The role of neuroimaging in neurotoxicity after chimeric antigen receptor T-cell therapy

Abstract

Chimeric antigen receptor T-cell (CAR-T) therapy has become an established treatment for several haematological malignancies in relapse and is being evaluated for new indications. An important clinical challenge associated with the use of CAR-T therapy, however, is the common development of neurotoxicity. Different neurotoxicity syndromes have been reported. The best-known form of CAR-T neurotoxicity is immune effector cell-associated neurotoxicity syndrome, which can be associated with various findings on magnetic resonance imaging (MRI), including cerebral oedema and leptomeningeal enhancement. Other manifestations of neurotoxicity include movement disorders, myelopathy, cranial nerve palsies and ischaemic strokes. MRI plays a key role in the diagnosis and management of patients with suspected neurotoxicity. It can be used to support the diagnosis, exclude differential diagnoses and forms part of the grading of neurotoxicity. Other roles for MRI after CAR-T therapy include assessment of potential longer-term effects of therapy and neurotoxicity, and the evaluation of patients with emerging indications for CAR-T therapy. We recommend performing a baseline MRI brain prior to CAR-T therapy where feasible, as this greatly aids in the interpretation of neuroimaging findings. Here, we discuss the established and potential roles of neuroimaging in the context of neurotoxicity secondary to CAR-T therapy.

Plain language summary

Neuroimaging in CAR-T neurotoxicity

Chimeric antigen receptor T-cell therapy is a promising treatment for some cancers and autoimmune disorders, but side effects affecting the brain and spinal cord are a common challenge. Imaging, in particular Magnetic Resonance Imaging (MRI), plays a key role in the care of these patients. This review discusses the established and potential roles of imaging in this context, including assisting with the diagnosis, assessing severity, excluding alternate possibilities and evaluating for possible longer-term effects of treatment.

Keywords

CAR-T ICANS MRI neurotoxicity

Introduction

Chimeric antigen receptor T-cell (CAR-T) therapy has become an established treatment for several haematological malignancies in relapse, including lymphoma, acute lymphoblastic leukaemia and multiple myeloma. It is also being evaluated in some solid tumours and autoimmune diseases. An important clinical challenge associated with the use of CAR-T therapy, however, is the common development of neurotoxicity. Neurotoxicity has been reported in more than half of the patients receiving CAR-T therapy in some series.¹

Several different types of neurotoxicity secondary to CAR-T therapy have been reported.² The best-known is immune effector cell-associated neurotoxicity syndrome (ICANS), which is defined by the American Society for Transplantation and Cellular Therapy (ASTCT) as ‘a disorder characterised by a pathologic process involving the central nervous system following any immune therapy that results in the activation or engagement of endogenous or infused T cells and/or other immune effector cells’.³ The pathophysiology of ICANS is not fully understood, but endothelial disruption with subsequent blood–brain barrier disruption have been implicated.⁴ The frequency and severity of ICANS are linked to the specific product being used and ICANS appears to be more common in products associated with brisker CAR-T expansion, due to a higher peak level of CAR-T cells.⁵ The expansion kinetics are a product of CAR-T design, including the co-stimulatory and binding domains, antigen burden and prophylactic measures taken in the clinic such as the early use of corticosteroids.⁶ A range of presentations of ICANS have been described, including delirium, encephalopathy, aphasia, tremor and seizures.³

ICANS usually manifests around 3–6 days after CAR-T infusion, peaking around day 7 or 8, and then resolving over the next 1–2 weeks.¹ In rare cases, ICANS can be fatal. There is also a range of non-ICANS manifestations of neurotoxicity, including movement disorders, cranial nerve palsies, ischaemic strokes, myelopathy, peripheral neuropathy and Guillain–Barré syndrome.² Their typical time of onset varies, and they can occur at a similar time to ICANS, or weeks or months later.² Neurotoxicity (acute toxic leukoencephalopathy) can also occur secondary to fludarabine used for lymphodepletion prior to CAR-T administration, typically occurring 20–60 days after the final dose.² The relative frequency of the different neurotoxicity syndromes varies depending on the CAR-T agent. For example, movement and neurocognitive toxicity (discussed further below) appears to be specifically associated with B cell maturation antigen (BCMA) targeting CAR-T therapy for multiple myeloma.² Cerebellar toxicity has recently been reported in association with GPRC5D-targeted CAR-T therapy for multiple myeloma and appears to be on-target, off-tumour effect, related to expression of GPRC5D in the inferior olivary nucleus.⁷ CAR-T therapy aims to target antigens specifically expressed by cancer cells, but on-target, off-tumour toxicity can occur when the infused T-cells also attack healthy tissues that express the same target antigen.⁸

Magnetic resonance imaging (MRI) plays an important role in the diagnosis and management of patients with suspected neurotoxicity. It can be used to support the diagnosis, exclude differential diagnoses and forms part of the grading of neurotoxicity. In this narrative review, we discuss the established and potential roles of neuroimaging in the context of neurotoxicity secondary to CAR-T therapy, across various domains. This includes neuroimaging findings associated with various forms of CAR-T neurotoxicity, tips for distinguishing between neurotoxicity and relevant differential diagnoses, considerations for an appropriate MRI protocol, possible areas worthy of future investigation and considerations relevant to emerging indications for CAR-T therapy. While we have been guided by targeted PubMed literature searches where possible (using search terms such as ‘CAR-T’, ‘neurotoxicity’, ‘ICANS’ and ‘MRI’), there is a paucity of published literature on many of the topics we cover. In such areas, we have also drawn on our multi-disciplinary experience from our service over the last 6 years.

Indications for neuroimaging

Neuroimaging is typically prompted by neurological deterioration after CAR-T therapy. MRI will be prioritised in patients where the clinical symptoms are severe, atypical for ICANS, rapidly progressive, or where they follow an atypical time course for ICANS, so that alternative diagnoses need consideration. Neuroimaging is usually negative in patients with less-severe ICANS⁹ and may not be required when there is a confident clinical diagnosis based on the manifestations and time course of symptoms. MRI of the brain will generally be the first step, though MRI of the spine should be considered particularly when symptoms are suggestive of long tract involvement or where there is back pain.^2,10 Subsequent investigations, including follow-up imaging, should be tailored to the individual patient, based on the specific imaging findings, the level of confidence regarding the cause of the patient’s deterioration and the degree of clinical improvement. Lumbar puncture is usually indicated after neuroimaging if the cause of the patient’s deterioration remains unclear, or if imaging findings or clinical symptoms suggest infection as a differential diagnosis. While data are currently limited, neuroimaging may also have a role in understanding the late effects of cellular therapies.

