Neurologic Complications of Chimeric Antigen Receptor Therapy

Epidemiology and Pathogenesis

Beginning in 2017, there has been a steady increase in use of U.S. Food and Drug Administration (FDA)-approved chimeric antigen receptor (CAR) T-cell products for refractory relapsed hematologic tumors. Various host factors like disease burden, pretreatment inflammation, and pretreatment endothelial activation have been reported to predict the neurotoxicity. Conditioning regimen affects the toxicity, with 60 Cy/125 Flu preconditioning seeming to achieve maximum effect with minimum toxicity. CAR T-cell product characteristics such as the number of cells transfused, costimulatory domain differences, levels of various cytokines produced, and endothelial activation biomarkers influence the neurotoxicity. Fever > 38.9ºC, interleukin (IL-6) >16 pg/mL, and monocyte chemoattractant protein 1 > 343.5 pg/mL in the first 36 h predict neurotoxicity. Earlier higher-grade cytokine release syndrome (CRS) and C-reactive protein (CRP) predict neurotoxicity. In the multivariable predictive model, having a score of 6 out of total 14 points could predict neurotoxicity occurrence with 82% sensitivity and 70% specificity ( Table 36.1 ). This type of clinical modeling is practical and easily adaptable given the readily available clinical data to help decide which patient could be discharged early. Cytokine measurements, while useful to understand the process, are not readily available. While some generalizations could be made, individual products differences and additional host factors need to be taken in consideration, for example, patients with central nervous system (CNS) disease and patients with preexisting neurologic deficits (preexisting cognitive impairments).

Other models include combining Endothelial Activation and Stress Index (EASIX) score with additional parameters, such as ferritin and CRP. EASIX, defined as creatinine × lactate dehydrogenase (LDH)/platelets, is a marker of endothelial activation that is validated in the allogeneic hematopoietic cell transplant patients. Variations of EASIX (m-EASIX, replacing creatinine with CRP) can predict patients with higher-grade neurotoxicity.

These clinical models reinstate to some extent what has been published about pathogenesis. Pre- and post–CAR T-cell cytokine measurements in the serum and cerebrospinal fluid (CSF) corelate with immune effector cell-associated neurotoxicity syndrome (ICANS), especially higher-grade ICANS. Across multiple trials, mediators associated with ICANS are interferon (IFN)γ, IL-15, IL-6, IL-10, granulocyte colony-stimulating factor (GM-CSF), IL-2, IL-2Rα, IL-1RA, CXCL10, and Granzyme b. These mediators can be appropriately called inflammatory secretome . Some cytokine elevations like IL-4 were not shown to be associated with ICANS. In pediatric leukemia, CD 19 CAR T-cell trial ICANS alone patients showed deferential elevation of IL-2, soluble IL-4 receptor, hepatocyte growth factor, and IL-15 compared to patients with CRS alone. There is trafficking of CAR T-cells into CNS but no documentation of tumor toxicity in the neural tissue seen at autopsy. Blood-brain barrier becomes more permeable, and this is reflected in the CSF studies showing increased protein and increased CSF/albumin ratio. Documented CSF cytokine elevations mirrored serum elevations, arguing against local production and favoring a spillover from a permeable blood-brain barrier. Trials in hematologic malignancies suggest direct correlations of ICANS to the occurrence and severity of CRS. This strong association of cytokines and ICANS explains the reduced occurrence of ICANS in solid tumor trials given lack of robust CAR T-cell proliferation and limited CRS seen. Blinatumomab, immune effector engaging cell therapy, causes severe CRS and neurotoxicity in 40% to 50% patients with documented elevations in cytokines. Cytokines cause neurotoxicity via various mechanisms. Endothelial dysfunction, changes in cerebral perfusion, neuronal excitability, and glial solute handling are some of the proposed mechanisms. Clinical conditions with similar mechanisms include preeclampsia, eclampsia, sepsis, severe influenza with cerebral edema, and posterior reversible encephalopathy syndrome. In most patients this is a monophasic presentation, with probable relation to a combination of factors (peak and rate of cytokine elevations, degree of blood brain barrier distruption, timing of endothelial dysfunction). Measuring cytokines in clinical practice at the current time remains impractical. Close monitoring of clinical effects of cytokines such as CRS and good clinical bedside neurologic examination are felt to serve these patients best, as balance is achieved between tolerable effects of proliferating and tumor-killing aspects of CAR T-cells from the detrimental effects of cytokines. Extreme and fatal cerebral edema, which fortunately is very rare and only seen during initial trials, reflects the detrimental effects of cytokines on endothelium, microglia, perivascular space, and astrocytes. As management of post–CAR T-cell infusion is standardized and early aggressive CRS is undertaken, it is possible that later-ensuing higher-grade ICANS could be less frequent. Knowledge of drug kinetics is important, as products such as tisagenlecleucel with lesser peak and prolonged proliferative indices have fewer incidences of ICANS compared to axicabtagence ciloleucel. Prophylactic use of IL-6 with tocilizumab in adults with non-Hodgkin lymphoma treated with axicabtagene ciloleucel on day 2 decreased ≥ 3 CRS but with increased incidence of grade ≥ 3 ICANS. Tocilizumab being IL-6 R blocker increases IL-6 levels paradoxically. Direct IL-6 blocker siltuximab is preferred in patients with ICANS alone while the use of tocilizumab is recommended when there is concomitant ICANS and CRS. Tocilizumab is dosed 8 mg/kg intravenously (IV) for up to three doses in 24 h to a maximum of four doses. Siltuximab (not FDA approved) is dosed 11 mg/kg IV once every 3 weeks. There is no role for tocilizumab in patients with isolated ICANS and no CRS . It is a predicament, as CRS itself is a risk factor for ICANS. Recent small sample study showed prophylactic tocilizumab can mitigate CRS without increasing ICANS. Consensus for now is not to use tocilizumab for ICANS.

