurotoxicity.

ICE Score and ASTCT Grading

The ICE score quantifies encephalopathy across four domains: orientation (0-3), naming (0-1), following commands (0-1), and writing (0-1), yielding a total of 0-10. A score of 10 indicates normal cognition; 7-9 corresponds to ASTCT grade 1 ICANS; 3-6 to grade 2; 1-2 to grade 3; and 0 to grade 4 ^[1]C4. Serial ICE assessments every 8-12 hours during the high-risk window (days 0-14 post-infusion) detect deterioration before overt neurologic decline. The ASTCT consensus further incorporates seizure, motor weakness, raised intracranial pressure, and cerebral edema into grade 3-4 definitions, triggering ICU transfer and aggressive immunosuppression ^[1]C4.

CAR-HEMATOTOX and Hematologic Reserve

Although developed for hematologic toxicity, the CAR-HEMATOTOX model incorporates baseline platelet count, hemoglobin, ANC, CRP, and ferritin, variables that also correlate with ICANS risk through shared inflammatory pathways ^[36]B3b. In the validation cohorts (n=200), a high CAR-HEMATOTOX score (≥2) predicted prolonged severe neutropenia (≥14 days) with 89% sensitivity and 68% specificity (AUC 0.89) ^[36]B3b. Patients with high scores had longer neutropenia (12 vs 5.5 days), higher severe thrombocytopenia (87% vs 34%), and anemia (96% vs 40%) ^[36]B3b. These cytopenias reflect marrow reserve depletion that parallels neuroinflammatory vulnerability.

Metabolic and Body Composition Biomarkers

Pretreatment metabolomics identifies a signature of low hydroxyproline, low glutamine, low proline, low glycine, and low isoursodeoxycholate associated with higher-grade and earlier-onset ICANS ^[87]D5. Higher baseline glucose and lower cholesterol/glutamate predict faster CRS onset, which often precedes ICANS ^[87]D5. Visceral adiposity independently drives severe CRS: patients with grade ≥2 CRS had higher median BMI, waist circumference, waist-to-height ratio, and visceral adipose tissue (VAT); these parameters also correlated with earlier CRS onset and higher peak IL-6 ^[88]D5. In multivariate logistic regression, BMI, waist circumference, WtHR, and VAT each increased the probability of grade ≥2 CRS ^[88]D5. ROC-derived thresholds for VAT and WtHR discriminated patients with markedly elevated peak IL-6 ^[88]D5. Body composition did not influence ICANS severity or onset directly ^[88]D5, but by amplifying CRS it indirectly elevates neurotoxicity risk.

ctDNA and Tumor Burden

Circulating tumor DNA (ctDNA) kinetics provide dynamic risk stratification. In a prospective multicenter trial of axi-cel for large B-cell lymphoma (n=72), higher pretreatment ctDNA concentrations correlated with both CRS and ICANS development ^[85]D5. At day 28, detectable ctDNA conferred a median PFS of 3 months versus not reached and median OS of 19 months versus not reached ^[85]D5. Among patients with partial response or stable disease on imaging, 15 of 17 (88%) with detectable ctDNA relapsed versus 1 of 10 (10%) with undetectable ctDNA ^[85]D5. ctDNA was detected at or before radiographic relapse in 29 of 30 (97%) patients ^[85]D5. All durably responding patients achieved undetectable ctDNA by 3 months ^[85]D5.

Clinical Prognostic Factors for CAR-T Outcomes

A systematic review of 79 studies in DLBCL identified ECOG PS, IPI score, disease histology, Ann Arbor stage, and elevated LDH as consistent predictors of efficacy and safety outcomes including ICANS ^[92]B2a. Real-world meta-analysis of axi-cel vs tisagenlecleucel confirmed that axi-cel carries higher grade ≥3 ICANS risk (OR 3.95, 95% CI 3.05-5.11) despite superior OS (HR 0.60) and PFS ^[93]C4. TP53 mutation status, while prognostic in chemotherapy-treated LBCL, does not independently worsen ICANS incidence or severity in CAR-T recipients ^[26]B3b.

Bispecific Antibody Context

For bispecific antibodies (mosunetuzumab, epcoritamab), ICANS incidence is lower than with CAR-T but remains clinically significant. In a phase 2 mosunetuzumab-CHOP study (n=40), 5 patients (12.5%) experienced neurologic events potentially consistent with ICANS (max grade 3) ^[83]B2b. Real-world epcoritamab data (n=157) showed any-grade ICANS in 12% (grade ≥3: 0.6%) ^[86]D5. Multivariable analysis identified ECOG PS >2, elevated LDH, refractory disease, and clinical trial ineligibility as independent predictors of inferior PFS and OS ^[86]D5.

Prognostic Factor Summary

Pearl: A high CAR-HEMATOTOX score (≥2) or detectable day-28 ctDNA identifies patients who need intensified neurologic monitoring and early intervention; both are calculable before or early in the CAR-T course and outperform single clinical variables ^{[36]B3b[85]D5}.

Step 1: Initial Assessment and Severity Classification

Patients with suspected ICANS require immediate neurologic evaluation using the ICE score (10-point scale assessing orientation, naming, following commands, writing, and attention). Grade 1 (ICE 9-10) warrants supportive care and close monitoring. Grade 2 (ICE 7-8) or higher mandates urgent neurology consultation, continuous cardiac monitoring, and preparation for pharmacologic intervention ^[99]B2a. Obtain STAT non-contrast head CT to exclude hemorrhage or edema; MRI brain with diffusion-weighted imaging is preferred if CT is negative and clinical suspicion persists. Lumbar puncture with opening pressure, cell count, protein, glucose, and cytokine panel (IL-6, IFN-γ) is indicated when infection cannot be excluded. Continuous EEG is recommended for grade ≥2 ICANS to detect non-convulsive status epilepticus, which occurs in up to 40% of severe cases ^[109]D5.