Neuroimaging techniques

As would be expected, MRI is much better at identifying changes related to neurotoxicity than CT (computed tomography).⁹ CT will generally only be able to detect more florid changes, such as diffuse cerebral oedema or larger areas of haemorrhage or infarction. Therefore, CT should generally be reserved for patients who are severely unwell and not suitable for MRI, or when MRI is contraindicated. In all other settings, MRI is the neuroimaging modality of choice and therefore is the focus of our discussion. Fluorine-18-fluorodeoxyglucose positron-emission tomography (FDG-PET) does not have a specific role in the evaluation of ICANS acutely, but can provide information in patients who develop tremor as a subacute complication of CAR-T therapy,^11–15 which is discussed further below.

Findings of neurotoxicity on MRI

Several different clinical neurotoxicity syndromes have been described after CAR-T therapy.² In line with this, a wide variety of neuroimaging findings have been described.^9,16 These can be grouped both anatomically and pathologically.⁹ Practically, any part of the CNS can be affected, including the white matter, grey matter (including deep grey matter and cortex), leptomeninges, spinal cord, cranial nerves and cauda equina^2,9 (Figure 1). Similarly, a variety of pathological processes can be observed, including cerebral oedema, ischaemia, vascular changes, haemorrhages and evidence of inflammation⁹ (Figure 2). A summary of the MRI findings associated with different CAR-T neurotoxicity syndromes is provided in Table 1. A range of mechanisms for these imaging findings have been postulated, depending on the specific imaging finding.¹⁶ Such possible mechanisms including compromise of the blood–brain barrier (e.g. leptomeningeal enhancement),¹⁶ demyelination (e.g. hyperintensity in the supratentorial white matter and cerebellum on T2-weighted fluid-attenuated inversion recovery (T2-FLAIR) imaging),¹⁶ endothelial activation (e.g. haemorrhages)^17,18 and an excitotoxic effect (e.g. cortical diffusion restriction).^16,19 Given that it is not uncommon to observe a combination of imaging findings, it is likely that more than one mechanism may be occurring in any given patient, especially in more severe cases.

Figure 1.

Serial T2-FLAIR images at the level of the lateral ventricles performed prior to CD19 CAR-T therapy for diffuse large B-cell lymphoma (a), at the time of ICANS (b), 2 weeks later (c) and a further 3 weeks later (d). Extensive FLAIR hyperintensity in the white matter (b) is consistent with diffuse oedema secondary to ICANS in the absence of similar changes at baseline (a). This includes involvement of both external capsules (arrows) and subtle FLAIR hyperintensity in both thalami (arrowheads). The oedema subsequently recedes (c) and resolves (d). However, the lateral ventricles have visible enlarged over this period (d, asterisks), implying associated white matter volume loss.

Figure 2.

The axial T2-FLAIR image (left) in this patient with clinical features of ICANS after CD19 CAR-T therapy for transformed marginal zone lymphoma demonstrates diffuse white matter oedema, consistent with grade 4 ICANS. The SWI image (right) also demonstrates multiple microhaemorrhages in a similar distribution, considered to represent a manifestation of ICANS in this patient.

Table 1.

Summary of MRI findings in various CAR-T neurotoxicity syndromes.

Neurotoxicity syndrome	Neuroimaging findings
ICANS	Cerebral oedema
	White matter (periventricular, external capsules, corpus callosum)
	Grey matter (basal ganglia, thalami, brainstem, and cerebral cortex)
	Leptomeningeal FLAIR hyperintensity and/or enhancement
	Other findings (e.g., haemorrhages, vascular changes) may also occur
Myelopathy	T2 hyperintensity ± diffusion restriction in spinal cord
	Often associated with changes of ICANS intracranially
Ischaemia	Diffusion restriction (in absence of other ICANS features)
Parkinsonism	FLAIR hyperintensity ± diffusion restriction in basal ganglia
	Hypometabolism in basal ganglia and/or cerebral cortex on FDG-PET
Cranial nerve involvement	Most commonly facial nerves (unilateral or bilateral)

CAR-T, Chimeric antigen receptor T-cell; FLAIR, fluid-attenuated inversion recovery.

Due to these varied manifestations and postulated mechanisms, there is no single pathognomic appearance of CAR-T neurotoxicity on MRI. Herein we touch on both the spectrum of appearances that occur with ICANS, as well as neuroimaging findings that may be observed with other forms of neurotoxicity after CAR-T therapy.

Neuroimaging findings in ICANS

The most common MRI manifestation of ICANS is areas of hyperintensity on T2-FLAIR imaging, which may affect white matter and/or grey matter (see Figures 1 and 2), and may be focal or diffuse.⁹ A pattern commonly described in severe ICANS consists of bilateral involvement of the basal ganglia, thalami, external capsules and/or brainstem (see Figure 1).^16,17,20 FLAIR hyperintensity in the periventricular white matter may be seen in conjunction with involvement of deeper structures as noted above (see Figure 1), or in isolation.^16,20 Transient FLAIR hyperintensity involving the splenium of the corpus callosum has also been described, in some cases occurring together with diffusion restriction.^20–22 Leptomeningeal FLAIR hyperintensity and/or enhancement are also commonly observed as a feature of ICANS as opposed to a different type of neurotoxicity.^{16,17,23–25} Haemorrhages, myelopathy or areas of diffusion restriction (indicative of ischaemia) may occur through other mechanisms, but often occur together with the more typical imaging features of ICANS, and can be a consequence of severe ICANS¹⁰ (see Figures 2 and 4). From an imaging perspective, it is more practical to consider such additional acute neuroimaging changes as being compatible with a diagnosis of ICANS in the appropriate clinical setting.