Changes in other constituents of nervous system were seen in animal models and in patients at postmortem. Brain microvascular endothelial changes, changes in pericytes, microglial activation, astrocyte activation and changes, neuronal dysfunction, and changes in excitatory neurotransmitters are all described. To a large extent, these changes likely reflect the inflammatory secretome (cytokine elevations, cause and effect) rather than primary detrimental effects of CAR T-cells on neuronal tissue. In a very significant proportion of patients, there is good response to steroids, suggestive of a reversible nature. The clinical correlates of the aforementioned pathology and utility of ancillary neurologic tests will be described in the next section. Published incidence of neurotoxicity with currently available products is detailed in Table 36.2 .

Clinical Presentation and Assessment

The two-rainbow model shown in Fig. 36.1 highlights the occurrence of earlier CRS that generally predates and predicts the incidence of ICANS. CRS constitutes clinical and laboratory features of systemic inflammation. Fever, hypoxia and hypotension, myalgias, arthralgia, and elevation of CRP and ferritin are seen. Encephalopathy is the main clinical presentation of CAR T-cell neurotoxicity. This is like toxic metabolic encephalopathy or delirium with reduced attention span as the predominant finding. The American Society for Transplantation and Cellular Therapy in 2018 by consensus proposed the term ICANS as a disorder characterized by pathologic process involving the CNS following any immune therapy that results in the activation or engagement of endogenous or infused T-cells and/or other immune effector cells. Symptoms or signs can be progressive and may include aphasia, altered level of consciousness, impairment of cognitive skills, motor weakness, seizures, and cerebral edema. This includes not just CAR T-cell but also immune effector-cell therapy such as blinatumomab. Clinical manifestations recognized over multiple trials include headache, meningismus, frontal release signs, personality changes, language dysfunction (inability to verbalize, inability to complete sentences), inability to do sequential multistep tasks, involuntary movements, and tremors. These could be graded per the Common Terminology Criteria for Adverse Events. There are five main domains in ICANS (see Fig. 36.1 and Table 36.3 ) and four grades. Grades 3 and 4 are higher grades. The most common encephalopathy is graded using the Immune Effector Cell associated Encephalopathy (ICE) score which would be the first domain ( Table 36.4 ). Other domains are level of: (2) consciousness (whether a patient is easily arousable or in deep coma), (3) seizures (easily controlled or prolonged refractory seizures), (4) motor weakness (any new motor weakness), and (5) raised intracranial pressure (from earlier focal edema seen on imaging to more overt clinical findings like VI nerve palsy, elevated intracranial pressure seen with ommaya or lumbar puncture, papilledema, and significant neuroimaging findings). Any new motor weakness and any signs for increased intracranial pressure constitute higher grades, namely 3 and 4. Use of CARTOX app from Android and Apple Store is recommended. This is a useful bedside tool to readily remind to and grade the aforementioned five elements. Consensus-based treatment algorithms are also detailed in the app and continuously upgraded. Use of these grading scales has standardized neurotoxicity assessment. This leads to