Step 2: First-Line Pharmacologic Intervention

Corticosteroids are the management of choice for moderate-to-severe ICANS (grade ≥2). The ASCO guideline recommends dexamethasone 10 mg IV every 6 hours for grade 2-3 ICANS and methylprednisolone 1 g IV daily for grade 4 ICANS ^[99]B2a. Initiate steroids immediately upon recognition of grade ≥2 ICANS; do not wait for imaging or LP results. Tocilizumab 8 mg/kg IV (max 800 mg) over 1 hour is indicated for concurrent CRS but has limited CNS penetration and does not reverse established neurotoxicity ^{[99]B2a[109]D5}. Administer tocilizumab only if CRS grade ≥2 coexists; avoid routine use for isolated ICANS.

Step 3: Escalation for Steroid-Refractory ICANS

If no improvement within 24 hours of high-dose steroids, escalate to anakinra 100 mg IV every 6 hours (or 2 mg/kg IV q6h) ^{[107]C4[111]D5}. Anakinra crosses the blood-brain barrier and targets IL-1-mediated neuroinflammation. Case reports describe resolution of grade 4 ICANS within 48 hours of anakinra initiation ^[107]C4. For life-threatening cerebral edema (grade 4 with radiographic evidence), consider extracorporeal cytokine adsorption (CytoSorb) as rescue therapy; a single case reported complete neurologic recovery after 4 days of CytoSorb with >95% IL-6 reduction ^[107]C4. Siltuximab (anti-IL-6) 11 mg/kg IV may be considered if anakinra fails, though evidence is limited to case series ^[111]D5.

Step 4: Monitoring and Titration

• Neurologic checks: q1h for grade ≥2 until ICE score stabilizes ≥9, then q4h × 24h.
• ICE score: document every shift; improvement of ≥2 points predicts recovery.
• EEG: continuous for grade ≥3; repeat at 24h if non-convulsive seizures detected.
• Inflammatory markers: CRP, ferritin, IL-6 q12h while on steroids; trend guides taper.
• Steroid taper: once ICE score ≥9 for 24h, reduce dexamethasone by 25% daily; total course typically 7-10 days ^[99]B2a.
• Seizure prophylaxis: levetiracetam 500 mg IV q12h for grade ≥2; continue 30 days post-resolution if EEG abnormal.

Step 5: Resolution and Disposition

ICANS resolution defined as ICE score 10 sustained ≥48h off steroids. Median time to resolution is 4 days (IQR 2-8) across products ^[40]B3b. Patients with grade ≥3 ICANS require ICU-level care until neurologic baseline restored. Outpatient follow-up within 2 weeks for neurocognitive screening (MoCA) and biomarker surveillance (NfL, GFAP) ^[27]B3b.

Drug Comparison Table

Dosing Table

Treatment Failure Protocol

1. No ICE improvement at 24h on high-dose steroids → add anakinra 100 mg IV q6h.
2. No improvement at 48h on anakinra + steroids → consider CytoSorb extracorporeal adsorption (if available) or siltuximab 11 mg/kg IV.
3. Progressive cerebral edema on imaging → neurosurgical consultation for ICP monitoring; consider hyperosmolar therapy (mannitol 0.5-1 g/kg or 3% saline) while arranging CytoSorb.

What NOT to Do

• Do NOT use tocilizumab as monotherapy for isolated ICANS without concurrent CRS, limited BBB penetration, no proven neurotoxicity benefit ^{[99]B2a[109]D5}.
• Do NOT delay steroids awaiting MRI/LP, every hour of delay in grade ≥2 ICANS correlates with worse outcomes ^[99]B2a.
• Do NOT taper steroids before ICE score ≥9 sustained 24h, premature taper associates with rebound neurotoxicity ^[99]B2a.
• Do NOT use prophylactic antiepileptics in grade 1 ICANS, no evidence of benefit, adds sedation burden ^[109]D5.

Controversies and Guideline Disagreement

Pearl: Initiate dexamethasone 10 mg IV q6h immediately for any grade ≥2 ICANS, do not wait for imaging or tocilizumab. Reserve tocilizumab for concurrent CRS ≥ grade 2. Escalate to anakinra at 24h if ICE score fails to improve by ≥2 points ^{[99]B2a[107]C4}.

Step 1: Assess for Persistent or Delayed Neurotoxicity

After acute ICANS resolves, screen for delayed neurotoxicity, movement disorders, cranial nerve palsies, parkinsonism , and neurocognitive decline, which may emerge weeks to months post-infusion. In the CARTITUDE-4 trial, 4.5% of cilta-cel recipients developed ICANS (all grade 1-2), 9.1% had cranial nerve palsy (grade 2, 8.0%; grade 3, 1.1%), and 2.8% had CAR-T-related peripheral neuropathy (grade 1-2, 2.3%; grade 3, 0.6%) ^[53]A1b. A real-world comparison of ide-cel and cilta-cel found delayed neurotoxicity (events not captured as ICANS) was significantly more frequent with cilta-cel (OR 20.07, 95% CI 4.46-90.20) ^[38]C4. Obtain a baseline neurocognitive assessment (MoCA or equivalent) at day 30 and repeat at 3, 6, and 12 months.

Step 2: Corticosteroid Tapering and Immunosuppression Weaning

For patients who required corticosteroids for grade ≥2 ICANS, taper over 4-6 weeks once ICE score returns to 10 and neuroimaging is stable. The ASCO guideline recommends corticosteroids for moderate-to-severe ICANS but does not specify a taper schedule; practice patterns favor a slow taper to avoid rebound neuroinflammation ^[99]B2a. Monitor for adrenal insufficiency if cumulative prednisone -equivalent dose exceeds 20 mg/day for >3 weeks.

Step 3: Refractory or Relapsed ICANS, Escalation Options

If ICANS recurs or persists despite corticosteroids:

• Anakinra 100 mg SC daily (2 mg/kg in pediatrics) until ICE score ≤1 for 48 hours. In a real-world liso-cel cohort, 9 patients received anakinra for grade ≥3 ICANS ^[126]B3b.
• Siltuximab 11 mg/kg IV every 3 weeks for corticosteroid-refractory ICANS. A multicenter retrospective analysis reported ICANS grade improvement in 60% of 17 steroid-refractory patients ^[122]C4.
• Tocilizumab 8 mg/kg IV (max 800 mg) for concurrent CRS; limited utility for isolated ICANS ^[99]B2a.

