Immune checkpoint inhibitors (CPIs) (anti-cytotoxic T-lymphocyte antigen-4, anti-programmed death 1, and anti-programmed death-ligand 1) have transformed the landscape of cancer therapy. However, their increasing use has unleashed immune-related adverse events in various organs, among which neurologic ones, while rare, are increasingly being recognized and remain incompletely characterized. Herein, we report five patients with nonmelanoma cancers who developed weakness after receiving CPIs. The etiology was attributed to radiculoneuritis (one patient), myositis (one patient), Miller Fisher/myasthenia gravis (MG) (one patient), neuropathy/myositis/MG (one patient), and myositis/MG (one patient). Weakness developed after a median of two doses (range: 1–3) and 4 weeks (range: 3–10) from initiation of therapy. Two patients had severe manifestations without improvement while the other three experienced partial improvement despite discontinuation of the CPI (s) and initiation of immunosuppressive therapy. A review of literature identified 62 similar cases. This report highlights the challenges in the diagnosis and management of neurologic adverse events related to the use of CPIs. It also addresses the crucial need for early recognition, proper workup, and better biomarkers to help improve the outcomes of these adverse events.
The use of checkpoint inhibitors (CPIs), targeting cytotoxic T-lymphocyte antigen-4 (CTLA-4) and/or programmed death-1 (PD-1), in cancer therapy, is rapidly expanding, and with it, immune-related adverse events (irAEs) affecting various organs are being increasingly recognized. Among those, neurological irAEs remain incompletely characterized but in general have a favorable prognosis. Analysis of 59 clinical trials identified their incidence at 3.8%, 6.1%, and 12% for anti-CTLA-4, ani-PD-1, and combined blockade, respectively. A single-institution retrospective data analysis of 347 patients receiving CPIs reported an overall incidence rate of 2.9%. However, they are hard to diagnose, occasionally lead to severe manifestations, and often complicate cancer therapy.
Materials and Methods
The neurology inpatient consultation service at our institution encountered five patients who developed various degrees of weakness after initiating checkpoint inhibition (CPI) immunotherapy. We summarize the cases below along with a literature review of patients presenting with similar CPI-induced symptoms.
Our review of literature was conducted in December 2018 by searching PubMed for relevant keywords including “immunotherapy,” “checkpoint inhibitor,” and individual CPIs combined with “myasthenia gravis,” “myositis,” or “neuromuscular.” Only full-text articles in English language were included.
Cases series patients
A 60-year-old male received anti-CTLA-4 therapy for recurrent metastatic adenoid cyst carcinoma of the submandibular gland. The drug was discontinued after three doses when he developed common terminology criteria for adverse events (CTCAE) grade 2 eyelid swelling with biopsy showing lacrimal gland granulomatous inflammation. Ten weeks after starting ipilimumab, he was hospitalized with fatigable weakness in his neck extensors. He had normal creatinine kinase (CK) and myasthenia gravis (MG) antibody panel serum levels. Cerebrospinal fluid (CSF) analysis showed mild protein (65 mg/dL) and immunoglobulin G (IgG) index elevations without pleocytosis. Motor and sensory nerve conduction studies (NCSs) and repetitive nerve stimulation (RNS) were normal. Electromyography (EMG) of paraspinal muscles was consistent with CTCAE grade 1 radiculoneuritis. He refused any immunosuppressive therapy and had partial improvement with physical therapy. Re-staging scans revealed remission of the primary malignancy and stable metastases in the lungs, liver, bones, and lymph nodes.
