Association of Autoimmune Encephalitis With Combined Immune Checkpoint Inhibitor Treatment for Metastatic Cancer

Importance  Paraneoplastic encephalitides usually precede a diagnosis of cancer and are often refractory to immunosuppressive therapy. Conversely, autoimmune encephalitides are reversible conditions that can occur in the presence or absence of cancer.

Objective  To report the induction of autoimmune encephalitis in 2 patients after treatment of metastatic cancer with a combination of the immune checkpoint inhibitors nivolumab and ipilimumab.

Design, Setting, and Participants  A retrospective case study was conducted of the clinical and management course of 2 patients with progressive, treatment-refractory metastatic cancer who were treated with a single dose each (concomitantly) of the immune checkpoint inhibitors nivolumab, 1 mg/kg, and ipilimumab, 3 mg/kg.

Exposures  Nivolumab and ipilimumab.

Main Outcomes and Measures  The clinical response to immunosuppressive therapy in suspected autoimmune encephalitis in the setting of immune checkpoint inhibitor use.

Results  Autoantibody testing confirmed identification of anti–N-methyl-D-aspartate receptor antibodies in the cerebrospinal fluid of 1 patient. Withdrawal of immune checkpoint inhibitors and initiation of immunosuppressive therapy, consisting of intravenous methylprednisolone sodium succinate equivalent to 1000 mg of methylprednisolone for 5 days, 0.4 mg/kg/d of intravenous immunoglobulin for 5 days, and 2 doses of rituximab, 1000 mg, in 1 patient and oral prednisone, 60 mg/d, in the other patient, resulted in improved neurologic symptoms.

Conclusions and Relevance  Immune checkpoint inhibition may favor the development of immune responses against neuronal antigens, leading to autoimmune encephalitis. Early recognition and treatment of autoimmune encephalitis in patients receiving immune checkpoint blockade therapy will likely be essential for maximizing clinical recovery and minimizing the effect of drug-related toxic effects. The mechanisms by which immune checkpoint inhibition may contribute to autoimmune encephalitis require further study.

Quiz Ref IDImmune checkpoint blockade for cancer therapy aims to enhance antitumor immunity. Nivolumab is a fully human IgG4 antibody that blocks programmed cell death protein 1 and potentiates activation of T cells.¹ Similarly, ipilimumab is a fully human monoclonal antibody that binds and inhibits cytotoxic T-lymphocyte–associated antigen 4, an inhibitory receptor on T cells.² Both therapies have demonstrated improved tumor-related outcomes in multiple types of cancer.³

Although these therapies hold great promise in treating various malignant neoplasms, checkpoint inhibitors uncommonly trigger varied immune-related adverse events of the central and peripheral nervous systems.^4-9Quiz Ref ID We describe 2 patients who developed autoimmune encephalitis, including anti–N-methyl-D-aspartate receptor (anti-NMDAR) encephalitis, shortly after treatment with the combination of nivolumab and ipilimumab for metastatic cancer. Administration of immunosuppressive therapy and cessation of combination checkpoint inhibition led to marked neurologic improvement. Although causality cannot be proven, these cases illustrate important factors for consideration in the use of immune checkpoint inhibitors.

Box Section Ref ID

Written consent was provided by the first patient and the wife of the second patient, as he was deceased at time of manuscript preparation. The Johns Hopkins University Institutional Review Board waived approval.

A woman in her mid-50s with a history of metastatic melanoma had previously received adoptive T-cell transfer therapy (NCT01993719) (Figure 1A) with partial response initially (per the Response Evaluation Criteria in Solid Tumors, guideline version 1.1¹⁰), followed subsequently by disease progression, including new metastases to the brain treated with stereotactic radiosurgery. She received 1 dose each (concomitantly) of nivolumab, 1 mg/kg, and ipilimumab, 3 mg/kg (NCT02186249). During the next week, the patient reported fever, generalized body aches, nausea, and vomiting. Within 2 weeks, she developed syncopal episodes, memory loss, gait disturbance, and abnormal behaviors, including unresponsiveness and inappropriate laughing. The patient was hospitalized. Vital signs showed evidence of dysautonomia with hypotension and bradycardia. By 18 days after receiving the infusion of nivolumab and ipilimumab, results of her neurologic examination revealed disorientation, inattention, bradykinesia, and hyperreflexia. Results of extensive serologic evaluations for metabolic derangements were unremarkable. Computed tomographic scan of the head showed no acute pathologic conditions.