Step 4: Infection Prophylaxis and Immunoglobulin Replacement

Prolonged cytopenias and B-cell aplasia mandate:

• IVIG 400 mg/kg monthly while IgG <400 mg/dL or recurrent infections.
• Antimicrobial prophylaxis: acyclovir / valacyclovir , fluconazole /posaconazole, and TMP-SMX (or atovaquone) for PJP prophylaxis until CD4+ >200/µL and off corticosteroids ≥1 month ^[117]A1b.
• G-CSF for neutropenia <500/µL persisting >14 days.

Step 5: Surveillance for Late Effects and Second Malignancies

• Neurocognitive testing every 6 months for 2 years, then annually.
• MRI brain if new focal deficits, seizures, or progressive cognitive decline.
• Second primary malignancy (SPM) screening: CBC with differential every 3 months for 3 years, then every 6 months; skin exam annually. Real-world data show SPM incidence of 4% (1% myeloid) after ide-cel ^[69]B2b and non-significantly higher odds with cilta-cel (OR 1.77, 95% CI 0.89-3.56) ^[38]C4.
• Cardiac monitoring: echocardiogram at 6 and 12 months if anthracycline exposure or prior radiation.

Step 6: Consolidation and Subsequent Therapy

For patients in complete response, no consolidation is standard. For those with residual disease or early relapse, options include:

• Allogeneic HCT, considered in fit patients with high-risk features (see Section 9).
• Bispecific antibodies (teclistamab, talquetamab) or CAR-T re-treatment, sequencing data suggest BsAb before CAR-T improves subsequent CAR-T response (CR 53.7% vs 29.4%) ^[116]A1a.
• Clinical trial enrollment for novel targets (GPRC5D, CD19×CD20, etc.).

Drug Comparison Table

Dosing Table

Treatment Failure Protocol

1. ICANS recurrence after steroid taper → restart dexamethasone 10 mg IV q6h + anakinra 100 mg SC daily.
2. No improvement in 48 h → add siltuximab 11 mg/kg IV.
3. Persistent grade ≥3 ICANS >7 days → MRI brain with contrast, EEG, neurology consult; consider high-dose methylprednisolone 1 g IV daily ×3 days.
4. Refractory to all above → clinical trial or compassionate-use agents (e.g., JAK inhibitors, anti-IL-1, anti-GM-CSF).

What NOT to Do

• Do NOT use tocilizumab for isolated ICANS without concurrent CRS, no proven benefit and may worsen neurotoxicity by increasing IL-6 CNS penetration ^[99]B2a.
• Do NOT taper corticosteroids faster than 4 weeks after grade ≥2 ICANS, rebound neuroinflammation risk.
• Do NOT omit PJP prophylaxis while on corticosteroids ≥20 mg/day prednisone equivalent.

Controversies and Guideline Disagreement

Pearl: After acute ICANS resolves, the long-term game is surveillance, screen for delayed neurotoxicity (movement disorders, cranial neuropathies , cognitive decline) at 30 days, 3, 6, and 12 months; taper steroids over 4-6 weeks; maintain infection prophylaxis until immune reconstitution; and enroll relapsed patients in trials exploring BsAb→CAR-T sequencing, which doubles CR rates ^[116]A1a.

Step 1: Pre-Infusion Risk Stratification and Bridging Therapy

Patients proceeding to CAR-T or bispecific antibody therapy require systematic assessment of disease burden, organ function, and comorbidity profile. A normal lactate dehydrogenase (LDH) level before lymphodepletion independently predicts higher response rates, longer progression-free survival, and improved overall survival across multiple CAR-T products ^[126]B3b. Elevated LDH at leukapheresis identifies patients who benefit most from cytoreductive bridging.

Bridging therapy is standard during the manufacturing interval. Radiation to all metabolically active sites (comprehensive bridging radiotherapy) associates with superior 2-year progression-free survival (HR 0.38, 95% CI 0.22-0.63) and overall survival (HR 0.45, 95% CI 0.25-0.83) versus focal approaches, without increasing severe CRS or ICANS ^[127]C4. LDH normalization after bridging correlates with outcomes comparable to patients with persistently normal LDH ^[127]C4. Polatuzumab-based regimens yield higher bridging response rates (ORR 52%, CR 35%) than chemotherapy or radiation alone ^[126]B3b.

Step 2: Lymphodepletion Regimen Selection

Fludarabine 30 mg/m² daily × 3 days plus cyclophosphamide 300 mg/m² daily × 3 days remains the label-standard regimen for most CD19 CAR-T products. Bendamustine 90 mg/m² daily × 2 days offers comparable efficacy with significantly lower any-grade CRS (31% vs 55%) and any-grade ICANS (1% vs 33%) in real-world liso-cel cohorts ^[126]B3b. The choice balances institutional formulary, fludarabine availability, and patient-specific infection risk.

Step 3: Acute Toxicity Management Protocol

CRS grading and first-line intervention, Follow ASTCT consensus grading. For grade ≥2 CRS, administer tocilizumab 8 mg/kg IV (max 800 mg) per dose; repeat every 8 hours if needed, maximum 4 doses. Add dexamethasone 10 mg IV every 6-12 hours for CRS refractory to tocilizumab or concurrent grade ≥2 ICANS ^[99]B2a.

ICANS grading and first-line intervention, For grade ≥2 ICANS, initiate dexamethasone 10 mg IV every 6 hours; escalate to methylprednisolone 1 g IV daily for grade ≥3 ICANS. Tocilizumab is not recommended for isolated ICANS without concurrent CRS ^[99]B2a.