earlier detection and helps recognize differences in CAR T-cell products. Clinical assessments continue to remain important bedside tools and are superior to any ancillary tests. Additional easy bedside tests starting from baseline like ability to follow multistep sequential commands (e.g., clap hand × 5, pick up 3 crackers and give to the nurse) could pick up subtle early changes. It is unclear if these subtle findings and early personality changes do warrant early limited corticosteroid usage. In patients with learning disability, Cornell Assessment for Pediatric Delirium could be used as supplemental tool.

Clinical assessments covering the earlier domains are important and should be done before and daily after CAR T-cell infusion. For patients with known CNS tumor and prior existing neurologic diseases (preexisting cognitive impairments, prior stroke, heavily treated patients with extensive white matter changes on magnetic resonance imaging [MRI] and with “chemo brain”), neurologic consultation is recommended before infusion. Post–CAR T-cell infusion new strokes are mostly related to new cardiac arrhythmia from the systemic effects of CRS. CNS bleeding complications (parenchymal, subdural, epidural) can be seen because of significant thrombocytopenia, also from greater hemorrhagic stroke conversion, and hemorrhagic conversion of ICANS-related parenchymal changes. CNS bleeding is graded per CTCAE grading and felt not to constitute ICANS if related to coagulopathy post–CAR T-cell infusion.

Median peak incidence of ICANS is generally 4 to 9 days after infusion and could last between 5 to 12 days. Higher, grade 3–4 toxicity can vary 11% to 64%. Differences in product characteristics (costimulatory dominions, rate and peak of CAR T-cell expansion) and patient characteristics (tumor burden) influence the toxicity. Frequencies and toxicity grades seen in trials leading to approvals compared to standard-of-care real-world experience at tertiary centers are similar.

Motor weakness could be related to new strokes from cardiac arrythmia. Transitory focal motor weakness from endothelial dysfunction could occur similar to reversible vasoconstriction syndrome, although to date, no abnormal angiogram findings have been published. Electroencephalogram (EEG) findings (focal slowing), transcranial dopplers (TCD) findings (elevations in blood flow at peak ICANS compared to baseline), and positron emission tomography (PET) findings (hypometabolism) suggest endothelial and vascular basis for this weakness. Similar conditions, such as eclampsia with documented vascular abnormalities, suggest vascular basis. New motor weakness should be treated as high-grade neurotoxicity, with steroids. For new motor weakness, spine MRI should be performed, given a rare possibility of changes within the spinal cord.

Seizures seen could be convulsive (overt clinical manifestations tonic/clonic) or nonconvulsive (NCSE). In NCSE, patients have prominent mental status changes and have limited or no motor manifestations and require EEG for diagnosis.

Differential diagnosis of CAR T-cell encephalopathy (ICE) are detailed in Table 36.5 .

Evaluation

Bedside clinical assessment tools remain the key tools with nurses at bedside, being a very integral part of the caring team. Care of these patients in designated hospital floors where nurses have completed adequate training helps in early recognition of clinical changes along with earlier transfer to higher care, intensive care unit (ICU), and earlier initiation of treatments.