During the next few days, the patient became stuporous, with episodic agitation. Magnetic resonance imaging (MRI) of the brain showed stable encephalomalacia at sites of prior radiosurgery with no additional metastases (eFigure in the Supplement). No changes were noted at previously irradiated tumor sites on brain MRI. Cerebrospinal fluid (CSF) demonstrated a monocytic pleocytosis (white blood cells, 8/µL; 100% lymphocytes, 0% monocytes, and 0% neutrophils; institutional normal range, 0-5/µL [to convert white blood cells to ×10⁹/L, multiply by 0.001; and lymphocytes to proportion of 1.0, multiply by 0.01]). Results of cytologic tests showed no evidence of malignant neoplasm. Cerebrospinal fluid protein and glucose levels, results of cytologic tests, and IgG index were normal. Oligoclonal bands were present and matched in the CSF and serum. Results of polymerase chain reaction were negative for herpes simplex virus in CSF. Results of an extensive evaluation of blood and CSF revealed no evidence of infection. Serial electroencephalography showed intermittent bilateral slowing, then a subclinical seizure of left temporo-occipital origin. Continuous electroencephalographic monitoring showed intermittent periods of rhythmic epileptiform activity in the left temporal lobe without clinical correlate. The patient remained stuporous. Subsequent analysis of CSF demonstrated a persistent monocytic pleocytosis (white blood cells, 6/µL; 100% lymphocytes, 0% monocytes, 0% neutrophils), resolution of the oligoclonal bands, and no other abnormality.

Given the high suspicion for autoimmune encephalitis, paraneoplastic antibody testing was performed using CSF and serum, and the patient was treated empirically with high-dose intravenous methylprednisolone sodium succinate equivalent to 1000 mg/d of methylprednisolone for 5 days, followed by 0.4 mg/kg/d of intravenous immunoglobulin for 5 days, without significant improvement. Analysis demonstrated IgG NMDAR antibodies in the CSF only (first performed by Athena Laboratories, Marlborough, Massachusetts; confirmed in the laboratory by one of us [J.O.D.]). The patient was treated with 2 doses of intravenous rituximab, 1000 mg, resulting in gradual improvement in mental status for 4 weeks. The patient’s score on the Montreal Cognitive Assessment 6 months after discharge was 28 of 30 (normal, ≥26), suggesting no cognitive impairment, and she had otherwise normal neurologic examination results. She experienced an initial partial response (per the Response Evaluation Criteria in Solid Tumors, guideline version 1.1¹⁰) 4 months after receiving the checkpoint inhibitors.

The patient received 2 doses of rituximab. One month after the second dose, she developed evidence of disease progression in a single external iliac lymph node, which was treated with stereotactic body radiation therapy. The patient has not received further anticancer or immunosuppressive agents and remains in stable condition 12 months after nivolumab and ipilimumab treatment.

A man in his mid-60s with metastatic small cell lung cancer that had progressed despite multiple prior therapies was treated with nivolumab, 1 mg/kg, and ipilimumab, 3 mg/kg (NCT 01928394) (Figure 1B). Within 4 days, the patient developed short-term memory loss and progressive difficulty ambulating. Magnetic resonance imaging of the brain demonstrated new nonspecific T2 hyperintensities in the right mesial temporal lobe (Figure 2A). The patient was hospitalized. Results of neurologic examination on admission were unremarkable. Serologic studies showed hyponatremia (sodium, 124 mEq/L [to convert to millimoles per liter, multiply by 1.0]; institutional normal range, 135-148 mEq/L), consistent with a history of syndrome of inappropriate antidiuretic hormone (serum sodium, 124-139 mEq/L in the preceding 6 months). Results of serologic evaluations of other metabolic derangements were otherwise normal. Gradual correction of the hyponatremia led to subjective clinical improvement, and the patient was discharged home.