Refractory CRS/ICANS, Siltuximab 11 mg/kg IV improved CRS grade in 75% and ICANS grade in 60% of patients previously treated with tocilizumab or steroids respectively ^[122]C4. Anakinra 100 mg SC daily (up to 2 mg/kg/day) shows activity in steroid-refractory ICANS; 7 of 9 anakinra recipients in a liso-cel cohort had grade ≥3 ICANS ^[126]B3b.

Step 4: Product-Specific Toxicity Profiles and Monitoring

Cilta-cel carries higher severe CRS (OR 6.80, 95% CI 2.28-20.33) and delayed neurotoxicity (OR 20.07, 95% CI 4.46-90.20) versus ide-cel, without difference in any-grade ICANS ^[38]C4. Teclistamab demonstrates 18-month PFS 69.8% vs 26.9% for standard triplet therapy (HR 0.29, 95% CI 0.23-0.38) but with 41.6% grade 3-4 infections ^[117]A1b.

Step 5: Sequencing and Salvage Strategies

Meta-analysis of 2,122 patients indicates bispecific antibody therapy before CAR-T improves subsequent CAR-T complete response rates (53.7% vs 29.4%; p=0.0008) and objective response rates (RR 1.62, 95% CI 1.24-2.11) ^[116]A1a. Bispecific combination after CAR-T yields higher CR rates (46.4%) than bispecific monotherapy (29.4%) ^[116]A1a.

For multiple myeloma progressing after anti-BCMA CAR-T, anti-GPRC5D CAR-T achieves 84% ORR (95% CI 68-94%) with only 5% grade 3 CRS and 3% grade 1 ICANS ^[120]B2b. Talquetamab bridging to BCMA CAR-T yields 71% response rate, 88% post-CAR-T response (54% CR), with 2% grade 3 ICANS and no grade ≥3 CRS ^[123]C4.

Step 6: ICU-Level Support and Late Complications

Twenty-seven percent of CAR-T recipients require ICU admission within median 4.5 days of infusion; 90-day mortality 22.4% ^[118]B2b. Independent mortality predictors: frailty (HR 2.51), bacterial infection (HR 2.12), life-saving therapy within 24 hours (HR 1.80) ^[118]B2b. Grade ≥3 infections occur in 41-45% of real-world CAR-T cohorts ^{[69]B2b[117]A1b}.

Second primary malignancies reported in 4% of ide-cel recipients (1% myeloid) ^[69]B2b; non-significant increase with cilta-cel (OR 1.77, 95% CI 0.89-3.56) ^[38]C4. Movement disorders and cranial nerve palsies occur with cilta-cel (9% grade 2, 1% grade 3) ^[53]A1b.

Controversies and Guideline Disagreement

Pearl: Normalize LDH before lymphodepletion, it is the strongest modifiable predictor of CAR-T efficacy across products; use comprehensive bridging radiotherapy to achieve it, and select bendamustine lymphodepletion when feasible to reduce CRS/ICANS without compromising response ^{[126]B3b[127]C4}.

Step 1: Assess Hematologic Reserve and Bleeding Risk Before Infusion

Baseline cytopenias and inflammatory markers identify patients who will develop prolonged neutropenia, thrombocytopenia, and bleeding after CAR T-cell infusion. The CAR-HEMATOTOX model incorporates platelet count, hemoglobin, absolute neutrophil count (ANC), C-reactive protein, and ferritin measured before lymphodepletion; a high score predicts severe neutropenia ≥14 days (12 vs 5.5 days), severe thrombocytopenia (87% vs 34%), and anemia (96% vs 40%) ^[36]B3b. Baseline platelet-to-lymphocyte ratio (PLR) ≥198 identifies patients at higher risk for severe immune effector cell-associated toxicities including ICANS ^[138]C4. Endothelial activation quantified by the modified Endothelial Activation and Stress Index (mEASIX), derived from platelets, CRP, and LDH, predicts consumptive coagulopathy, CRS grade ≥2, and inferior progression-free and overall survival ^[102]D5.

Step 2: Transfusion Support During Aplasia

Profound neutropenia (ANC <100/µL) occurs in 72% of patients and is prolonged (≥21 days) in 64% after axicabtagene ciloleucel or tisagenlecleucel ^[36]B3b. Median duration of severe neutropenia (ANC <500/µL) is 9 days ^[36]B3b. Transfuse red cells for hemoglobin <7 g/dL (or <8 g/dL with cardiac comorbidity) and platelets for counts <10×10⁹/L (or <20×10⁹/L with fever, bleeding, or invasive procedure). Growth factor support with granulocyte colony-stimulating factor (G-CSF) is reasonable for prolonged severe neutropenia; thrombopoietin-receptor agonists (e.g., romiplostim, eltrombopag) may be considered for refractory thrombocytopenia ^[76]D5. Bone marrow biopsy is indicated for cytopenias persisting >90 days to exclude residual disease or secondary marrow neoplasm ^[76]D5.

Step 3: Anticoagulation and Thrombosis Prophylaxis

One-year cumulative incidence of venous thromboembolism (VTE) after axicabtagene ciloleucel is 6.3%; bleeding incidence is 11.0% ^[132]B2b. Median time to either event is 28 days post-infusion ^[132]B2b. Therapeutic anticoagulation at the time of infusion increases bleeding risk (HR 3.74, 95% CI 1.33-10.49) ^[132]B2b. Thrombocytopenia grade ≥3 (platelets <50×10⁹/L) at infusion also raises bleeding risk (HR 3.55, 95% CI 1.12-11.26) ^[132]B2b. A CAR-HEMATOTOX score ≥2 before therapy similarly predicts bleeding (HR 3.62, 95% CI 1.23-10.69) ^[132]B2b. In a Mayo Clinic series of 97 axi-cel recipients, only 2 (2.1%) developed VTE despite 57% receiving concurrent anticoagulation (53 as prophylaxis) ^[140]C4. Routine therapeutic anticoagulation is not recommended solely for CAR T-cell therapy ; prophylactic-dose low-molecular-weight heparin is appropriate for immobilized inpatients. Hold or reduce anticoagulation when platelets fall <50×10⁹/L or active bleeding occurs.

