Case series review of neuroradiologic changes associated with immune checkpoint inhibitor therapy

Abstract

While immuno-oncotherapy (IO) has significantly improved outcomes in the treatment of systemic cancers, various neurological complications have accompanied these therapies. Treatment with immune checkpoint inhibitors (ICIs) risks multi-organ autoimmune inflammatory responses with gastrointestinal, dermatologic, and endocrine complications being the most common types of complications. Despite some evidence that these therapies are effective to treat central nervous system (CNS) tumors, there are a significant range of related neurological side effects due to ICIs. Neuroradiologic changes associated with ICIs are commonly misdiagnosed as progression and might limit treatment or otherwise impact patient care. Here, we provide a radiologic case series review restricted to neurological complications attributed to ICIs, anti-CTLA-4, and PD-L-1/PD-1 inhibitors. We report the first case series dedicated to the review of CNS/PNS radiologic changes secondary to ICI therapy in cancer patients. We provide a brief case synopsis with neuroimaging followed by an annotated review of the literature relevant to each case. We present a series of neuroradiological findings including nonspecific parenchymal and encephalitic, hypophyseal, neural (cranial and peripheral), meningeal, cavity-associated, and cranial osseous changes seen in association with the use of ICIs. Misdiagnosis of radiologic abnormalities secondary to neurological immune-related adverse events can impact patient treatment regimens and clinical outcomes. Rapid recognition of various neuroradiologic changes associated with ICI therapy can improve patient tolerance and adherence to cancer therapies.

Immuno-oncotherapies (IO), such as immune checkpoint inhibitors (ICIs), have demonstrated utility in treating some of the most common systemic malignancies that are associated with high risk for central nervous system (CNS) involvement including melanoma, renal cell, lung, and breast cancers.^1–4 For this reason, neuroradiologic imaging changes can be incorrectly presumed to be metastatic involvement of these systemic cancers in the brain and spinal cord. Immune checkpoints are immunomodulatory pathways that when inhibited lead to an enhanced immune response against tumor antigens. Currently, approved and commercially available ICIs target cytotoxic T-lymphocyte–associated antigen 4 (CTLA-4) (ie, ipilimumab), programmed cell death protein 1 (PD-1) (ie, nivolumab, pembrolizumab, and cemiplimab), and programmed cell death protein ligand 1 (PD-L1) (ie, atezolizumab, avelumab, and durvalumab). Other agents in development target lymphocyte activation gene-3 (LAG3), T-cell immunoglobulin and mucin-domain containing-3 (TIM3), and B- and T-cell lymphocyte attenuator (BTLA).^5–7 Immune-related adverse events (irAEs) are rare but can lead to significant morbidity and mortality. Common irAEs include pulmonary, gastrointestinal, and cutaneous manifestations.⁷ More serious complications include cardiologic, pulmonary, and neurotoxic reactions.⁷

This case series review elucidates the rare neurological irAEs with identifiable neuroradiologic features that commonly mimic brain or spinal cord metastases. This work emphasizes the importance of early surveillance and rapid recognition of neurological changes in patients treated with ICIs. The aim of this review series is to improve clinician identification and understanding of CNS/PNS irAEs.

Incidence of Neurological irAEs

The incidence of ICI-related neurotoxicity is reported as low as 1% but can be a fatal irAE.⁸ There are relatively few reports on the incidence of neurological irAEs. Cuzzubbo et al. evaluated 59 clinical trials (including 9208 patients) using ICIs and noted the incidence of neurological irAEs secondary to CTLA-4 inhibitors to be 3.8%, PD-1 inhibitors 6.1%, and CTLA-4 inhibitors combined with PD-1 inhibitors to be 12.0%.⁹ They noted most neurological AEs (55%) were low-grade (1/2) headaches or nonspecific complaints.⁹ The incidence of high-grade (3/4) neurological irAEs was highest in patients treated with anti-CTLA-4 therapy (0.7%), as compared to those treated with anti-PD-1 therapy (0.4%).⁹ Interestingly, in trials assessing the combination of both antibodies, the incidence of high-grade neurological irAEs is similar to anti-CTLA-4 therapy alone (0.7%).⁹

Time to Onset of Neurological irAEs

Cuzzubbo et al.⁹ analyzed 27 patients receiving ICI treatment who later developed nervous system irAEs at a median onset of 6 weeks from the start of therapy (range 1-74 weeks).¹⁰ For all cases, symptom onset was acute to subacute and related to tumor response.¹⁰ Spain et al. (2017)¹¹ and Zimmer et al. (2016)¹² reported that 80% and 75% of patients, respectively, developed their neurological irAEs within the first 4 months of immunotherapy, largely during induction phase of treatment. It is especially important to remain vigilant for neurological signs of irAEs within the first 4 months of therapy.