A 65-year-old female received anti-PD-L1 and anti-CTLA-4 for recurrent leiomyosarcoma in her ribs. Three weeks after receiving her first and only dose of immunotherapy, she was hospitalized with CTCAE grade 3 weakness in the upper and lower extremities plus CTCAE grade 2 myalgias, difficulty breathing, bilateral eye ptosis, dysphagia, dysphonia, dysarthria, tongue weakness, and bilateral facial weakness. Pulmonary function tests were borderline with forced vital capacity (FVC) of 12 mL/kg and negative inspiratory force of − 14, but she did not require intubation. She had CTCAE grade 4 CK elevation (19,794 U/L) without electrocardiogram changes or troponin elevation. MG/Lambert–Eaton syndrome (LES) antibody panel revealed elevated acetylcholine receptor (Ach-R) binding antibody (0.55 nmol/L), Ach-R modulating antibody (94%), and striational muscle (SM) antibody titer (1:15360). NCSs were normal but RNS showed postexercise decrements, suggestive of neuromuscular junction (NMJ) dysfunction. EMG showed muscle membrane irritability in the left arm and leg muscles consistent with myositis. Therefore, she was treated with intravenous (IV) solumedrol (1 g daily) followed by five cycles of plasma exchange (PEx) and then standard dose intravenous immunoglobulin (IVIg) at 0.4 g/kg daily for 5 days without improvement. Her CK levels trended down, but her MG antibody panel was not repeated as it does not typically reflect response to therapy in MG. She was eventually discharged home on an oral steroid taper and with a percutaneous endoscopic gastrostomy (PEG) after a 3-month hospitalization. Computed tomography (CT) scans during hospitalization showed further enlargement in the rib lesions. No further immunosuppressant therapy was deemed necessary given the decline in her functional status with a plan to follow-up outpatient.
A 68-year-old male received anti-PD-1 therapy for recurrent metastatic squamous cell carcinoma of the tongue. Four weeks later, having received two doses of the drug, he developed CTCAE grade 2 neck muscle weakness and left eyelid drooping and was eventually hospitalized with CTCAE grade 4 CK elevation (11,274 U/L). MG/LES antibody panel showed elevated anti-SM antibody titer (1:61,440). By then, he had developed left ptosis, left facial droop, neck extensor weakness, and bilateral proximal leg weakness. NCS and RNS were unremarkable, while EMG of the cervical paraspinal, left arm, and leg muscles was consistent generalized myositis. CK levels gradually normalized and he experienced significant clinical improvement from prednisone (80 mg daily for 25 days), but it was recommended to not restart immunotherapy given the involvement of bulbar muscles. CT scan after hospital discharge showed increase in the pulmonary lesions.
A 66-year-old male received anti-PD-1 therapy for recurrent metastatic renal carcinoma. Three weeks later, having received two doses of the drug, he was hospitalized for CTCAE grade 2 generalized upper extremity and facial muscle weakness, ophthalmoplegia, fatigable right eye ptosis, and hoarseness. Workup revealed CTCAE grade 4 CK elevation (6503 U/L) level, grade 3 ALT (458 U/L) elevation, and grade 2 AST (171 U/L) elevation. In addition, Ach-R binding antibody (4.40 nmol/L), Ach-R modulating antibody (99%), and anti-SM antibody titer (1:122,880) levels were elevated consistent with MG. NCS showed mild axonal sensorimotor polyneuropathy, RNS testing was normal, and EMG showed generalized myositis. He was started on IV solumedrol (70 mg daily) followed by five cycles of PEx and three doses of IVIg, with partial recovery and normalization of CK levels. Further, immunosuppressive was deferred to outpatient follow-up, but his residual weakness precluded him from receiving any further therapies, and he expired 3 months later from disease progression, with CT scan during hospitalization showing increase in the size of his primary and metastatic lesions as well as a new omental lesion.