Step 4: Cytoreductive Bridging Before CAR T-Cell Infusion

Bridging radiotherapy achieves rapid cytoreduction in CNS lymphoma. In 12 patients receiving CNS bridging radiotherapy before commercial CAR T, mean lesion size reduction was 74.0% (95% CI 62.0-86.0) at a median of 12 days from radiotherapy completion to CAR T infusion ^[130]D5. Best CNS response included 8 complete responses and 1 partial response; 3 patients experienced CNS relapse outside the radiation field ^[130]D5. Grade ≥3 CRS occurred in 1 of 12 patients; grade ≥3 ICANS in 3 of 12 ^[130]D5. For systemic disease, lymphodepleting chemotherapy (fludarabine/ cyclophosphamide ) remains standard; a 2-day regimen (fludarabine 40 mg/m² + cyclophosphamide 500 mg/m² × 2 days) yields similar efficacy and CRS/ICANS rates as a 3-day regimen but with less prolonged thrombocytopenia (2% vs 27% platelet recovery >60 days) ^[131]C4. Intensive multi-agent bridging (e.g., VTD-PACE, DCEP) in multiple myeloma delays platelet and neutrophil recovery without improving progression-free survival after idecabtagene vicleucel ^[136]B3b.

Step 5: Monitoring and Escalation Triggers

• Daily CBC with differential through day +30; continue weekly until ANC >1×10⁹/L and platelets >50×10⁹/L without transfusion.
• Fibrinogen and platelet trends: a >10% day-over-day platelet drop or fibrinogen <250 mg/dL precedes high-grade ICANS in >75% of cases ^[137]B3b.
• mEASIX at baseline and during CRS/ICANS to track endothelial activation and coagulopathy risk ^[102]D5.
• Serum neurofilament light chain (NfL) before infusion: NfL >74.8 pg/mL predicts moderate-to-severe ICANS (sensitivity 0.88, specificity 0.50, AUC 0.71) ^[128]B2b.
• Bone marrow biopsy if cytopenias persist >90 days or new dysplasia appears ^[76]D5.

Drug / Modality Comparison Table

Dosing Table

Treatment Failure Protocol

• Refractory thrombocytopenia (>90 days, platelets <20×10⁹/L despite TPO-RA, transfusions, infection control): bone marrow biopsy → consider stem cell boost (autologous CD34+ cells) or eltrombopag escalation ^[76]D5.
• Persistent neutropenia (>90 days, ANC <500/µL despite G-CSF): evaluate for marrow infiltration, myelodysplasia, or viral suppression (CMV, HHV-6) ^[76]D5.
• Recurrent VTE on prophylactic anticoagulation: escalate to therapeutic anticoagulation if platelets >50×10⁹/L and no active bleeding; consider IVC filter if anticoagulation contraindicated.
• Life-threatening bleeding (intracranial, hemodynamic instability): platelet transfusion to >100×10⁹/L, 4-factor PCC, tranexamic acid 1 g IV q6h, hold anticoagulation, neurosurgical consultation.

What NOT to Do

• Do NOT use therapeutic anticoagulation routinely for VTE prophylaxis in CAR T-cell recipients; bleeding risk outweighs low VTE incidence (2.1% in axi-cel series) ^{[140]C4[132]B2b}.
• Do NOT omit baseline mEASIX or CAR-HEMATOTOX assessment; both independently predict severe hematologic toxicity ^{[36]B3b[102]D5}.
• Do NOT delay bone marrow biopsy for cytopenias >90 days; secondary myeloid neoplasms occur ^[76]D5.
• Do NOT use intensive multi-agent bridging (VTD-PACE/DCEP) expecting PFS benefit in myeloma; it delays hematologic recovery without improving survival ^[136]B3b.

Controversies and Guideline Disagreement

Pearl: Baseline CAR-HEMATOTOX and mEASIX scores stratify bleeding and cytopenia risk before CAR T infusion; a 2-day fludarabine/cyclophosphamide lymphodepletion regimen reduces prolonged thrombocytopenia without compromising efficacy; reserve therapeutic anticoagulation for documented VTE, prophylactic dosing suffices for most inpatients ^{[36]B3b[102]D5[131]C4[132]B2b}.

From CTCAE to Standardized ICANS Grading

Early CAR-T trials graded neurotoxicity using CTCAE v4.03, a scale not designed for immune effector cell syndromes. In the JULIET trial of tisagenlecleucel, CTCAE identified neurotoxicity in 58.5% of 106 infused patients (40.6% grade 1-2, 17.9% grade 3-4) ^[143]B2b. Retrospective regrading of the same dataset by expert panel using the CARTOX-derived modified CRES (mCRES) and the ASTCT ICANS criteria reduced the any-grade incidence to 17.1% each, reclassifying 31 patients (46%) from CTCAE grades 1-3 to grade 0 because their events, headache, anxiety, dizziness, peripheral neuropathy , insomnia, were deemed non-attributable encephalopathy ^[143]B2b. This discrepancy exposed CTCAE's subjectivity and over-capture of non-specific neurologic events, prompting adoption of the ASTCT ICANS consensus grading (ICE score, consciousness, seizures, motor findings, cerebral edema) as the universal standard ^[143]B2b.

Landmark Toxicity Signals That Shaped Monitoring

The JCAR015 (ROCKET) trial in adult ALL was halted after fatal cerebral edema, establishing that severe ICANS could be rapidly lethal and mandating intensive neurologic monitoring ^[143]B2b. In contrast, the JULIET lymphoma trial reported no grade 5 neurotoxicity, suggesting disease context and CAR construct influence severity ^[143]B2b. ZUMA-1 (axicabtagene ciloleucel) later showed higher rates of grade ≥3 ICANS (28% in patients ≥65 years) than tisagenlecleucel, reinforcing product-specific risk profiles ^[126]B3b.