Mechanisms and Patterns of Neurological irAEs

The mechanism of neurotoxicity in ICIs remains to be clearly defined; however, it has been proposed that autoreactive T cells contribute to processes within endoneurial microvessels causing a microvasculopathy and inflammation within the perineural spaces.¹³ Patterns of neurotoxicity might be different depending on the ICIs used.¹⁴ Neuromuscular junction dysfunction (myasthenia gravis), non-infectious encephalitis, and myelitis occur more commonly with PD-1/PD-L1 inhibitors than anti-CTLA-4 agents.^11,14 In contrast, Guillain-Barré syndrome (GBS) and non-infectious meningitis occur more commonly with anti-CTLA-4 agents.^11,14 It has been reported that overall adverse events are more common with combined use of CTLA-4 antagonists and PDL-1 inhibitors.¹¹

Case reports and case series have further helped clarify the spectrum of neurologic complications associated with ICI therapy.¹⁵ PNS toxicities, such as peripheral neuropathies, radiculopathies, and GBS, appear to be more common complications when compared to CNS side effects.¹⁵ Though less common, CNS toxicities have also been reported with myelitis, meningitis, and encephalopathies of varying types being documented in scattered cases.¹⁵ CNS irAEs more commonly associated with anti-CTLA-4 include, but are not limited to, transverse myelitis, aseptic meningitis, meningoencephalitis, necrotizing encephalitis, and brainstem encephalitis.¹⁰ CNS irAEs more commonly associated with anti-PD-1 therapies include limbic encephalitis, inflammatory demyelination, and brainstem encephalitis.¹⁰

Clinical Case Presentations

Diagnosis is made based on radiologic imaging and clinical signs and symptoms. Patients with new-onset moderate to severe neurological signs or symptoms after treatment with ICIs should be considered for holding or discontinuation of therapy.¹⁶ Neurological irAEs should be a diagnosis of exclusion. When appropriate, magnetic resonance imaging (MRI), electromyography (EMG), cerebrospinal fluid (CSF) examination, and serum studies should be completed to exclude other causes such as vascular disease, progressive tumor disease involving the CNS structures, infection, paraneoplastic syndrome, or toxic/metabolic causes.¹¹ We recognize these clinical outcomes as potentially related to the host’s immune response to ICIs and attempt to distinguish these from the classic definition of paraneoplastic syndromes where an immune response is triggered by the neoplasm itself.

Rapid diagnosis and treatment are essential as neurological irAEs can lead to serious sequelae or death. A multidisciplinary team should participate in decision management for patients treated with ICIs to avoid discontinuation of vital therapies for tumor control.^10,17 We caution the readers to note that in most of the cases, brain and body imaging as well as CSF analysis were sought in accordance with standard of care to best identify tumor progression. Often, these patients were not reasonable candidates for neurosurgical intervention in order to support radiological findings. In some cases, removal of the presumed offending ICI therapy and documenting clinical improvement were sufficient to diagnose irAE related to ICIs. However, in these cases, we worked to obtain follow-up CNS imaging at the earliest timepoint possible in relationship to clinical improvement. We included MRI brain perfusion-weighted imaging sequences in all patient cases with a history of prior brain irradiation to assist with discerning radiation necrosis from tumor progression. Evaluation of relative cerebral blood volume (rCBV) through MRI perfusion does not, however, help to discern radiation necrosis from nonspecific and inflammatory lesions secondary to ICIs—as they both demonstrate low rCBV.

Nonspecific Changes

Okada et al. described immunotherapy-related neuroimaging changes that are benign and nonspecific in patients within 6 months of treatment on IO which should not be confused as progression.¹⁸ These lesions can be solitary or diffuse. This criterion is referred to as “immunotherapy response assessment in neuro-oncology (iRANO).” ¹⁸

Parenchymal

A 61-year-old female with a ROS1-rearranged non-small–cell lung cancer (NSCLC) was initially treated with crizotinib then switched to ceritinib following progressive body disease. Multiple subsequent progressions and brain metastasis were treated with carboplatin, pemetrexed, and ceritinib. The patient was eventually treated with an ICI, pembrolizumab, followed by whole-brain radiotherapy (WBRT) with stereotactic radiosurgery (SRS) boost for multiple brain metastases. Routine follow-up brain MRI demonstrated nonspecific enhancing lesions in the left cerebellum without associated mass effect or elevation in rCBV on perfusion-weighted imaging. These new lesions were not noted to be within the SRS treatment area. Clinical presentation was notable for grade 1 limb edema and complaints of vivid dreams; otherwise, no other CNS effects were noted. Subsequent MRIs taken 8 months after initiation of ICI therapy showed an interval resolution of the left cerebellar enhancing lesion. This lesion was not determined to be symptomatic and deemed nonspecific radiologic changes secondary to ICI therapy as described by Okada et al. (Figure 1A-C).