Clinical assessment of five domains, ICANS including ICE score are detailed earlier. Useful ancillary testing include neuroimaging. Computed tomography (CT) scan while readily accomplished and could be repeated more frequently is useful to rule out cerebral edema, bleeding complications. Early edema on CT scan is hard to delineate in younger patients, unless there is preinfusion comparative scan. MRI provides more detailed findings (smaller strokes, signal abnormalities, Figs. 36.2 and 36.3 ). Hemorrhages could be petechial, large parenchymal, subdural, epidural, hemorrhagic transformation of strokes, hemorrhagic transformation of ICANS related changes. Typical posterior reversible encephalopathy syndrome (PRES) and atypical PRES findings have been detailed in recent publications. These changes likely reflect vascular/endothelial dysfunction. Typical PRES involves changes in the posterior regions of the brain (occipital, parietooccipital, or lesser frequent frontal). Atypical PRES could involve deeper structures like brain stem, thalamus substantia nigra, cerebellum, corpus callosum (see Fig. 36.3 ). The reversible nature of these findings with mostly complete clinical improvement are in keeping with what we know in patients with typical PRES seen in non CAR T-cell conditions. Strokes, could be spontaneous and in patients with cardiac arrythmia. These strokes can be small and clinically silent, and other reasons for mental status changes need to be pursued if there is no motor weakness. NCSE is seen as well in some of these small stroke volume patients to explain the mental status changes related to ICANS rather then the small stroke. Other MRI findings include cerebral venous thrombosis, meningeal enhancement, and encephalitis type imaging patterns. Lab data (paraneoplastic antibodies, CSF oligoclonal bands) and clinical presentation do not support autoimmunity.

Rapid- and severe-onset ICANS and sudden significant Δ change (ICE score > 6, ICANS > 2) would require more frequent clinical assessment and ancillary tests (neuroimaging, EEG, fundus examination, CSF analysis when safe).

EEG at bedside remains a very useful practical tool. Experience at multiple centers proved the utility of EEG. It is recommended to have at least a 30- minute study and, depending on hospital resources and patient’s condition, prolonged continuous EEG recording could be considered. This is a paramount test to identify NCSE, with patients having mental status changes and no overt clinical motor manifestations. NCSE presentation mimics delirium or toxic/metabolic encephalopathy. Rapid diagnosis and treatment of NCSE with benzodiazepine can result in dramatic mental status improvement ( Figs. 36.4 and 36.5 ). Some products (axicabtagene ciloleucel) that are associated with greater frequency and higher-grade ICANS show a higher incidence of EEG findings and would need more frequent EEG evaluation. EEG abnormalities include focal EEG abnormalities (focal slowing matching PET hypometabolism and focal neurologic findings), global EEG slowing, focal or generalized epileptiform discharges, and frontal rhythmic slowing. Features of toxic metabolic encephalopathy are seen like periodic waves (older terminology triphasic waves). The degree of slowing seen (mild, moderate, and severe) and additional aforementioned EEG findings (epileptiform discharges) could match the clinical grading ( Fig. 36.6 ). Higher clinical grades are associated with greater EEG abnormalities (presence of NCSE, epileptiform discharges, moderate to severe degree background slowing). Clinical improvements also match resolution of the aforementioned EEG findings in the author’s experience, representing transitory neuronal dysfunction related to cytokine and endothelial dysfunction.

Transcranial Doppler and Positron Emission Tomography Scan

Depending on the hospital resources, these could provide additional information. TCD can show an increase in blood flow velocities but would need daily studies starting from CAR T-cell infusion day, and changes likely reflect endothelial dysfunction. PET scan findings include focal hypometabolism. Cerebral vessel studies such as CT angiogram and MR angiogram at peak of higher-grade ICANS with focal clinical and EEG findings failed to demonstrate obvious narrowing of cerebral vasculature.