Evolution of Prophylactic and Preemptive Strategies

Early practice treated ICANS reactively with corticosteroids after onset. The ZUMA-1 experience with axicabtagene ciloleucel demonstrated that prophylactic tocilizumab did not prevent ICANS despite reducing CRS ^[143]B2b. Subsequent real-world data with lisocabtagene maraleucel showed bendamustine lymphodepletion (vs fludarabine/ cyclophosphamide ) associated with significantly lower any-grade ICANS (1% vs 33%) and grade ≥3 ICANS (0% vs 13%) without compromising efficacy ^[126]B3b. This shifted lymphodepletion choice toward bendamustine where feasible, especially in older or comorbid patients.

Fractionated Dosing and Bridging Radiotherapy

Fractionated CAR-T dosing emerged as a strategy to mitigate peak cytokine exposure. The CART19-BE-02 trial of varnimcabtagene autoleucel used adaptive intra-patient dose escalation (0.1, 0.3, 0.6, 2.0 × 10⁶ CAR T cells/kg) with 24-hour intervals contingent on safety; only 1 of 32 patients (3%) developed grade ≥3 ICANS ^[34]B2b. Concurrently, bridging radiotherapy before CAR-T was evaluated for cytoreduction. The ILROG multicenter study of 172 patients showed comprehensive bridging RT (to all active sites) associated with superior PFS (HR 0.38, 95% CI 0.22-0.63) and OS (HR 0.45, 95% CI 0.25-0.83) and lower grade ≥3 ICANS rates when LDH normalized after RT ^[127]C4. LDH normalization after bridging therapy emerged as a prognostic surrogate for reduced neurotoxicity risk.

Refractory ICANS and Novel Agents

For steroid-refractory ICANS, anakinra (IL-1 receptor antagonist) and siltuximab (anti-IL-6) entered practice based on mechanistic rationale and case series. In the liso-cel real-world cohort, 9 patients received anakinra (7 with grade ≥3 ICANS) and 2 received siltuximab ^[126]B3b. These agents remain off-label but are incorporated into consensus algorithms for ICANS unresponsive to high-dose corticosteroids.

Current Paradigm: Risk-Adapted Monitoring and Grading

Today, ASTCT ICANS grading is mandatory for clinical trials and standard of care. Monitoring protocols incorporate serial ICE scoring, EEG for non-convulsive seizures, and neuroimaging for cerebral edema. Lymphodepletion selection (bendamustine preferred for lower ICANS), fractionated dosing for high-risk products, and bridging strategies aimed at LDH normalization represent the evolutionary trajectory from reactive management to risk-adapted prevention.

Pearl: The shift from CTCAE to ASTCT ICANS grading halved reported neurotoxicity incidence by excluding non-attributable events; current prevention focuses on bendamustine lymphodepletion, fractionated dosing, and bridging cytoreduction targeting LDH normalization, all strategies that lower peak inflammatory exposure without sacrificing efficacy ^{[143]B2b[126]B3b[34]B2b[127]C4}.

Respiratory Monitoring

Respiratory failure is the most common indication for ICU admission after CAR-T infusion. In the CARTTAS cohort, 27% of patients required vasoactive support within 24 hours of ICU admission, and hypoxia was a leading driver of ICU transfer ^[118]B2b. Continuous pulse oximetry and serial forced vital capacity (FVC) measurements are recommended during the first 14 days. Intubation criteria follow standard ARDS thresholds: PaO₂/FiO₂ < 150 on non-invasive support, FVC < 15 mL/kg, or progressive hypercapnia with altered mental status.

Autonomic Complications

Autonomic instability manifests as labile blood pressure, arrhythmias, ileus, and urinary retention. In the CARTTAS study, 27% of ICU patients required vasoactive drugs, and 12% developed bacterial infections that compounded hemodynamic instability ^[118]B2b. Sinus tachycardia is near-universal during CRS; new-onset atrial fibrillation or ventricular arrhythmias warrant telemetry and electrolyte repletion. Ileus correlates with high-grade ICANS and opioid use, abdominal radiographs and bowel regimen (polyethylene glycol 17 g daily) are indicated. Urinary retention occurs in up to 15% of patients with grade ≥3 ICANS; bladder scanning every 8 hours and clean intermittent catheterization prevent overflow injury.

DVT/PE Prophylaxis

All hospitalized CAR-T recipients should receive pharmacologic VTE prophylaxis unless contraindicated. Enoxaparin 40 mg SC daily (30 mg if CrCl < 30 mL/min) is preferred; unfractionated heparin 5000 units SC q8h is an alternative for renal impairment. Mechanical prophylaxis (sequential compression devices) is added for ICU patients. Prophylaxis continues through hospitalization and for 2 weeks post-discharge in patients with persistent immobility or active malignancy ^[118]B2b.

Pain Management

Pain syndromes include mucositis (grade 3 in 2% after bridging RT ^[127]C4), neuropathic pain from ICANS, and musculoskeletal pain from cytokine flare. Multimodal regimen: scheduled acetaminophen 1 g q6h, gabapentin 300 mg q8h titrated to 1200 mg q8h for neuropathic component, and opioid PCA ( hydromorphone 0.2 mg demand, 6-min lockout) for breakthrough. Avoid NSAIDs in thrombocytopenia (platelets < 50×10⁹/L).

Rehabilitation

Early mobilization begins once hemodynamically stable (MAP ≥65 mmHg off vasopressors) and afebrile for 24 hours. Physical therapy twice daily targeting sitting balance, transfer training, and ambulation. Occupational therapy addresses activities of daily living and cognitive screening (MoCA) for ICANS survivors. Speech-language pathology evaluates swallow safety before oral intake in patients with grade ≥2 ICANS.

Hospital-Acquired Complications

Bacterial infection developed in 12% of ICU CAR-T patients and independently doubled 90-day mortality (HR 2.12, 95% CI 1.11-4.08) ^[118]B2b. Prevention bundle: chlorhexidine bathing daily, elevation of head of bed 30-45°, daily sedation vacation and spontaneous breathing trial, Foley catheter removal within 24 hours of ICU transfer, and antibiotic stewardship (de-escalate at 48 hours if cultures negative). Pressure injury prevention: Braden scale q12h, turning q2h, silicone foam dressings over bony prominences. UTI prevention: closed drainage, catheter removal as soon as feasible, asymptomatic bacteriuria not treated.