Figure 1.

Case of nonspecific asymptomatic changes (A-C). (A) Within the first month of ICI therapy. Sagittal view of T1 FLAIR propeller post-contrast demonstrating normal brain. (B) Within 3 months of initiating ICI therapy. Sagittal view of T1 3D SPGR post-contrast demonstrating the appearance of left cerebellar lesion (yellow circle). The initial radiographic impression read, “potentially immunotherapy-related treatment-related, however, metastatic disease cannot be ruled out.” (C) Eight months after discontinuing ICI therapy. Sagittal view of T1 3D SPGR post-contrast demonstrating resolution of the previously noted enhancing left cerebellar lesion. Case of peri-tumoral changes (rows D-G). (Row D) 6 months after starting ICI. Interval size increase of the lesion in the left parieto-occipital junction (Row D, left). The edema surrounding this lesion has also increased yellow circle. A 5-mm enhancing lesion (T1 post-contrast not shown) with associated peritumoral edema in the right occipital lobe appears slightly smaller than on prior study (top right, red oval). (Row E) Also at 6 months after the start of ICI—correlate perfusion-weighted imaging does not demonstrate elevation in rCBV associated with either the bilateral occipital lobe areas yellow circle left and red oval on right. More consistent with a treatment-related effect. (Row F) 14 months after starting ICI. The left parieto-occipital lesion has slightly decreased in size, with less surrounding T2/FLAIR signal abnormality (middle left). Slight interval increase in the size of the right occipital lobe peritumoral edema without changes in T1 contrast (not shown) (middle right). (Row G) 19 months’ follow-up after starting ICI with noted further decrease in size of T2 FLAIR signal abnormality in the left temporal occipital region (bottom left). Further interval fluctuation in size and prominence of right occipital lobe FLAIR abnormality representing nonspecific peritumoral vasogenic edema (bottom right).

Encephalitic Changes

When considering a differential diagnosis, encephalitis must be differentiated from infectious diseases, metabolic diseases, brain metastasis, cerebrovascular diseases such as cerebral hemorrhage, cerebral infarction, and metastatic tumor of the pia mater.¹⁰ This case would require clinical and routine serological examination, serum protein electrophoresis, and CSF/serum virology assessment. Neuroaxis imaging and routine CSF including cytology, and potentially flow cytometry should be completed.¹⁰

Peri-tumoral T2 FLAIR

A 61-year-old male with metastatic adenocarcinoma of the lower left lung, previously underwent lung radiation therapy and SRS for noted brain metastases in the bilateral occipital lobes, followed by four cycles of pemetrexed and carboplatin systemic therapy. Recurrent brain metastasis and body progression were noted 3 years later and treated with brain SRS with maintenance bevacizumab for vasogenic edema, and initiation of nivolumab ICI therapy. A brain MRI 6 months after starting ICI showed a presumed mixed response with increased size of the left parietooccipital lesion and decrease in size of the right occipital lobe lesion. Subsequent MRIs showed decreasing size of the left enhancing lesion with less surrounding T2 FLAIR (fluid-attenuated inversion recovery) and interval increases in size of the nonspecific T2 FLAIR associated with the right lesion. Without clinical decline and no evidence of new metastatic disease, the patient continued on every 3-week dosing of ICI with stability. We determined the lesion was a nonspecific neuroradiologic change secondary to ICI therapy as described by Okada et al.¹⁸ (Figure 1, rows D-G).