Movement and neurocognitive toxicity

A subacute-onset neurotoxicity with Parkinsonian features, labelled ‘movement and neurocognitive toxicity’, is a well-recognised complication after BCMA CAR-T therapy.^2,11–15,26 It is postulated to represent an off-tumour effect of infused CAR-T cells against BCMA-expressing cells in the basal ganglia, as localised RNA expression of BCMA has been demonstrated in the basal ganglia of normal brains^13,14. There are notable differences compared to idiopathic Parkinson’s disease, however, such as a lack of response to levodopa.^2,13,15 In most reported cases, MRI has not shown any additional abnormalities,^12,13,15 though two cases with new MRI abnormalities have been reported. Gudera et al reported FLAIR hyperintensity in the caudate head, putamen and globus pallidus bilaterally, which resolved on later imaging.¹⁴ Similarly, Couturier et al reported one case with asymmetric globus pallidus FLAIR hyperintensity and diffusion restriction.¹⁵ Hypometabolism in the basal ganglia and/or cerebral cortex on FDG-PET has been described by several authors,^13–15,26 though one report has described increased (rather than decreased) FDG uptake in the caudate nuclei, putamina and thalami.¹¹ Outside the context of CAR-T therapy, the ‘swallow tail’ sign on SWI has been a useful MRI feature for the diagnosis of idiopathic Parkinson’s disease.^27,28 Studies reporting movement and neurocognitive toxicity after CAR-T therapy have not commented on this feature and it is worthy of investigation, though one would expect this sign to be negative, given the different pathogenesis.

Other neurotoxicity syndromes

There is also a range of neuroimaging findings associated with other CAR-T neurotoxicity syndromes.² Cranial nerve involvement usually occurs after BCMA CAR-T therapy for multiple myeloma and most commonly affects the facial nerves, either unilaterally or bilaterally² (Figure 3). Ischaemic strokes, myelopathy (Figure 4) and Guillain–Barré syndrome are other possible manifestations of CAR-T neurotoxicity, which are considered distinct from ICANS if other features of ICANS are absent²; the imaging appearances are equivalent to those occurring outside the context of CAR-T therapy.

Figure 3.

Axial post-contrast T1WI in a patient presenting with facial palsy after BCMA targeted CAR-T therapy for multiple myeloma, demonstrating bilateral facial nerve enhancement (arrowheads) involving the canalicular segments and geniculate ganglia.

Figure 4.

Sagittal MRI images of the cervical spine in a patient with symptomatic myelopathy after CD19 CAR-T therapy for diffuse large B-cell lymphoma, occurring together with grade 4 ICANS. The T2-weighted image (left) shows extensive T2-hyperintensity within the cervical spinal cord (arrowheads), which was also present throughout the thoracic spinal cord (not shown). The lack of enhancement on post-contrast T1WI with fat saturation (right) supports that the changes are related to CAR-T therapy rather than an alternate cause such as spinal cord involvement by lymphoma.

Evolution of MRI changes

The MRI findings of neurotoxicity are often considered to be transient, but this depends on the specific MRI changes. In general, the typical evolution of MRI findings in neurotoxicity is not well documented,²⁹ as repeat imaging may not be specifically indicated if the patient follows the expected pattern of improvement. A follow-up MRI would predominantly be performed if clinical findings persist for longer than would be expected or the patient deteriorates.

Oedema is expected to resolve over the course of a few weeks (see Figure 1). FLAIR hyperintensity affecting the splenium of the corpus callosum, with or without diffusion restriction, has also been reported to be transient.^20,22 Leptomeningeal enhancement is another feature that is expected to resolve,²⁵ provided that it is indeed related to ICANS. On the other hand, ischaemic changes (manifest as diffusion restriction) may progress to infarction and subsequently evolve accordingly, as occurs outside the context of ICANS. This may manifest as enhancement if imaging is performed in the subacute phase, and then later evolve to encephalomalacia, sometimes also demonstrating evidence of cortical laminar necrosis.¹⁷ At the other end of the spectrum, parenchymal haemorrhages can be expected to persist in some form, based on the location and size of haemorrhage. Microhaemorrhages are likely to appear similar at follow-up, subarachnoid haemorrhage may redistribute, potentially with a residuum of superficial siderosis, while lobar haemorrhages would be expected to partially contract.

Grading of ICANS

Presently, ICANS is most commonly graded according to the ASTCT guideline using a 4-grade system.³ Abnormal brain MRI findings are most common in higher grades of ICANS (grades 3–4),^9,24 but usually absent in grades 1–2 ICANS.^9,20 Importantly, MRI findings are included within the ICANS grading system.³ Specifically, focal cerebral oedema on neuroimaging indicates at least grade 3 ICANS, while diffuse cerebral oedema is evaluated as grade 4 ICANS³ (see Figures 1 and 2). As such, MRI has the potential to identify a higher grade of ICANS independent from clinical features.

Of the MRI findings that can be observed with neurotoxicity, as described above, only cerebral oedema is included within the ASTCT grading of ICANS.³ The ASTCT statement explicitly excludes intracranial haemorrhage from ICANS grading, as it can instead be due to a coagulopathy.³ Notably, cerebral oedema is not defined, other than for distinguishing between focal and diffuse oedema.³ For example, it does not address a potential distinction between involvement of white matter versus grey matter.³ We would expect that white matter oedema is included, but this is less clear for areas of FLAIR hyperintensity involving the grey matter, which is a common neuroimaging manifestation of ICANS.^20,24,30 While FLAIR hyperintensity in central structures such as the basal ganglia and thalami can have various causes,³¹ our practice is to include this appearance within the spectrum of cerebral oedema that would inform ICANS grading, assuming there is no better explanation. It is also unclear whether diffusion restriction, reflecting cytotoxic oedema due to ischaemia, is allowed within the definition. Clarifying the definition of oedema in future iterations will be important to optimise ICANS grading.

Differential diagnoses

In addition to assisting with grading of ICANS and prognostication, MRI is valuable for distinguishing between ICANS (or other forms of neurotoxicity) and relevant differential diagnoses. These relevant differential diagnoses typically comprise CNS involvement by the patient’s malignancy and CNS infection. An imaging feature that is common to all three differential diagnoses is the presence of leptomeningeal enhancement, which can make the distinction between these three processes challenging. It is important for radiologists to be aware that leptomeningeal enhancement is a common manifestation of ICANS, and it should not be considered diagnostic of CNS disease or infection. If other MRI features that are more typical of ICANS are observed together with leptomeningeal enhancement, this suggests that the leptomeningeal enhancement is also likely to represent a manifestation of ICANS. Additionally, leptomeningeal enhancement related to ICANS is expected to resolve over a relatively short interval; thus, follow-up imaging can help confirm the diagnosis.