Within 2 weeks, he was rehospitalized with lightheadedness, disorientation, memory loss, and right arm paresthesias. Results of a neurologic examination revealed lethargy, poor recall, mild intention tremor, hyperreflexia, and ataxic gait. Results of metabolic and nutritional assessment were notable only for hyponatremia (sodium, 126 mEq/L). Results of subsequent serologic studies for other metabolic abnormalities were unremarkable. Results of repeated brain MRI were unchanged. No changes were noted at previously irradiated tumor sites on brain MRI. Analysis of CSF demonstrated monocytic pleocytosis (white blood cells, 18/µL; 89% lymphocytes, 11% monocytes, 0% neutrophils), an elevated protein level (98 mg/dL [to convert to grams per liter, multiply by 10.0]; institutional normal range, 15-45 mg/dL), a normal glucose level and IgG index, matched oligoclonal bands in the CSF and serum, and no evidence of bacterial or viral infection. Cytologic testing of CSF was not performed. Given the concern for autoimmune limbic encephalitis, the patient was treated with oral prednisone, 60 mg/d, and demonstrated dramatic improvement in his level of arousal and gait. The hyponatremia also resolved. No other metabolic derangements were detected on serologic analyses. Serum was positive for antiglial nuclear antibody, with a markedly elevated titer (>1:15,000). No other paraneoplastic autoantibody was detected in serum; CSF paraneoplastic autoantibody testing was not performed. The patient was discharged home with a slow tapering regimen of oral prednisone, starting at a dosage of 60 mg/d. Nivolumab and ipilimumab treatments were withheld. One month later, the patient’s neurologic functioning was nearly back to baseline. He experienced a partial response (per the Response Evaluation Criteria in Solid Tumors, guideline version 1.1¹⁰) 8 weeks after treatment with nivolumab and ipilimumab.

Nineteen weeks later, MRI of the brain showed resolution of the previously noted right mesial temporal lobe abnormality (Figure 2B), but also showed 2 new brain metastases (body of the corpus callosum extending along the right septum pellucidum and along the ependyma of the third ventricle). Stereotactic radiosurgery was performed, but the patient experienced marked functional decline and entered hospice care and died.

We describe 2 patients with metastatic cancer who developed autoimmune encephalitis after receiving combination therapy with immune checkpoint inhibitors. Quiz Ref IDAlthough causality cannot be proven in these cases, several features suggest that these syndromes were triggered by immune checkpoint blockade. The timing of the onset of neurologic symptoms after administration of nivolumab and ipilimumab suggests immune-related adverse events rather than classic paraneoplastic neurologic disorders (PNDs). Most tumor-induced PNDs are subacute and progressive, commonly preceding the detection of tumor by months to years.¹¹ Here, both patients were treated for metastatic cancer for extended periods without evidence of a PND. Both patients received only 1 dose of combination immune checkpoint inhibitors. Furthermore, anti-NMDAR encephalitis, as seen in the first patient, is believed to be caused by direct effects of pathogenic antibodies^12-15 rather than T-cell–mediated responses, which are implicated in classic PNDs. Finally, both patients had marked clinical improvement after immunosuppressive treatment. These features suggest that immune checkpoint inhibition favored the development of immune responses against neuronal antigens. That encephalitis has been reported following use of ipilimumab in metastatic pancreatic adenocarcinoma may also lend credence to this association, although autoantibody testing was not performed in this report.¹⁶ Further studies are required to confirm if there is a causal association between immune checkpoint blockade and autoimmune encephalitis.

The induction of antibodies to NMDARs in a patient with metastatic melanoma, as in case 1, is of particular interest. Melanoma is not commonly associated with PNDs, although there are reports of melanoma-associated retinopathy¹⁷ and cerebellar degeneration.^18,19 The identification of anti-NMDAR antibodies in a patient with melanoma has not been described, to our knowledge. N-methyl-D-aspartate receptors are expressed on melanocytes²⁰ and anti-NMDAR encephalitis after excision of melanocytic nevis has been recently described.²¹ Furthermore, the GRIN2A gene (OMIM 138253), which encodes the NMDAR subunit GluN2A, is highly mutated in patients with malignant melanoma,²⁰ leading to aberrant NMDAR complex formation.²² These findings may imply a convergence of mechanisms, including aberrant tumor expression of NMDARs along with a propensity for autoimmunity caused by immune checkpoint inhibitors.