Peri-device and peri-resection cavity

A 57-year-old male with NSCLC treated with carboplatin and paclitaxel. Chemotherapy side effects led to second-line treatment with pembrolizumab ICI monotherapy. Noted right temporal lobe brain metastases and leptomeningeal carcinomatosis were treated with resection of single lesion followed by WBRT and SRS boost, followed by Ommaya reservoir placement and intrathecal (IT) methotrexate (MTX) chemotherapy. Brain MRI surveillance was completed every 3 months. Within 4 months on ICI therapy, he developed skin changes, elevated thyroid labs, onset of worsening limb weakness, and intermittent confusion and headaches occurred. Brain MRI demonstrated interval increase in volume of enhancement around the Ommaya catheter without elevated rCBV and expected diffuse leukoencephalopathy related to WBRT and IT. CSF was negative for malignant cells but continued to demonstrate elevated white blood cells and proteins, but normal glucose. Multidisciplinary tumor board discussion debated IT therapy-related necrosis vs peri-device encephalitis secondary to immunotherapy. IT therapy was initially held with the patient’s continued decline. Ultimately, the decision was made to hold ICIs, and high-dose steroids were initiated. Despite holding therapies and no evidence of systemic progression, the patient continually declined over the next several months with worsening encephalitic radiologic changes and death. Here, we present a case of dual causality with progression of the neoplastic process leading to the patient’s death—concurrent with the inflammatory CNS process likely secondary to ICI therapy (abnormal CSF and brain imaging timed with ICI-related thyroid and skin changes) as the cause for the neuroradiological presentation with progressive focal leptomeningeal enhancement (Figure 2).

Figure 2.

Case of peri-device and peri-resection cavity changes. MRI brain left-side T1 post-contrast and right-side perfusion-weighted imaging. (Row A) Peri-device changes (left) sagittal image Ommaya reservoir (yellow arrow) with enhancement along catheter (red arrows), without elevated rCBV (red oval) (right), described as possible disseminating necrotizing leukoencephalopathy (DNL) vs peri-device associated inflammation secondary to ICI, no concern for malignancy. (Row B) Peri-resection cavity changes (left) axial image with diffuse linear enhancement along the cavity (yellow arrows) without nodularity or associated elevated rCBV (right). (Row C) (left) axial image with continued progression of enhancement with increased nodularity (yellow arrows) around resection cavity and mild elevation of rCBV (white arrowheads) on perfusion study (right)—likely combined treatment-related encephalitis and tumor progression.

Remote traumatic brain injury (TBI) with encephalomalacia (metastatic mimicry)

A 70-year-old male with metastatic clear cell renal cell carcinoma status post left-sided radical nephrectomy and first-line pazopanib followed by nivolumab ICI therapy at progression. The patient had early onset of skin changes and bronchitis presumed secondary to ICI use; however, therapy was continued. After 3 months on ICIs, the patient developed seizure-like activity and a left visual field cut on examination in the emergency department. MRI of the brain showed a mass-like lesion in the right occipital lobe concerning for metastatic disease. Patient underwent a right occipital craniotomy and pathology revealed a diffuse benign lymphocytic infiltrate with reactive predominant T-lymphocytes positive for T-cell receptor β and γ gene rearrangements. Pathology determined the findings to be consistent with immune-mediated reactive T-cell response. A second opinion from the National Institutes of Health pathologist concurred. Further history from the patient revealed a remote TBI and resultant area of right occipital encephalomalacia discovered on skull-based portion of body PET CT (not shown here). This was a case of encephalitic metastatic mimicry as referenced in our prior publication Bross et al. with adapted images.¹⁹ (Figure 3).

Figure 3.

Case of metastatic mimicry. Preoperative MRI 1 year after starting anti-PD-1 immunotherapy. (A) Axial T1 post-contrast, (B) T2 FLAIR, (C) T2 fat saturation demonstrate a septated “mass” lesion with heterogeneous enhancement and lobulated contour centered within the right occipital lobe with surrounding vasogenic edema and mild local mass effect. (D) Diffuse perivascular infiltrate of small-sized monotonous lymphocytes with mature chromatin (scale bar 100 µm). (E) CD4-positive staining of T cells (scale bar 100 µm).

Hypophyseal Changes

Hypophysitis is more common with anti-CTLA-4 than PD-1/PD-L1 inhibitors.²⁰ Anti-CTLA-4 can cause direct damage to the pituitary gland as pituitary cells naturally express CTLA-4.²¹ Onset of hypophysitis usually occurs 4-10 weeks after initiation of ICI treatment.²² This type of toxicity typically occurs in men and older patients.²³

