Neurotoxicities After CAR T-Cell Immunotherapy

Timing of Neurotoxicity

Neurologic signs and symptoms that are associated with CAR T-cell treatment typically follow a monophasic time course. Manifestations such as delirium, language disturbance, tremor, transient focal weakness, behavioral disturbances, ataxia, peripheral neuropathy, visual changes and generalized weakness, seizures, and acute cerebral edema usually develop in the first 10 days after CAR T-cell infusion. Headache is often present in conjunction with neurotoxicity but may also be present in the absence of neurologic symptoms. In many cases, the findings worsen over the course of several days and usually take a similar amount of time to resolve ( Fig. 7.1 ). In most patients, several different signs and symptoms occur at the same time, but confusion, language dysfunction, or delirium are often the first manifestations to appear and the last to resolve. In a study of neurotoxicity in 133 adults with ALL, NHL, or CLL who received T-cells expressing a CD19-targeting CAR with a 4-1BB costimulatory domain, neurotoxicity onset occurred a median of 4 days after CAR T-cell infusion, with a peak on day 7, and a duration of 5 days. In a similar study of 53 adults who received CD28-costimulated CD19-targeted CAR T-cells for treatment of ALL, neurotoxicity started a median of 5 days after CAR T-cell infusion, the first severe neurotoxicity symptom occurred at a median 9 days postinfusion, and resolution occurred a median of 11 days after onset.

The kinetics of presentation may differ between studies investigating different CAR T-cell therapies and studies where dedicated assessment tools are employed to identify early signs of subtle neurotoxicity. Most signs and symptoms of neurotoxicity resolve within 21 days of CAR T-cell infusion, although in rare cases they can persist longer, either owing to irreversible neurologic injury, prolonged headaches, or cognitive and attentional complaints. Rarely, neurotoxicity onset has occurred after 14 days due to delayed CAR T-cell expansion and CRS. One-month outcomes on neurocognitive screening tests were overall stable or improved in a cohort of 22 pediatric patients treated with CD22-targeting CAR T-cells. However, long-term cognitive and neurodevelopmental outcomes have not yet been reported in the CAR T-cell–treated patient population. Comprehensive long-term follow-up as well as prospective neurocognitive testing would be required to determine whether acute neurotoxicity causes chronic changes that affect brain health.

Signs and Symptoms

The most common manifestation of neurotoxicity is a transient impairment of attention and cognitive processing, typically described as delirium or confusion. Language disturbance is also frequent and is one of the more characteristic signs of ICANS. This can easily be missed in subtle cases or when patients are systemically very ill. Impaired language production can be a manifestation of true aphasia, which is due to dysfunction of cortical and subcortical language modules, but abnormal language output can also occur related to impaired attention or arousal, as occurs in delirium and depressed level of consciousness. Validated mental status exam tools are required to make the distinction. Of note, no imaging abnormalities have been reported in neurotoxicity patients that specifically localize to language centers of the brain such as Broca's or Wernicke's areas without affecting other brain regions.

Headache is very common and often an early sign of neurotoxicity. Although this temporal association suggests that headache can be a symptom of neurotoxicity, it is nonspecific and can be seen without other associated neurologic symptoms. Therefore, headache does not contribute to the criteria for diagnosis and grading of ICANS.

Seizures have primarily been reported in patients with ALL who received CD19-directed CAR T-cells, with incidences ranging from 0% to 30% after CD28-costimulated and 3%–14% after 4-1BB CAR treatment. Only isolated cases of seizure have been reported from other CAR T-cell trials (BCMA, 1 of 57 patients treated ; EGFRvIII for glioblastoma, 1 of 10 patients ), although there may be additional cases that are not reported here because not all authors have provided comprehensive listings of toxicities. The reason for the higher seizure risk in ALL patients is unknown but could be related to younger patient age (which appears less likely, since both pediatric and adult ALL trials report similar seizure incidences), more severe CRS, or higher rates of preexisting CNS comorbidities.

Other frequently reported signs and symptoms include tremor, visual hallucinations, behavioral disturbances, ataxia, peripheral neuropathy, visual changes, and generalized weakness. The overall pattern is that of a global, toxic-metabolic encephalopathy.

Focal pathology leading to focal neurologic dysfunction, such as unilateral limb weakness or cranial neuropathy, is unusual, and other etiologies should be considered. These include CNS infection, hemorrhagic or ischemic stroke, and malignant CNS involvement. Since patients with relapsed/refractory hematologic malignancies are at overall increased risk of such neurologic complications, further study will be needed to determine whether their incidence is affected by CAR T-cell therapy.