Enhancement of cranial and/or spinal nerves may also mimic leptomeningeal carcinomatosis² (see Figure 3). In one series of 21 patients who developed cranial nerve palsy after BCMA CAR-T, the median time of onset was 22 days,³² a little later than is typical of ICANS.¹ No MRI findings other than cranial nerve enhancement were reported in the above series.³² Therefore, when cranial nerve enhancement is observed, features that would suggest a diagnosis other than neurotoxicity (in particular leptomeningeal carcinomatosis) include the presence of other MRI features of leptomeningeal disease in addition to cranial nerve enhancement, involvement of cranial nerves other than the facial nerves, or changes occurring after CD19 CAR-T therapy. Ependymal enhancement has not been reported as a feature of ICANS to our knowledge and would also suggest an alternate diagnosis, in particular progression of the patient’s haematologic malignancy (Figure 5) or an infective ventriculitis.

Figure 5.

Post-contrast T1WI in a patient 7 weeks after CD19 CAR-T therapy for diffuse large B-cell lymphoma, demonstrating new ependymal enhancement along the temporal horn of the right lateral ventricle, which suggests a diagnosis other than CAR-T neurotoxicity. The diagnosis of leptomeningeal involvement by lymphoma was confirmed by lumbar puncture.

Areas of diffusion restriction without enhancement will generally be most in keeping with neurotoxicity, but such an appearance can also occasionally occur with intracranial involvement by haematologic malignancies (though enhancement is observed in most cases). If there is doubt, a short-interval follow-up MRI should clarify the diagnosis – the diffusion restriction would be expected to recede if due to infarction (as a manifestation of neurotoxicity), but would persist or progress if due to malignancy. Intra-axial enhancing lesions will generally be suggestive of tumour, but there are some caveats. First, knowledge of the patient’s underlying malignancy will be relevant. For example, parenchymal lesions are a common manifestation of secondary CNS involvement by systemic lymphoma,^33,34 but are rare in multiple myeloma.^34,35 An alternative explanation for intra-axial enhancement could be subacute infarction, especially if MRI is performed in the subacute phase, after the peak of the patient’s clinical syndrome, perhaps because MRI had not been feasible earlier due to the patient’s clinical condition.

In patients with features of a myelopathy, infective myelitis, especially viral, is an important possible differential diagnosis^10,36 and warrants appropriate investigation. Most reported cases of myelopathy as a form of neurotoxicity have occurred together with or shortly after clinical manifestation of severe ICANS,¹⁰ thus the absence of clinical or imaging features of ICANS should raise concern.

MRI sequences in ICANS

The literature on the MRI assessment of CAR-T neurotoxicity has focused on the MRI findings, and there is little information on the optimal MRI protocol. There is no single consensus protocol, as, for example, has been published for clinical trials in intracranial metastases and gliomas.^37,38 To our knowledge, direct comparison between specific sequences and imaging parameters has not been performed in the context of CAR-T neurotoxicity. As such, we do not aim to recommend a specific imaging protocol, but rather discuss the relative merits of these options based on an understanding of their utility in related disease processes and our own experience. As is the case for MRI in any context, the MRI sequences used in the investigation of neurotoxicity should be targeted to the clinical questions, possible imaging findings and relevant differential diagnoses. A consideration of possible longer-term sequelae is also worthwhile, as well as potential future research questions. Given the wide variety of possible imaging manifestations of neurotoxicity, the MRI protocol should be comprehensive. This discussion focuses on ICANS, but the imaging protocol may be varied if other types of neurotoxicity are suspected, and depending on factors such as the clinical status of the patient.

The T2-FLAIR sequence, to detect changes such as oedema, is key in the evaluation of ICANS. Given its importance, we suggest it be performed first, in case the patient is unable to complete a full examination. Diffusion-weighted imaging (DWI) is critical for detecting ischaemic changes, while a susceptibility-sensitive sequence such as susceptibility-weighted imaging (SWI) is important to detect haemorrhages. Pre- and post-contrast T1-weighted imaging (T1WI) is important for identifying findings such as leptomeningeal enhancement and excluding differential diagnoses such as intracranial involvement by the patient’s malignancy. We recommend that 3-dimensional (3D) T1WI be used, typically with 1 mm isotropic resolution. Pre-contrast T1WI should match the chosen post-contrast (C+) T1WI sequence (or at least one of these, if more than one C+ T1WI sequence is performed).³⁷ Standard axial T2-weighted imaging will typically be included and is a relatively fast sequence; we suggest performing this between contrast administration and C+ T1WI, as it is not affected by contrast administration.³⁷ A delay of approximately 4–8 min between contrast administration and performing C+ T1WI is recommended to optimally assess enhancement,^37,39 so while T2-weighted imaging may provide little additional information compared to the FLAIR sequence in this context, it does not add to the overall length of the scan. If additional time is necessary between administering contrast and C+ T1WI (if the T2 sequence is relatively quick), SWI can also be performed post-contrast without loss of information.⁴⁰ Identification of vascular changes (such as stenoses) in some previous reports^41–43 suggests value in performed MR angiography (MRA), though in our experience, we have found it to provide less additional value than the other sequences listed. The above sequences and possible imaging findings are summarised in Table 2.

Table 2.

MRI techniques to consider for inclusion in the MRI protocol of patients with suspected CAR-T neurotoxicity and possible imaging findings.

MRI technique	Possible findings in neurotoxicity
FLAIR	Focal or diffuse cerebral oedema, focal areas of FLAIR hyperintensity (e.g., basal ganglia, thalami)
DWI	Ischaemia, transient splenial lesions
SWI	Microhaemorrhages, subarachnoid or lobar haemorrhage
C+ T1WI	Leptomeningeal enhancement, cranial nerve enhancement, parenchymal volume loss (3D-GRE)
C+ FLAIR	Leptomeningeal enhancement
MRA	Vascular changes
MRI spine	T2-hyperintensity in the spinal cord, cauda equina enhancement

FLAIR, fluid-attenuated inversion recovery.