Hypophysitis

A 54-year-old male was treated for recurrent B-Raf proto-oncogene (BRAF)-positive metastatic melanoma with combined immunotherapy using both ipilimumab (Day 1 and Day 22) and nivolumab ICIs (Days 1, 15, 22, and 29). The patient developed grade 2 rash 10 days after he received the first doses of combined ICIs. He was treated with prednisone 1 mg/kg/day, and ICIs were temporarily held. He resumed combined immunotherapy on Day 22, then nivolumab therapy alone on Day 29 of cycle 1. Seven days later, the patient developed severe bilateral frontal and temporal headache associated with photophobia. He denied vision changes, nausea/vomiting, or tinnitus. MRI showed diffuse interval enlargement of the pituitary gland. The pituitary gland measured up to 10 mm in the craniocaudal dimension compared to 4 mm previously and demonstrated mildly heterogeneous enhancement. Overall, findings were most consistent with immunotherapy induced hypophysitis. Immunotherapy was held; the patient was hospitalized and treated with high-dose steroids and hormonal replacement therapy for hypophysitis and panhypopituitarism. All ICIs were discontinued going forward for this patient (Figure 4).

Figure 4.

Case of hypophysitis. Sagittal T1 post-contrast brain MRI. (A) One week prior to starting anti-CTLA-4 therapy demonstrating the absence of acute intracranial abnormalities. (B) Two months after initiation of therapy demonstrating enlargement of the pituitary gland (yellow arrow). (C) Two months after discontinuation of ICIs demonstrating interval resolution of pituitary enlargement.

Spinal Cord, Cranial, and Peripheral Nerve Changes

Optic nerve plus nonspecific parenchymal changes

A 66-year-old female with epidermal growth factor receptor (EGFR), negative anaplastic lymphoma kinase (ALK), and translocation positive stage IV NSCLC. The patient was initially treated with systemic crizotinib and subsequent progressions over the next 3 years were treated with ceritinib followed by alectinib, then carboplatin, and pemetrexed. The patient developed brain metastases which were treated WBRT with hippocampal avoidance and SRS boost to the right temporal lobe, and monthly nivolumab immunotherapy was added. After 4 cycles of ICIs, the patient had vision changes. MRI of the brain and orbits at this time (24 weeks after initial WBRT and SRS) showed homogeneous enhancement of the intracranial segments of the optic nerves and small nonspecific enhancing 6-mm left cerebellar lesions (not included in the original field of SRS). CSF collected at the time of abnormal cranial nerve imaging was normal. The differential included radiation-induced optic neuropathy, metastatic disease, or immunotherapy-related treatment changes. Tumor board consensus was to hold immunotherapy with short course of steroids. After noted improvement, the patient was changed to lorlatinib therapy. The patient was determined to have cranial nerve and diffuse parenchymal changes related to ICI therapy. Ten weeks later, the patient was noted to have onset of seizures, and an MRI brain with perfusion was obtained. While the optic nerve enhancement and nonspecific cerebellar lesions were less prominent, there was development of a right temporal lobe enhancing lesion with associated areas of elevated rCBV, indicating admixed tumor progression and some radiation necrosis in this area. Despite her tumor progression on new lorlatinib therapy, the improved clinical symptoms were observed with the interval resolution of the optic nerve and cerebellar enhancement of ICIs (Figure 5A-G).

Figure 5.

Case of optic neuritis and nonspecific changes (A-G). MRI T1 thin FAT SAT plus contrast (A) coronal and (B) axial orbits with homogeneous enhancement of the intracranial segments of the optic nerves (yellow arrows). (C) and (D) nonspecific enhancing lesions of the cerebellum. (E-G) Follow-up imaging (2.5 months later) demonstrated interval resolution of nonspecific imaging changes along the optic nerves (E, red circle) and in the cerebellum (G, yellow circle). However, new findings on follow-up identified a new enhancing lesion in the right temporal lobe with associated elevation in rCBV (E/F yellow arrows) and vasogenic edema on T2 sequences (not shown). Case of brachial neuritis (Parsonage-Turner Syndrome) (H, I). (H) Coronal T2-weighted images with fat saturation show increased T2 signal in the supraspinatus, infraspinatus, and teres minor (red arrowheads) muscles consistent with edema. MRI chest, brachial plexus (I) axial STIR with diffusely increased signal in the left infraspinatus muscle (yellow rectangle) consistent with denervation from brachial neuritis.