Pediatric CAR T-cell patients appear to have similar neurotoxicity profiles as adults. However, subtle cognitive dysfunction and language disturbance may be more difficult to detect in young children, requiring specific pediatric assessment tools. Proposed pediatric-specific measures include a neurologic symptom checklist that can be filled out by caregivers or an observational tool for mental status assessment that can be used instead of language-based instruments that require patient cooperation (see ICANS Diagnostic Criteria section).

Acute cerebral edema is a life-threatening complication that to date has only been described after treatment with CD19-directed CAR T-cells. In one clinical trial for adults with ALL, fatal cerebral edema occurred in five patients. This forced the termination of the trial. Deaths due to cerebral edema have been reported in clinical trials using CAR constructs containing either 4-1BB or CD28 costimulatory domains, in children and adults with ALL, and in adults with NHL and CLL. All published cases have been associated with CRS. Cerebral edema typically develops in the first week after CAR T-cell infusion and often at a time when CRS-associated signs such as fever and hypotension are abating after treatment with tocilizumab and/or steroids. The first signs of impending cerebral edema can occur with no preceding neurotoxicity, or as a worsening of existing neurologic signs and symptoms, occasionally after initial improvement. Within hours, symptoms can evolve from mild somnolence and confusion to those associated with increased intracranial pressure such as headache, nausea, vomiting, and decreasing level of consciousness. Blood pressure may rise due to autoregulatory responses to preserve cerebral perfusion. Patients then become comatose and require invasive ventilation and may develop seizures. Bradycardia is a late sign of impending herniation. Even if treated aggressively with measures directed toward CRS management and neuroprotection, edema can quickly become irreversible to the point of herniation and brain death.

Other fatal neurologic events after CD19 CAR T-cell therapy have included cortical necrosis, acute cerebral hemorrhage during resolving CRS, edema and hemorrhage of the brainstem and deep brain structures, and multifocal thrombotic microangiopathy. One case of chronic, progressive neurologic decline was reported in a patient with preexisting optic atrophy, and pathologic examination revealed severe diffuse white matter degeneration and neuronal loss. Since this case did not have the typical acute presentation of neurotoxicity, a distinct underlying pathophysiology may have contributed to the clinical findings.

Findings on Standard Clinical Blood Tests

Abnormalities in several standard clinically available markers have been identified in patients with neurotoxicity. However, many of these patients have concurrent CRS, and it is difficult to ascertain whether any of these abnormalities are specifically associated with neurotoxicity. We assessed clinical laboratory markers for CRS and neurotoxicity in a cohort of 133 adult patients after 4-1BB-costimulated CAR T-cell treatment. The most pronounced abnormalities were seen in patients with grade ≥4 CRS as well as in patients with grade ≥3 neurotoxicity. For example, the acute phase reactants CRP and ferritin were significantly higher in patients with grade ≥3 neurotoxicity during the first week after CAR T-cell infusion compared with patients with grade 0–2 neurotoxicity, and the increase in CRS was seen as early as 0–36 h after CAR T-cell infusion. The pattern of ferritin and CRP increases was the same when comparing patients with grade ≥4 CRS with those with grade 3 CRS. In a different study, day 3 and peak CRP levels, as well as day 3 but not peak ferritin levels, correlated with the severity of neurotoxicity. Other markers that were abnormal in patients with neurotoxicity and/or CRS in our cohort during the acute toxicity phase in the first 14 days after CAR T-cell infusion include decreased serum total protein and albumin levels, and derangements of coagulation assays (including elevated PT, PTT, and D-dimer, and decreased fibrinogen and platelet counts).

Findings on Standard Clinical Cerebrospinal Fluid Examination

Neurotoxicity is associated with increased cerebrospinal fluid (CSF) protein concentration and leukocyte counts, consistent with increased trafficking of cells and serum proteins across the blood-CSF barrier. The protein concentration in the CSF can be very high, occasionally above 1 g/dL, but the increase is transient and normalizes soon after resolution of neurologic symptoms. CSF cell counts are typically modestly elevated and rarely above 100 cells/μL. The cellular composition of the CSF is typically lymphocyte predominant, and presence of significant numbers of neutrophils should prompt investigation for other causes of neurologic dysfunction, such as infection. Although infectious workup is usually negative in patients with a typical course of neurotoxicity, empiric antibacterial and antiviral coverage, as well as CSF culture and viral PCRs, should be strongly considered.

CNS Imaging

Head imaging should be considered in all cases of neurotoxicity, but the clinical utility of normal or abnormal imaging results is not well studied. Magnetic resonance imaging (MRI) is preferred if the patient is stable enough to undergo the study, and contrast administration may be helpful for detecting evidence of inflammatory infiltrates by leptomeningeal enhancement and blood-brain barrier breakdown by enhancement of lesions in the brain parenchyma ( Fig. 7.2 ). Head CT is useful to rule out acute abnormalities such as hemorrhage but is not sensitive to most of the changes that are seen with neurotoxicity. In patients with mild acute neurotoxicity, imaging is frequently normal. The incidence of MRI abnormalities is higher in patients with severe neurotoxicity, and head imaging may be considered in such cases both to rule out other etiologies such as stroke and to monitor response to treatment.