There are different options for the post-contrast sequence(s), with different strengths and weaknesses. Three-dimensional (3D) magnetisation prepared gradient recalled echo (3D-GRE) C+ T1WI provides better grey-white matter differentiation than 3D turbo spin echo (3D-TSE) T1WI,⁴⁴ which facilitates the assessment of grey and white matter volume loss as a possible long-term sequela.⁴⁵ On the other hand, 3D-TSE is superior to 3D-GRE for the assessment of leptomeningeal enhancement^46,47 (Figure 6), as is C+ T2-FLAIR.^47–49 While the identification of leptomeningeal enhancement does not particularly distinguish between ICANS and key differential diagnoses such as leptomeningeal carcinomatosis and infection, the sequences should be sufficiently sensitive that leptomeningeal enhancement can be confidently identified or excluded. We suggest that C+ 3D-GRE alone is suboptimal for this and should be supplemented with (or replaced by) C+ 3D-TSE and/or C+ T2-FLAIR. In our practice, we typically include at least two post-contrast sequences where feasible. It is generally unclear which post-contrast sequences have been performed in literature reports of MRI findings in ICANS and the relative frequency of leptomeningeal enhancement as a manifestation of ICANS may be underestimated if this has been predominantly 3D-GRE T1WI.

Figure 6.

The axial post-contrast FLAIR image (left) in this patient with ICANS after CD19 CAR-T therapy for B-cell lymphoma demonstrates enhancement within several cerebral sulci, predominantly on the left. However, this enhancement is not identified with confidence on the post-contrast 3D-GRE image (right).

It is worth noting, however, that the patients for whom MRI is most valuable, namely those with higher grades of ICANS, may have difficulty staying still for a lengthy MRI examination. There are also various risks and challenges associated with performing MRI in critically unwell patients,⁵⁰ so it is prudent to decrease the duration of the scan, where this can be done without compromising its diagnostic value. A variety of techniques can be employed to decrease the duration of individual sequences.⁵¹ A highly targeted MRI protocol, as can be performed in settings such as ischaemic stroke,⁵² is more challenging in the setting of suspected neurotoxicity due to the wide variety of possible imaging findings, but some rationalisation of sequences is possible. For example, while 3D-T2-FLAIR has value, standard axial (or coronal) T2-FLAIR will often be adequate. Additionally, there is less need to assess parenchymal volume loss on 3D-GRE acutely, provided that at least one post-contrast sequence (3D-TSE or T2-FLAIR) is performed. It may also be feasible to omit MRA in patients without evidence of ischaemia on DWI.

The MRI protocol will also vary depending on the patient’s clinical presentation, including if other types of neurotoxicity than ICANS are suspected. For example, suspicion of myelopathy or Guillain–Barré syndrome will prompt an MRI of the spine, which should include sagittal T2 and pre- and post-contrast T1WI at a minimum. DWI of the spinal cord may also be performed if available, as spinal cord infarction is a possible explanation for the incomplete recovery seen in some patients with myelopathy after CAR-T therapy.¹⁰ Sequences focusing on the cranial nerves can be added in patients with clinical cranial nerve involvement. High-resolution SWI has the potential to help distinguish ‘movement and neurocognitive toxicity’ from idiopathic Parkinson’s disease, as discussed above, though literature on this is currently lacking.

Baseline MRI

Some MRI findings are quite typical of ICANS in the appropriate context, especially when severe. For some changes, however, it can be challenging to determine whether the changes reflect ICANS or are pre-existing if baseline pre-CAR-T neuroimaging has not been performed.⁵³ In general, pre-existing changes can be grouped into those related to previous treatments (with or without underlying CNS involvement) and those unrelated to the patient’s cancer (most commonly ischaemic). This is particularly the case for FLAIR hyperintensity in the periventricular white matter (Figure 7), as this is a common site for both treatment-related changes (e.g. after intrathecal methotrexate or whole-brain radiotherapy³⁴) and ischaemic changes. Importantly, most patients receiving CAR-T therapy have already received several previous lines of therapy, thus post-treatment changes are not uncommon. When there is uncertainty regarding the nature of white matter changes and pre-treatment baseline imaging is not available, subsequent imaging should clarify, as white matter changes secondary to ICANS are expected to recede, as discussed above, while pre-existing changes will persist.

Figure 7.

This axial 3D-FLAIR image performed as a baseline prior to CAR-T therapy shows FLAIR hyperintensities in the periventricular white matter and prominence of the lateral ventricles. If the patient subsequently underwent imaging for the investigation of ICANS without baseline imaging, these changes could be misinterpreted as white matter changes due to ICANS and hydrocephalus, respectively.

If feasible, we recommend performing a baseline brain MRI prior to CAR-T therapy as a routine. Where access is limited, patients with known CNS disease and/or those evaluated as being at higher risk of ICANS can be prioritised. We suggest an overall similar MRI protocol to what would be used in the context of ICANS, to allow the most accurate comparison if MRI is subsequently warranted due to suspicion of ICANS. The main exception is that MRA seems less important at baseline, as there would be little reason to expect significant vascular changes prior to CAR-T therapy.

Potential future roles of MRI

Being a new treatment, the potential longer-term sequelae of CAR-T therapy and neurotoxicity are unclear. Cognitive impairment is often observed in patients after CAR-T therapy.^54,55 However, there is currently little data on possible neuroimaging correlates, thus, this is an area worthy of further investigation. In many cases, there would be little indication for longer-term neuroimaging follow-up, especially if there was no CNS disease at baseline, thus, more chronic post-treatment sequelae would not be detected. We have observed that more severe MRI findings at the time of ICANS can be associated with visible brain parenchymal volume loss after a relatively short interval (see Figure 1). It seems unlikely that this would represent a ‘pseudoatrophy’ phenomenon as can occur in multiple sclerosis (MS) after commencing disease-modifying therapies,⁵⁶ as underlying brain inflammation would not be expected in these patients prior to commencing CAR-T therapy. Even without visible changes on standard MRI sequences, microstructural abnormalities have been reported using diffusion tensor imaging.²⁵ It seems distinctly possible that, even when ICANS does not cause visible MRI changes acutely, accelerated brain parenchymal volume loss could occur, as for example is commonly observed in MS.^57–60 Additionally, it is possible that changes such as leukoencephalopathy could occur down the track, similar to that commonly developing after treatments such as intrathecal methotrexate or whole-brain radiotherapy.³⁴

Emerging indications for CAR-T therapy

Building on its established role in haematological malignancies, CAR-T therapy is being investigated for use in solid organ malignancies and non-neoplastic autoimmune diseases. Of the two, initial results have been more promising in autoimmune diseases⁶¹ and there are many ongoing clinical trials.⁶² The possibility of ICANS is a particularly important clinical consideration when using CAR-T for autoimmune diseases, as the acceptable rate of toxicity may be lower, given that these diseases are not immediately life-threatening. Additionally, CAR-T therapy is not expected to reverse any accumulated deficits, and the goal is largely to stall the progression of the disease.⁶³ Fortunately, initial data suggest that the rates of ICANS when CAR-T is used for autoimmune diseases may be lower than when used in haematological malignancies.^61,62 Further research is required and neuroimaging retains an important role.