Complication Summary Table

Pearl: LDH normalization after bridging therapy predicts lower grade ≥3 ICANS and better survival; use serial LDH as a real-time cytoreduction biomarker to guide escalation of supportive care ^[127]C4.

Prognosis by Disease Context and CAR-T Product

Outcomes after ICANS are inseparable from the underlying malignancy, the CAR-T construct, and the intensity of acute toxicity. In large B-cell lymphoma (LBCL), real-world cohorts show 2-year progression-free survival (PFS) of 38-46% and overall survival (OS) of 53-63% depending on product ^{[40]B3b[127]C4}. Axi-cel yields 2-year PFS 46% and OS 63%; tisa-cel 30% and 45%; liso-cel 45% and 58% ^[40]B3b. In relapsed/refractory multiple myeloma , cilta-cel achieves median PFS not reached at 15.9 months with 12-month PFS 75.9% versus 11.8 months and 48.6% for standard care ^[53]A1b; ide-cel shows median PFS 8.8 months in standard-of-care use ^[69]B2b. For Burkitt lymphoma , median PFS is only 2.3 months and median OS 6.0 months despite a 58% initial response rate ^[124]C4. Richter transformation carries median PFS 4.7 months and median OS 8.5 months ^[125]C4.

Impact of ICANS Grade on Survival

Severe ICANS (grade ≥3) correlates with higher non-relapse mortality (NRM). In the CARTTAS ICU cohort, 90-day mortality was 22.4% (95% CI 17.1-27.7%) among CAR-T recipients requiring intensive care; grade 3-4 ICANS was present in 35% of evaluated patients and frailty, bacterial infection, and early life-saving therapy independently predicted death ^[118]B2b. In LBCL, NRM at 1 year is 7% after liso-cel, driven by early ICANS/CRS deaths then infections ^[126]B3b. No ICANS-related deaths occurred in the bridging radiotherapy cohort despite 24% grade ≥3 ICANS ^[127]C4.

Delayed Neurotoxicity and Long-Term Sequelae

Cilta-cel carries a distinct delayed neurotoxicity syndrome (movement disorders, parkinsonism , cranial nerve palsies) not captured by standard ICANS grading. In CARTITUDE-4, 9.1% of cilta-cel recipients developed cranial nerve palsy (grade 2-3) and 2.8% had CAR-T-related peripheral neuropathy ^[53]A1b. Real-world comparison shows cilta-cel associated with higher delayed neurotoxicity (OR 20.07, 95% CI 4.46-90.20) versus ide-cel ^[38]C4. Long-term neurocognitive outcomes remain understudied; no Parkinsonism was reported in 821 ide-cel recipients ^[69]B2b.

Prognostic Biomarkers

Pre-lymphodepletion LDH ≤ upper limit of normal predicts superior PFS (HR 0.25, 95% CI 0.12-0.55) and OS (HR 0.30, 95% CI 0.13-0.68) after liso-cel ^[126]B3b. LDH normalization after bridging radiotherapy identifies patients with 1-year PFS 66-74% versus 17% if LDH remains elevated ^[127]C4. High CAR-HEMATOTOX score predicts severe infections with bispecific antibodies post-CAR-T ^[160]C4.

Natural History Without Effective Salvage

After CAR-T failure in DLBCL, historical median OS is 5 months ^[160]C4. Odronextamab post-CAR-T achieves median OS 10.2 months (12-month OS 45.5%) and CR rate 31.7% ^[160]C4. Sequencing bispecific antibodies before CAR-T improves subsequent CAR-T complete response rates (53.7% vs 29.4%) ^[116]A1a.

Pearl: LDH normalization after bridging therapy and pre-lymphodepletion LDH ≤ ULN are the strongest validated prognostic surrogates for durable remission; grade ≥3 ICANS signals elevated early NRM but not necessarily worse long-term survival if rescued. ^{[126]B3b[127]C4[118]B2b}

Pediatrics

EASIX and m-EASIX scores predict CRS and ICANS in pediatric and AYA patients after CD19- CAR T-cell therapy , with m-EASIX at day +3 achieving 90.6% AUC for severe CRS and 82.2% AUC for severe ICANS ^[161]C4. In a 76-patient cohort (median age), CRS occurred in 61.8% and ICANS in 22.4%; severe CRS (grade 3-4) in 18.4% and severe ICANS in ^[161]C4. Lymphodepletion used fludarabine 120 mg/m² and cyclophosphamide 1000 mg/m² for tisagenlecleucel; 75 mg/m² and 900 mg/m² for investigational product ^[161]C4. Point-of-care fresh CD19-CAR_Lenti at 3×10⁶ CAR⁺ T cells/kg yielded grade 1-2 CRS in 68% and grade 1-2 ICANS in 11% (2/19), both self-limited ^[75]C4. Children with trisomy 21 had prolonged late ICAHT (>14 days in 3/3) but no increase in clinically relevant infections ^[75]C4. Neurofilament light chain (NF-L) in CSF rose significantly in the two patients who developed ICANS, suggesting a potential early biomarker ^[75]C4.

Pregnancy

No data on CAR T-cell therapy or ICANS management in pregnancy are reported in the provided evidence. Teratogenicity of lymphodepletion (fludarabine/cyclophosphamide), tocilizumab , siltuximab, anakinra, and corticosteroids is not addressed. Delivery planning and breastfeeding safety are not reported.

Elderly

In lymphoma patients receiving CD19 CAR-T, cardiovascular event incidence was 28.7% in the elderly versus 13.5% in adults (statistically significant) ^[167]A1b. CV events and hypotension strongly associate with CRS severity ^[167]A1b. Real-world ide-cel in renal impairment (median age 71 vs 67 years) showed similar ICANS rates (16% vs 13%) and grade ≥3 ICANS (2.5% vs 2.6%) regardless of renal function ^[165]D5.