The most common and characteristic MRI pattern during neurotoxicity is one of symmetric T2 hyperintensities and swelling of the thalami and other deep gray matter structures, which is indicative of interstitial edema. These lesions were noted to be reversible in patients who underwent repeat imaging after resolution of neurologic symptoms. The deep gray matter lesions are typically nonenhancing and without restricted diffusion, but there may be associated microhemorrhages. This pattern is often seen in patients with decreased level of consciousness but can also occur in patients with mild, nonspecific symptoms. Similar symmetric deep gray matter edema, with or without diffusion restriction, may be seen in many other neurologic disorders, such as hypoxic-ischemic brain injury, a number of toxic-metabolic etiologies, central posterior reversible encephalopathy syndrome (PRES), acute disseminated encephalomyelitis, and acute necrotizing encephalopathy.

Patchy reversible T2 hyperintensities can also be seen in the cerebral white matter and can be associated with patchy enhancement that resolves over time. This can affect the supratentorial or the cerebellar white matter, similar to what is seen in inflammatory disorders of the white matter, such as multiple sclerosis or CNS vasculitis. An unusual involvement of the extreme and external capsule white matter has been reported by several groups. Acute T2 hyperintensities, with or without patchy enhancement, have also been described in areas of known CNS injury from other causes such as catheter placement or hemorrhage that were present prior to CAR T-cell treatment.

Diffuse leptomeningeal enhancement can occur during neurotoxicity, indicating meningeal inflammation that is likely associated with increased cell and protein trafficking across the blood-CSF barrier. Diffusion restriction, which is associated with cytotoxic edema, has been seen in regions of cortical gray matter and can be reminiscent of PRES. This can evolve into cortical laminar necrosis, which is irreversible and leads to chronic neurologic deficits that refer to the cortical location of the lesion. Finally, global cerebral edema can be identified on imaging and is associated with devastating neurologic injury.

It is not known whether the different imaging patterns are all indicative of the same underlying pathophysiology or whether separate processes contribute. Individual patients frequently manifest multiple of the findings listed above, often without a clear clinical correlate of specific symptomatology or severity of symptoms. This supports the conclusion that they may all be part of the same process, which may manifest differently based on individual patient characteristics.

Since neurotoxicity is a clinical diagnosis, imaging typically plays a supportive role in management of CAR T-cell patients, both to rule out alternative diagnoses and to monitor the evolution of known abnormalities. Preexisting head imaging abnormalities such as diffuse T2 hyperintensities in the supratentorial white matter are common in patients who have undergone chemotherapy and/or radiation, and comparison with prior or follow-up imaging is needed to distinguish these chronic changes from acute toxicities. Consideration can be given to obtaining head imaging prior to CAR T-cell treatment in all patients; however, it is unclear whether this would change management. Since no studies have defined criteria for obtaining head imaging, either before, during, or after treatment, the true incidence of MRI abnormalities in neurotoxicity is unknown. It is also not known whether the presence or absence of abnormalities on head imaging correlates with neurologic and overall outcomes after CAR T-cell therapy and whether treatment decisions should be made based on imaging findings. To clarify these issues, prospective imaging studies using standardized neurologic assessment criteria will be needed.

Electroencephalography

The role of electroencephalography (EEG) monitoring in neurotoxicity is not yet well defined, although its routine use for risk assessment has been advocated. EEG can either be useful for detecting subtle trends in background patterns, which may be an early warning sign for deterioration, or to rule out subclinical seizures. In neurotoxicity patients who underwent EEG, the most common finding was diffuse slowing, a nonspecific indicator of encephalopathy that is common in critically ill patients. Causes of diffuse slowing can include medication effects, toxic metabolites, or hypoxia/ischemia. The presence of interictal epileptiform discharges is associated with a higher risk of seizures, although their absence does not rule out the possibility of seizures. Nonconvulsive seizures or subclinical status epilepticus have been reported in CAR T-cell patients, although this is typically only seen after clinical seizures are treated with antiseizure medications, and EEG monitoring is started to assess treatment response. Only rarely, subclinical seizures are found to be the cause of depressed mental status without other clinical indicators of seizure activity. To compare the true incidence of epileptiform abnormalities and seizures between different CAR T modalities and patient populations, defined EEG monitoring criteria will be needed. Ideally, prospective studies should use standardized criteria for neurotoxicity grading to trigger initiation of EEG monitoring and develop protocols to guide its duration.