Treatments used for haematological malignancies often induce changes on MRI, such as leukoencephalopathy secondary to methotrexate.⁶⁴ In contrast, therapies used for the treatment of autoimmune diseases are less likely to cause visible post-treatment changes on MRI. However, some of the conditions that have been treated with CAR-T therapy specifically target the CNS, such as MS or neuromyelitis optica spectrum disorder, or often affect the CNS, such as systemic lupus erythematosus (SLE).^65,66 Indeed, it is important to recognise that CAR-T would typically be considered in patients with more severe disease refractory to conventional therapies, and such patients will inevitably have greater severity of CNS involvement than the average prevalent population. Performing baseline pre-CAR-T MRIs is essential in systemic autoimmune diseases involving the CNS. First, a pre-CAR-T baseline is important for distinguishing between neurotoxicity and progression of the patient’s autoimmune disease. Second, it allows evaluation of the effectiveness of CAR-T at treating disorders affecting the CNS. A parallel to this need can be drawn with the radiological appearance of progressive multifocal leukoencephalopathy and immune reconstitution inflammatory syndrome, which has occurred among patients treated for MS with natalizumab.⁶⁷ Of note, the development of a new T2-hyperintense spinal cord lesion has been reported in one patient treated with CAR-T for MS, and it was unclear whether this represented disease progression or was secondary to the treatment,⁶⁸ highlighting the importance of ongoing MRI follow-up to provide insights into the effects of this emerging treatment.

In terms of an imaging protocol tailored to autoimmune diseases, performing 3D-T2-FLAIR will be particularly important and this is already an established part of the imaging protocol in conditions such as MS.⁶⁹ While not necessary for haematologic malignancies, an MRA should be included at baseline for conditions which can be associated with a vasculitis, such as SLE.⁷⁰ Similarly, a baseline MRI of the spine would be important for conditions commonly affecting the spinal cord, such as MS, neuromyelitis optica and SLE,⁶⁸ to aid in the future distinction between progression of the autoimmune disease and a treatment-related effect.

Conclusion

Neuroimaging plays a key role important in the management of patients with suspected neurotoxicity secondary to CAR-T therapy. In the acute setting, it is important to support accurate diagnosis of neurotoxicity and grading of ICANS. While only some MRI features are pathognomonic of ICANS, it is practical to consider all acute neuroimaging changes as being compatible with a diagnosis of ICANS in the appropriate clinical setting. Obtaining baseline MRI prior to commencement of CAR-T therapy is important, at least for patients at high risk of ICANS, to help differentiate radiological appearances of neurotoxicity from pre-existing brain damage. Neuroimaging will also be valuable for evaluating potential long-term effects of CAR-T therapy and in identifying neurotoxicity in patients who receive CAR-T therapy for autoimmune disease of the CNS.

Footnotes

Acknowledgements

None.

Declarations

ORCID iDs

Arian Lasocki

Simon J. Harrison

Michael Dickinson

Tomas Kalincik

References

Gust

Ponce

Liles

, et al. Cytokines in CAR T cell-associated neurotoxicity. Front Immunol 2020; 11: 577027.

Graham

Velasco

Alarcon Tomas

, et al. Non-ICANS neurological complications after CAR T-cell therapies: recommendations from the EBMT Practice Harmonisation and Guidelines Committee. Lancet Oncol 2025; 26: e203–e213.

Lee

Santomasso

Locke

, et al. ASTCT Consensus grading for cytokine release syndrome and neurologic toxicity associated with immune effector cells. Biol Blood Marrow Transplant 2019; 25: 625–638.

Schubert

Schmitt

Wang

, et al. Side-effect management of chimeric antigen receptor (CAR) T-cell therapy. Ann Oncol 2021; 32: 34–48.

Jacobson

Munoz

Sun

, et al. Real-world outcomes with chimeric antigen receptor T Cell therapies in large B cell lymphoma: a systematic review and meta-analysis. Transplant Cell Ther 2024; 30: 77.e71–e77.e15.

Bachy

Le Gouill

Di Blasi

, et al. A real-world comparison of tisagenlecleucel and axicabtagene ciloleucel CAR T cells in relapsed or refractory diffuse large B cell lymphoma. Nat Med 2022; 28: 2145–2154.

Mailankody

Devlin

Landa

, et al. GPRC5D-targeted CAR T cells for myeloma. N Engl J Med 2022; 387: 1196–1206.

Sun

Hao

Yang

, et al. Immunotherapy with CAR-Modified T cells: toxicities and overcoming strategies. J Immunol Res 2018; 2018: 2386187.

Lapidus

Anderson

Harrison

, et al. Neuroimaging findings in immune effector cell associated neurotoxicity syndrome after chimeric antigen receptor T-cell therapy. Leuk Lymphoma 2022; 63: 2364–2374.

10.

Deschênes-Simard

Santomasso

Dahi

PB.

Clinical features, pathophysiology, and management of acute myelopathy following CAR T-cell therapy. Blood 2024; 144: 2083–2094.

11.

Aliakbar

Manouvakhova

Wong

, et al. Case report: treatment of parkinsonism secondary to ciltacabtagene autoleucel using a combination dopaminergic regimen. Front Immunol 2024; 15: 1444010.

12.

Graham

Lee

Wiggin

, et al. Chemotherapy-induced reversal of ciltacabtagene autoleucel-associated movement and neurocognitive toxicity. Blood 2023; 142: 1248–1252.

13.