Brachial neuritis/neuralgic amyotrophy (Parsonage-Turner syndrome)

A 58-year-old male with BRAF-positive metastatic melanoma was treated with nivolumab ICI therapy every 2 weeks for 1 year. Six months after finishing his immunotherapy, he complained of sharp pains shooting down both thumbs, weakness, and numbness in the thenar eminence bilaterally. Physical exam showed positive Tinel sign, increased reflexes in the left upper extremity, decreased strength more proximally in the left upper extremity with mild decrease in strength on wrist extension on the left. MRI of the left shoulder showed diffuse edema like signal within the supraspinatus, infraspinatus, and teres minor muscles consistent with Parsonage-Turner syndrome. EMG showed electrodiagnostic evidence for mild carpal tunnel syndrome in the left hand and borderline carpal tunnel syndrome in the right hand. Five weeks after initiating a course of steroids and physical therapy, the patient demonstrated interval improvement of clinical symptoms and a partial recovery of upper extremity strength. In this case, the patient likely experienced a late-delayed reaction to prolonged ICI use. While we documented his changes at 6-most post-completion of ICIs on review of his medical chart, the patient had been complaining of subtle and subjective weakness of the upper extremity without noted pain. It is unclear if there were findings sooner as no shoulder imaging was obtained until symptoms involved pain (Figure 5H, I).

Lumbar radiculopathy and Conus inflammation with vascular abnormalities

A 64-year-old male with relapsed Hodgkin lymphoma was treated with several lines of salvage including ifosfamide, carboplatin, and etoposide (ICE) and brentuximab vedotin both as a single agent and in combination with bendamustine with progression. After failing these initial lines of therapy, the patient was initiated on nivolumab ICI. Two to three months after starting ICI therapy, the patient noted onset of progressive radiating lower lumbar pain and weakness with associated bilateral foot numbness. Physical examination revealed 1+/4 Achilles and patellar reflexes. Routine lumbar X-rays demonstrated intervertebral disc degeneration and multilevel facet osteoarthritis. A PET scan at that time showed no concerning abnormalities or metastatic bone involvement. Multidisciplinary tumor board case reviews considered the following differentials: inflammation, demyelination, neoplastic involvement, ischemic injury, and vascular malformation. Vascular imaging obtained at about 6 months after onset of initial symptoms detected a small incidental aortic aneurysm and a dural A-V fistula (dAVF). The dAVF was treated with embolization and ICIs were continued and the patient’s pain was not notably improved, but there was improvement in foot numbness. Six months later, for a total of 1.25 years on ICIs, the patient suffered liver toxicity and nivolumab was discontinued. With continued follow-up over the next 12 months’ off ICIs, the patient’s leg weakness and lumbar pain gradually improved. Nivolumab was restarted with progression identified on surveillance PET/CT, and no further neurological complications were documented. We speculated that persistent microvascular inflammation secondary to ICIs led to the development or exacerbation of a dAVF and endoneurial irritation as the cause of his symptoms as described in Manousakis et al.¹³ The patient improved to baseline neurological function while off ICIs, and was able to resume therapy without neurological concerns for 9-month follow-up on re-initiation of ICIs by the time of this submission (Figure 6).

Figure 6.

Case of vascular malformation and spinal cord expansion. MRI sagittal STIR (yellow arrows) 6 months after initiation of ICI (A), T2 (B), and axial T2 (C) with noted diffusely increased signal within the mildly enlarged distal spinal cord primarily in the region of the conus with noted primary central cord edema. Findings are nonspecific and differential includes inflammation, demyelination, infection, or neoplastic etiologies. MRA/MRI T/L spine 3 months symptomatic. (D) Sagittal T2 image with continued mild distal expansion and dilated perimedullary vessels (yellow arrows) (E) sagittal and (F) axial T1 post-contrast with intense contrast enhancement of the distal spinal cord and conus medullaris (red arrows), and (G) enhancement of the dorsal perimedullary veins (thick yellow arrows). Findings are characteristic of a spinal dAVF (type 1 AVM).