Quiz Ref IDMore important, both patients had brain metastases before the onset of autoimmune encephalitis. It is difficult to estimate the contribution of this factor to the development of neurologic immune-related adverse events following immune checkpoint inhibition. Disruption of the blood-brain barrier or the damage resulting from stereotactic radiosurgery may have facilitated T-cell and B-cell interactions with neuronal epitopes. This hypothesis will require further investigation. There is a growing appreciation for anti-NMDAR encephalitis following infection with herpes simplex virus^23,24; herpes simplex virus was not detected in the first patient’s CSF, nor was there a history thereof. Most cases of anti-NMDAR encephalitis are autoimmune, with only 38% found in association with tumors, most commonly ovarian teratoma.^12,25 Previous studies showed NMDAR expression in ovarian teratomas²⁶; we were unable to examine whether or not NMDARs were expressed by the melanoma in case 1. Nonetheless, the observations in case 1 suggest a role for T-cell activation in the development of antibody-mediated autoimmune encephalitis, such as anti-NMDAR encephalitis. Furthermore, immune checkpoint inhibition may unmask or accelerate preexisting autoimmune reactions that target neuronal epitopes, leading to autoimmune encephalitis. Conversely, serum detection of antiglial nuclear antibody, as in case 2, is considered a marker of an underlying malignant neoplasm²⁷ and was not likely a pathogenic antibody-mediated process involving this antibody. Antiglial nuclear antibody has been associated with numerous PNDs,²⁸ making it less relevant per se to the autoimmune limbic encephalitis observed in case 2. However, the notion that immune checkpoint inhibition may have accelerated autoimmune reactions to this and other targets warrants further investigation.

As immune checkpoint inhibitors are used with increasing frequency in patients with malignant tumors, health care professionals should consider immune-related adverse events triggered by immune checkpoint inhibition among possible diagnoses of new-onset neurologic syndromes of unclear etiologic causes. Quiz Ref IDEarly recognition and management of these neurologic immune-related adverse events will be essential for maximizing clinical recovery and minimizing the effect of drug-related toxic effects.

Accepted for Publication: April 1, 2016.

Corresponding Author: Ellen M. Mowry, MD, MCR, Department of Neurology, Johns Hopkins University School of Medicine, 600 N Wolfe St, Pathology Bldg, Ste 627, Baltimore, MD 21287 (emowry1@jhmi.edu).

Published Online: June 6, 2016. doi:10.1001/jamaneurol.2016.1399.

Author Contributions: Drs Williams and Benavides contributed equally to the manuscript, and Drs Probasco and Mowry contributed equally to the manuscript. Drs Probasco and Mowry had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Williams, de Ávila, Lipson, Probasco, Mowry.

Acquisition, analysis, or interpretation of data: Benavides, Patrice, Dalmau, de Ávila, Le, Lipson, Probasco, Mowry.

Drafting of the manuscript: Williams, Benavides, Lipson, Probasco.

Critical revision of the manuscript for important intellectual content: Benavides, Patrice, Dalmau, de Ávila, Le, Lipson, Probasco, Mowry.

Administrative, technical, or material support: Le, Lipson.

Study supervision: Dalmau, de Ávila, Probasco, Mowry.

Conflict of Interest Disclosures: Dr Benavides reported receiving a fellowship grant from Mallinckrodt Pharmaceuticals. Dr Dalmau reported receiving royalties from Euroimmun for a patent for the use of N-methyl-D-aspartate receptor as an autoantibody test. Dr de Ávila reported being employed by and owning stock in Bristol-Myers Squibb. Dr Le reported receiving research funding from Merck, BMS, and Aduro Biotech to run clinical trials. Dr Lipson reported serving as a paid consultant for Amgen, Castle Biosciences, Merck, and Bristol-Myers Squibb. Dr Mowry reported that Teva Pharmaceuticals provided free glatiramer acetate for a randomized clinical trial of vitamin D supplementation, of which Dr Mowry is principal investigator. Dr Mowry reported serving as site principal investigator for a trial for Sun Pharmaceuticals, and reported that Johns Hopkins University School of Medicine receives funding from Biogen Idec to conduct research, and funding from the National Multiple Sclerosis Society. No other disclosures were reported.