Van Oekelen

Aleman

Upadhyaya

, et al. Neurocognitive and hypokinetic movement disorder with features of parkinsonism after BCMA-targeting CAR-T cell therapy. Nat Med 2021; 27: 2099–2103.

14.

Gudera

Baehring

Karschnia

Parkinsonism following chimeric antigen receptor T cell therapy. JAMA Neurol. Epub ahead of print 12 August 2024.

15.

Couturier

Escoffre

Leh

, et al. Parkinson-like neurotoxicity in female patients treated with idecabtagene-vicleucel. Hemasphere 2024; 8: e131.

16.

Pinto

Liu

Nelson

Jr. , et al. Neuroimaging of complications arising after CD19 chimeric antigen receptor T-cell therapy: a review. J Neuroimaging 2023; 33: 703–715.

17.

Gust

Hay

Hanafi

, et al. Endothelial activation and blood-brain barrier disruption in neurotoxicity after adoptive immunotherapy with CD19 CAR-T Cells. Cancer Discov 2017; 7: 1404–1419.

18.

Holtzman

Xie

Bentzen

, et al. Immune effector cell-associated neurotoxicity syndrome after chimeric antigen receptor T-cell therapy for lymphoma: predictive biomarkers and clinical outcomes. Neuro Oncol 2021; 23: 112–121.

19.

Gust

Annesley

Gardner

, et al. EEG correlates of delirium in children and young adults with CD19-directed CAR T cell treatment-related neurotoxicity. J Clin Neurophysiol 2021; 38: 135–142.

20.

Santomasso

Park

Salloum

, et al. Clinical and biological correlates of neurotoxicity associated with CAR T-cell therapy in patients with B-cell acute lymphoblastic leukemia. Cancer Discov 2018; 8: 958–971.

21.

Karschnia

Jordan

Forst

, et al. Clinical presentation, management, and biomarkers of neurotoxicity after adoptive immunotherapy with CAR T cells. Blood 2019; 133: 2212–2221.

22.

Lee

Kochenderfer

Stetler-Stevenson

, et al. T cells expressing CD19 chimeric antigen receptors for acute lymphoblastic leukaemia in children and young adults: a phase 1 dose-escalation trial. Lancet 2015; 385: 517–528.

23.

Shah

Johnson

Fenske

, et al. Intrathecal chemotherapy for management of steroid-refractory CAR T-cell-associated neurotoxicity syndrome. Blood Adv 2020; 4: 2119–2122.

24.

Strati

Nastoupil

Westin

, et al. Clinical and radiologic correlates of neurotoxicity after axicabtagene ciloleucel in large B-cell lymphoma. Blood Adv 2020; 4: 3943–3951.

25.

Tan

AP.

CAR T-cell therapy-related neurotoxicity in paediatric acute lymphocytic leukaemia. Pediatr Blood Cancer 2020; 67: e28635.

26.

Cohen

Parekh

Santomasso

, et al. Incidence and management of CAR-T neurotoxicity in patients with multiple myeloma treated with ciltacabtagene autoleucel in CARTITUDE studies. Blood Cancer J 2022; 12: 32.

27.

Chau

Todd

Wilcox

, et al. Diagnostic accuracy of the appearance of Nigrosome-1 on magnetic resonance imaging in Parkinson’s disease: a systematic review and meta-analysis. Parkinsonism Relat Disord 2020; 78: 12–20.

28.

Tseriotis

Eleftheriadou

Mavridis

, et al. Is the swallow tail sign a useful imaging biomarker in clinical neurology? A systematic review. Mov Disord Clin Pract 2025; 12: 134–147.

29.

Marino

Mader

Kardasis

, et al. Evolving magnetic resonance imaging (MRI) findings in immune effector cell-associated neurotoxicity syndrome. Cureus 2025; 17: e84407.

30.

Gust

Finney

, et al. Glial injury in neurotoxicity after pediatric CD19-directed chimeric antigen receptor T cell therapy. Ann Neurol 2019; 86: 42–54.

31.

Hegde

Mohan

Lath

, et al. Differential diagnosis for bilateral abnormalities of the basal ganglia and thalamus. Radiographics 2011; 31: 5–30.

32.

Van De Donk

NWCJ

Sidana

Schecter

, et al. Clinical experience with cranial nerve impairment in the CARTITUDE-1, CARTITUDE-2 Cohorts A, B, and C, and Cartitude-4 Studies of Ciltacabtagene Autoleucel (Cilta-cel). Blood 2023; 142: 3501–3501.

33.

Ghose

Elias

Guha

, et al. Influence of rituximab on central nervous system relapse in diffuse large B-Cell lymphoma and role of prophylaxis – a systematic review of prospective studies. Clin Lymphoma Myeloma Leuk 2015; 15: 451–457.

34.

Lasocki

Seymour

JF.

Central nervous system manifestations of systemic haematological malignancies and key differentials. Clin Radiol 2022; 77: 328–336.

35.

Lasocki

Gangatharan

Gaillard

, et al. Intracranial involvement by multiple myeloma. Clin Radiol 2015; 70: 890–897.

36.

Zhang

Wang

, et al. Case report: acute HHV6B encephalitis/myelitis post CAR-T cell therapy in patients with relapsed/refractory aggressive B-cell lymphoma. Front Neurol 2024; 15: 1334000.

37.

Kaufmann

Smits

Boxerman

, et al. Consensus recommendations for a standardized brain tumor imaging protocol for clinical trials in brain metastases. Neuro Oncol 2020; 22: 757–772.

38.

Ellingson

Bendszus

Boxerman

, et al. Consensus recommendations for a standardized Brain Tumor Imaging Protocol in clinical trials. Neuro Oncol 2015; 17: 1188–1198.

39.

Seo

Lim

Lee

, et al. Time optimization of gadobutrol-enhanced brain MRI for metastases and primary tumors using a dynamic contrast-enhanced imaging. BMC Med Imaging 2022; 22: 180.

40.

El-Koussy

Schenk

Kiefer

, et al. Susceptibility-weighted imaging of the brain: does gadolinium administration matter? Eur J Radiol 2012; 81: 272–276.

41.

Wang

Cao

, et al. Bilateral anterior cerebral artery occlusion following CD19- and BCMA-targeted chimeric antigen receptor T-cell therapy for a myeloma patient. Int J Hematol 2021; 114: 408–412.