Meningeal Changes

A 66-year-old male with high-grade carcinoma of the right parotid gland was initially treated with cisplatin and concurrent radiation therapy followed by casodex and pembrolizumab ICI therapy at progression with noted brain metastasis of right temporal lobe and bifrontal lobe. The right temporal lesion was treated with laser interstitial thermal therapy (LITT) then required open resection of tissue necrosis secondary to LITT. The patient developed LMC treated with WBRT and Ommaya reservoir placement and IT MTX. Paclitaxel therapy replaced ICI for a period of 1 year but was stopped due to toxicity and pembrolizumab was re-initiated. MRI brain 6 months after re-initiation of ICI therapy demonstrated multifocal enhancing signal abnormalities of the cerebellum and diffuse enhancement surrounding the resection cavity. Without convincing evidence of tumor progression and negative CSF cytology, changes were attributed to immunotherapy-related inflammation and necrosis. As the ICIs were controlling systemic disease decision to continue therapy after a brief holiday for multidisciplinary consensus. Follow-up imaging revealed worsening of MRI changes involving the dura without patient clinical worsening and overall stability of his body disease. Patient was not a good neurosurgical candidate, thus limiting diagnostic tissue confirmation. CSF cytology remained benign but proteins and white cells were elevated with fluid xanthochromia. The decision was made for patient to temporarily hold ICI therapy and initiate high-dose steroids with repeat MRI brain in 6-8 weeks with suspicion for pachymeningeal inflammation secondary to ICIs. Unfortunately, follow-up imaging at 8 weeks off ICIs and high-dose steroids demonstrated relatively asymptomatic tumor progression without resolution of the initially noted enhancement, new mass effect on the brain. The imaging report read, “Further progression of the large infiltrative mass in the right lateral temporal lobe with extension to the central skull base and along right tentorial leaflet with suggestion of perineural and transependymal tumor spread.” Salivary tumors often present with asymptomatic slow progression and can spread along the meninges and mimic a leptomeningeal process. This was not determined to be a case of neuroradiologic evidence of irAEs related to ICIs. The patient went to hospice and died shortly after repeat imaging (Figure 7, rows A-D).

Figure 7.

Case of pachymeningitis (rows A-D). (Row A) Axial T1 post-contrast MRI brain 3 months prior to re-initiation of therapy. (Row B) Axial T1 post-contrast MRI brain 6 months after re-initiation of ICI demonstrating multifocal leptomeningeal and dural signal abnormalities and enhancement. (Row C) Contrasted MRI brain axial (left) perfusion-weighted imaging without elevation in rCBV in corresponding areas of worsening enhancement (right). (Row D) Axial T1 post-contrast imaging at 8-week follow-up imaging with continued progression of tumor along the meningeal and subependymal surfaces (yellow arrows) and elevated rCBV in perfusion imaging on far right. Case of osseous changes (E-G). Sagittal T1 brain MRI (E) imaging 1 month prior to start of anti-PD-1 therapy with intact bone as compared to (F) 5 months after starting ICI therapy demonstrating new diffuse T1 hypo-intensities (white arrows) of bone marrow signal in imaged osseous structures, including calvarium, skull base, and imaged upper cervical spine. (G) MRI brain follow-up 3 months off ICIs with interval partial resolution of abnormal marrow signal.

Osseous Changes

Moseley et al. was the first to report potential skeletal irAEs and reported on six cancer patients treated on ICIs found to have new osteoporotic fractures or focal bone resorptive lesions after nonspecific musculoskeletal complaints. Two patients were treated with combination ipilimumab and nivolumab and the remaining patients were treated with anti-PD-1 monotherapy.²⁴ Patients in this series were not found to have preexisting osteoporosis.²⁴ Interestingly, patients with destructive or resorptive bone changes were also noted to have arthritic joint symptoms while on ICI therapy.²⁴

Skull-base bone changes

A 56-year-old male EGFR-negative, ALK-negative Stage II-A (right upper lung) and Stage I-A (left lower lung) NSCLC was initially treated with concurrent radiation and carboplatin/pemetrexed therapy. Right temporal brain metastasis was noted approximately 2.5 years after initial diagnosis and treated with SRS. Pembrolizumab ICI was added to carboplatin/pemetrexed briefly, but therapy was reduced to pembrolizumab/pemetrexed due to carboplatin toxicity. At 5 months of ICI therapy, the patient complained of left mandibular, hip, and knee pain. A brain MRI demonstrated new diffuse replacement of yellow marrow signal in osseous structures as compared to prior MRI, but a nuclear medicine bone scan showed no evidence of metastatic disease. The patient took a drug holiday off ICIs with resolution of symptoms without an associated MRI brain completed for comparison before the patient was restarted on ICIs for another 7 months with a varied treatment schedule due to symptoms. The patient was taken off ICI therapy permanently and monitored every 3 months for systemic and brain response. Follow-up MRI scan at 3 months off ICI therapy revealed partial resolution of the abnormal marrow signal (Figure 7E-G).