Renal Impairment

Teclistamab in relapsed/refractory multiple myeloma with CrCl <40 mL/min (21% of 384 patients, 18% severe <30 mL/min or dialysis) demonstrated comparable ICANS rates (16% vs 13%) and grade ≥3 ICANS (2.5% vs 2.6%) to patients without renal impairment ^[165]D5. Ide-cel with CrCl <50 mL/min (13% of 214 patients) showed similar ICANS incidence (23% vs 20%) ^[89]D5. Renal function did not worsen after either therapy ^{[89]D5[165]D5}. Higher baseline and day 30 grade ≥3 anemia and thrombocytopenia occurred with renal impairment on teclistamab ^[165]D5.

Immunocompromised / Post-Allogeneic Transplant

Allogeneic donor-derived CD19-CAR-T (1-3×10⁶ cells/kg) in 13 pediatric/young adult BCP-ALL patients post-allo-HSCT produced grade 1 CRS max and grade 2 ICANS; one acute GVHD controlled with steroids/ruxolitinib; no GVHD in 3 haploidentical donors ^[78]D5. All achieved MRD-negative CR; 8/13 maintained CR at median 12-month follow-up ^[78]D5. In point-of-care autologous CAR-T, 6/19 patients had prior allo-HSCT; they experienced higher rates of late ICAHT but comparable CRS/ICANS ^[75]C4.

Pearl: Pediatric ICANS prediction is strengthened by day +3 m-EASIX (AUC 82.2% for severe ICANS) and CSF NF-L elevation; elderly patients face disproportionate CV toxicity (28.7% vs 13.5%) linked to CRS severity; renal impairment does not increase ICANS risk with teclistamab or ide-cel; allogeneic CAR-T post-transplant yields manageable neurotoxicity (grade 2 max) without excess GVHD ^{[161]C4[75]C4[167]A1b[165]D5[89]D5[78]D5}.

Primary Prevention: Infection Prophylaxis and CRS/ICANS Mitigation

Infection risk after CAR-T and bispecific antibody therapy drives much of the preventable morbidity. The EBMT/JACIE best-practice recommendations endorse antiviral prophylaxis for varicella-zoster virus (VZV) and Pneumocystis jirovecii pneumonia (PJP) with trimethoprim-sulfamethoxazole as standard ^[178]D5. Monthly intravenous immunoglobulin (IVIG) replacement is advised for all patients receiving bispecific antibodies regardless of IgG level, a strategy that independently reduced grade ≥3 infections by 62% (HR 0.38, 95% CI 0.19-0.79) in a 158-patient real-world cohort ^[49]B3b. Extending bispecific dosing to every-other-week after cycle 3 (upon partial response or better) and to every 4 weeks after cycle 7 further lowered severe infection risk (HR 0.37, 95% CI 0.20-0.69) ^[49]B3b.

Prophylactic tocilizumab administered before step-up doses of bispecific antibodies cut CRS incidence without increasing infections; 96-100% of patients received it in the same cohort ^[49]B3b. For CAR-T, the COMMIT consensus panel recommends antimicrobial prophylaxis tailored to the depth and duration of cytopenias, including antibacterial (fluoroquinolone) and antifungal (posaconazole) agents during neutropenia ^[176]C4.

Secondary Prevention: Early Detection of ICANS and Recurrence

No validated biomarker prevents ICANS outright, but serial platelet and fibrinogen dynamics show promise. In 265 B-cell lymphoma patients, a platelet count <100 × 10⁹/L one day before onset predicted high-grade ICANS in >75% of cases (median decrease 11.6% vs 1.6%, p=0.001) ^[137]B3b. Rising fibrinogen (>400 mg/dL) the day before onset also correlated with ICANS (p=0.03) ^[137]B3b. These parameters are routinely available and can trigger intensified neurological monitoring.

Baseline brain MRI before CAR-T infusion identifies pre-existing CNS involvement; in 154 patients, urgent MRI during ICANS revealed abnormalities in 40.7% and two distinct patterns (central variant and stroke-like) that inform differential diagnosis ^[21]C4. Screening for CNS myeloma before BCMA CAR-T is advised in high-risk patients ^[81]D5.

Surveillance: Long-Term Monitoring and Vaccination

Immunoglobulin monitoring monthly for at least 12 months post-infusion guides IVIG replacement (target IgG >400 mg/dL) ^[49]B3b. CMV DNAemia surveillance on day 1 of each cycle is recommended for bispecific recipients; cumulative incidence reached 48% at 10 months with anti-BCMA agents ^[49]B3b. Preemptive antiviral therapy at viral loads 452-134,000 IU/mL prevented CMV disease in all eight clinically significant cases ^[49]B3b.

Vaccination: Inactivated vaccines (influenza, pneumococcal, recombinant zoster, SARS-CoV-2, RSV) should be updated before lymphodepletion and repeated per CDC/ACIP schedules after immune reconstitution (typically ≥6 months post-CAR-T) ^[49]B3b. Live vaccines are contraindicated during active immunosuppression.

Neurocognitive surveillance: Serial Montreal Cognitive Assessment (MoCA) at baseline, during ICANS, and at 1 month detects subtle deficits missed by ICE score]) ^[182]B2b. Handwriting and tremor changes are early clinical markers of incipient ICANS ^[182]B2b.

Late effects: Prolonged cytopenias, hypogammaglobulinemia, and B-cell aplasia require ongoing hematologic and infectious surveillance per EBMT long-term follow-up guidelines ^[178]D5. Cardiac monitoring (echo/ECG) is warranted given 17.8% CV event rate with CD19 CAR-T ^[167]A1b.

Controversies and Guideline Disagreement

Pearl: Monthly IVIG from cycle 1 of bispecific therapy and platelet/fibrinogen trending during the first 2 weeks post-CAR-T are the two highest-yield, actionable prevention strategies currently supported by outcome data ^{[49]B3b[137]B3b}.