42.

Beuchat

Danish

Rubin

, et al. EEG findings in CAR T-cell-associated neurotoxicity: Clinical and radiological correlations. Neuro Oncol 2022; 24: 313–325.

43.

Rubin

Danish

Ali

, et al. Neurological toxicities associated with chimeric antigen receptor T-cell therapy. Brain 2019; 142: 1334–1348.

44.

Danieli

Riccitelli

Distefano

, et al. Brain tumor-enhancement visualization and morphometric assessment: a comparison of MPRAGE, SPACE, and VIBE MRI techniques. AJNR Am J Neuroradiol 2019; 40: 1140–1148.

45.

Gil

Tabari

Nickel

, et al. Deep-learning-accelerated T1-MPRAGE MRI for quantification and visual grading of cerebral volume in memory loss patients. Radiol Adv 2025; 2: umaf022.

46.

Choi

Lee

, et al. Application of 3D fast spin-echo T1 black-blood imaging in the diagnosis and prognostic prediction of patients with leptomeningeal carcinomatosis. AJNR Am J Neuroradiol 2018; 39: 1453–1459.

47.

Jeevanandham

Kalyanpur

Gupta

, et al. Comparison of post-contrast 3D-T(1)-MPRAGE, 3D-T(1)-SPACE and 3D-T(2)-FLAIR MR images in evaluation of meningeal abnormalities at 3-T MRI. Br J Radiol 2017; 90: 20160834.

48.

Fukuoka

Hirai

Okuda

, et al. Comparison of the added value of contrast-enhanced 3D fluid-attenuated inversion recovery and magnetization-prepared rapid acquisition of gradient echo sequences in relation to conventional postcontrast T1-weighted images for the evaluation of leptomeningeal diseases at 3T. AJNR Am J Neuroradiol 2010; 31: 868–873.

49.

Kakeda

Korogi

Hiai

, et al. Pitfalls of 3D FLAIR brain imaging: a prospective comparison with 2D FLAIR. Acad Radiol 2012; 19: 1225–1232.

50.

McLean

Thompson

MRI and the critical care patient: clinical, operational, and financial challenges. Crit Care Res Pract 2023; 2023: 2772181.

51.

Eisenmenger

Peret

Roberts

, et al. Focused Abbreviated Survey MRI Protocols for Brain and Spine Imaging. Radiographics 2023; 43: e220147.

52.

Cortes-Albornoz

Clifford

, et al. A 3-minute ultrafast MRI and MRA protocol for screening of acute ischemic stroke. J Am Coll Radiol 2025; 22: 366–375.

53.

Kim

Yeom

Vasyliv

, et al. Brain MRI changes in children and young adults with B-cell acute lymphoblastic leukemia following chimeric antigen receptor T-cell therapy. Eur Radiol 2025

54.

Kennedy

Ahmed

Artz

, et al. Assessing cognitive function in transplantation and chimeric antigen receptor t cell therapy recipients: Expert recommendations from the survivorship, aging and biobehavioral special interest groups of the American Society for Transplantation and Cellular Therapy. Transplant Cell Ther 2025

55.

Kuznetsova

Rosenfeld

Sales

, et al. Cognitive impairment in hematology patients planned for chimeric antigen receptor T-cell therapy. Expert Rev Hematol 2025; 18(11): 987–997.

56.

De Stefano

Airas

Grigoriadis

, et al. Clinical relevance of brain volume measures in multiple sclerosis. CNS Drugs 2014; 28: 147–156.

57.

Kalkers

Vrenken

Uitdehaag

, et al. Brain atrophy in multiple sclerosis: impact of lesions and of damage of whole brain tissue. Mult Scler 2002; 8: 410–414.

58.

Mendelsohn

Pemberton

Gray

, et al. Commercial volumetric MRI reporting tools in multiple sclerosis: a systematic review of the evidence. Neuroradiology 2023; 65: 5–24.

59.

Khan

Bao

Shah

, et al. Effect of disease-modifying therapies on brain volume in relapsing-remitting multiple sclerosis: results of a five-year brain MRI study. J Neurol Sci 2012; 312: 7–12.

60.

Fox

Jenkins

Leary

, et al. Progressive cerebral atrophy in MS: a serial study using registered, volumetric MRI. Neurology 2000; 54: 807–812.

61.

Müller

Taubmann

Bucci

, et al. CD19 CAR T-cell therapy in autoimmune disease – a case series with follow-up. N Engl J Med 2024; 390: 687–700.

62.

Chinas

Alexopoulos

CAR T-cells meet autoimmune neurological diseases: a new dawn for therapy. Front Immunol 2025; 16: 1604174.

63.

Schett

Müller

Taubmann

, et al. Advancements and challenges in CAR T cell therapy in autoimmune diseases. Nat Rev Rheumatol 2024; 20: 531–544.

64.

Hassankhani

Bagley

Dahmoush

, et al. Role of neuroimaging in cancer-treatment neurotoxicity. AJR Am J Roentgenol 2025; 225(3): e2533139.

65.

Arinuma

Kikuchi

Wada

, et al. Brain MRI in patients with diffuse psychiatric/neuropsychological syndromes in systemic lupus erythematosus. Lupus Sci Med 2014; 1: e000050.

66.

Postal

Lapa

Reis

, et al. Magnetic resonance imaging in neuropsychiatric systemic lupus erythematosus: current state of the art and novel approaches. Lupus 2017; 26: 517–521.

67.

Honce

Nagae

Nyberg

Neuroimaging of natalizumab complications in multiple sclerosis: PML and other associated entities. Mult Scler Int 2015; 2015: 809252.

68.

Fischbach

Richter

Pfeffer

, et al. CD19-targeted chimeric antigen receptor T cell therapy in two patients with multiple sclerosis. Med 2024; 5: 550–558.e552.

69.

Wattjes

Ciccarelli

Reich

, et al. 2021 MAG NIMS-CMSC-NAIMS consensus recommendations on the use of MRI in patients with multiple sclerosis. Lancet Neurol 2021; 20: 653–670.

70.

Rodrigues

Galego

Costa

, et al. Central nervous system vasculitis in systemic lupus erythematosus: a case series report in a tertiary referral centre. Lupus 2017; 26: 1440–1447.