Discussion

Immunotherapy has evolved and significantly advanced cancer management in recent years. There are many types of immunotherapy including monoclonal antibodies, ICIs, oncolytic virus therapy, T-cell therapy, such as CAR T-cell therapy, and cancer vaccines.²⁵ Our review focuses on ICIs and the associated neurological adverse events that have characteristic associated neuroradiological findings. Because ICI-related neurotoxicity may be fatal, patients on ICI should be instructed to notify the oncology healthcare team if they develop any new signs or symptoms that might suggest neurological irAE. Radiological findings of neurological irAEs might precede neurological signs and symptoms and therefore close monitoring is warranted during the induction phase of therapy.^11,12 The decision to discontinue ICIs is complicated and frequently involves a multidisciplinary team discussion. Our literature review was limited by it being retrospective in nature. There are no prospective studies that report treatment approaches to radiological findings associated with neurological irAE. Another limitation was the narrow focus of this review to exclude other types of immunotherapies with neurological irAEs with radiological changes.²⁶ Juwhan Choi and Sung Yong Lee²⁷ conducted an important study reviewing some of the common organ-specific irAEs. Further studies are needed to provide reviews that focus on evaluation of diagnostic body imaging to determine system irAEs to better inform and educate providers caring for patients on these therapies.

Accurately discerning between surgical-related changes, tumor recurrence or pseudoprogression, nonspecific immune-oncotherapy–related lesions, and radiation necrosis or treatment effect is of critical importance.²⁸ Often, clinicians employ on routine and advanced brain tumor imaging along with the patient clinical assessment to make these determinations. Owen et al. evaluated patients who had been treated with body irradiation before or after initiation of immunotherapies and noted there was not an increased risk for irAEs.²⁹ Furthermore, in this study, the development of irAEs did not significantly correlate with survival when controlling for the duration of therapy in a landmark analysis.²⁹ We were unable to identify any published studies that assessed the potential risk for neuroradiological changes in patients on immunotherapies who were exposed to CNS irradiation, however.

Conclusion

This report addresses the potential to more rapidly recognize and consider specific neuroradiological features that could be associated with ICI therapy. To our knowledge, no other such report has been published that specifically addresses this issue. Patients treated with ICIs should be carefully monitored for neurological irAEs during treatment, particularly during the first 4 months of therapy. Because some of the neurological irAEs are nonspecific, early identification and treatment of neurological adverse events is a critical part of the management and surveillance plan. A high level of clinical suspicion is essential for early diagnosis. Although significant radiological findings might occur with immunotherapy, subtle and minimal radiological changes are often reported with a broad differential and require close monitoring and follow-up imaging. The decision to continue or discontinue ICIs in the setting of neurological irAE remains a challenge in our field and often demands multidisciplinary discussion which includes neuro-oncologists and neuroradiologists. While this is one case series of interesting related cases of patients with CNS tumors who were treated with ICI, we acknowledge that further investigations are necessary to determine the best imaging algorithms for accuracy of these modalities to differentiate between the various proposed causes for abnormal radiological signal.

Glossary

Abbreviations

ALK

anaplastic lymphoma kinase

BRAF

B-Raf proto-oncogene

BTLA

B- and T-cell lymphocyte attenuator

CAR

chimeric antigen receptor T cells

CNS

central nervous system

CSF

cerebrospinal fluid

CTLA-4

cytotoxic T-lymphocyte–associated antigen 4

dAVF

dural A-V fistula

EGFR

epidermal growth factor receptor

EMG

electromyography

FLAIR

fluid-attenuated inversion recovery

GBS

Guillain-Barré syndrome

ICI

immune checkpoint inhibitor

IO

immuno-oncotherapy

irAEs

immune-related adverse events

IT

intrathecal

LITT

laser interstitial thermal therapy

LAG3

lymphocyte activation gene-3

MRI

magnetic resonance imaging

MTX

methotrexate

NSCLC

non-small cell lung cancer

PNS

peripheral nervous system

PD-1

programmed cell death protein 1

PD-L1

programmed cell death protein ligand 1

rCBV

relative cerebral blood volume

SRS

stereotactic radiosurgery

TIM3

T-cell immunoglobulin and mucin-domain containing-3

WBRT

whole-brain radiotherapy

Funding

This work was supported by the Geisinger Foundation through philanthropic donations from Mr. Jeff Erdly, Mr. Jerry Sandel, Lowe Family, and Comp Family.

Author contributions statement. N.N.G., M.M., G.J.M., R.P.—Concept development as well as the below listed contributions. A.M., A.S., C.B., E.F., E.N.L., J.V., T.M., S.P.B.—Substantial involvement in the development of the draft manuscript, contributed important intellectual content, data interpretation and development of figures, critical revisions, and discussions around important intellectual content. Final approval of the planned version for publication. All authors agreed to be accountable for all aspects of the work for accuracy and are committed to the integrity of the final product.

Conflict of interest statement. No authors endorse any relevant conflicts of interest.