A 67-year-old male received anti-PD-1 therapy for recurrent metastatic renal carcinoma. Six weeks later, having received two doses of the drug, he was hospitalized with CTCAE grade 2 diplopia, ophthalmoplegia, hoarseness, dysphagia, and bifacial and neck flexion muscle weakness. He was intubated for impending respiratory failure with low FVC (8.5 mL/kg) and NIF (−5). He had elevated Ach-R binding antibody (2.01 nmol/L) and Ach-R modulating antibody (99%) consistent with MG. CSF examination revealed cytoalbumin dissociation (protein = 110 mg/dL, white blood cell = 0). NCS showed conduction blocks while RNS and EMG studies were normal, which along with ophthalmoplegia, and CSF findings was consistent with Miller Fisher syndrome, although his serum GQ1b IgG was negative. He was started on IV solumedrol (500 mg twice daily), pyridostigmine, and PEx with minimal improvement but was successfully extubated after seven cycles of PEx. MG antibodies were undetectable 17 days after admission, and he was given one dose of each of rituximab and infliximab. His respiratory status worsened again despite being on high-dose oral steroids and repeat EMG/NCS showing generalized myositis, without evidence of worsening demyelinating features or NMJ dysfunction on RNS. He was reintubated, received further PEx and IVIg without improvement, and ended up receiving a tracheostomy and a PEG tube before being discharged to a long-term acute care facility with outpatient follow-up to discuss if further immunosuppressive therapy is indicated. CT scan during hospitalization showed stable primary and metastatic lesions.
A literature review for published cases of CPI-induced myositis and/or MG identified 62 cases of irAEs of myositis, MG, or both.[4–45]Table 3 summarizes the baseline characteristics and kinetics of the neurotoxicities and outcome of irAEs in this literature review.
CPI therapy has revolutionized cancer therapy but has also contributed to the emergence of fairly unique, and at times unpredictable toxicities, typically occurring in the skin, liver, gastrointestinal tract, or endocrine system. CPI-induced neurotoxicities are uncommon with neuropathy, myositis, and MG, representing the most common neurological irAEs. They generally have a favorable prognosis but can occur concomitantly with other organ toxicities, such as MG with myocarditis which worsens their prognosis. Furthermore, we show here that they can be severe enough to preclude further cancer therapy and/or lead to life-threatening sequelae despite expedited intervention.
Our study is limited by the lack of tissue diagnosis. Muscle and/or nerve biopsy is generally preferred for more accurate diagnosis but was not attempted in any of our patients due to various factors: patient/family preference, highly suggestive serologic/electrodiagnostic results, and unclear impact on management or patient outcome if done. It is possible, albeit unlikely, that the MG manifestations seen in patients 2, 4, and 5 are paraneoplastic manifestations of the primary malignancy. Both leiomyosarcoma and renal cell carcinoma have been associated with paraneoplastic myasthenic syndrome in rare cases.[48,49] However, the development of neurologic symptoms shortly after initiation of immunotherapy and not after cancer diagnosis fits more with the timeline of CPI-induced neurotoxicities, especially given the rarity of such paraneoplastic myasthenic associations with these tumors. Our literature review conclusions are also limited by the retrospective nature of our search, but it is the largest to compile such cases to date with 62 cases reviewed, and our search did yield several important insights into these poorly described irAEs.
The mechanisms of irAEs are not completely understood but are dose related, particularly in CTLA-4 blockade cases, and vary with the histology of the malignancy treated, especially in PD-1/PD-L1 blockade cases. They seem to be driven by the same immune mechanisms that drive them to enact their therapeutic effects.
Muscle and nerve biopsies along with the laboratory data (positive antibodies) gave some clues to the pathogenesis and treatment paradigms. Targeting lymphocytes (T- and B-cells) and cytokines has been a main stay for immune-mediated toxicities. Treatments for autoimmune neuromuscular diseases described in this case series when seen in noncancer patients include immunomodulation (steroids, steroid-sparing agents such as rituximab, IVIg, and PEx) to affect lymphocytes, cytokines, and the complement system.
The kinetics of neurological irAEs is not fully elucidated. In our case series, they were similar to those seen in the literature review with neurotoxicity occurring after a median of two doses and a latency of 4 weeks [Tables 2 and 3] in both. The largely favorable prognosis of these toxicities seen in the literature review with improvement in 82% of the cases including 20% with complete recovery [Table 3] was not replicated in our case series where 40% had no improvement and none had complete recovery [Table 2]. This is not too surprising since the majority of the patients in our case series had bulbar muscle involvement which correlates with poor outcome.
The occurrence of an irAE is believed by some to correlate with efficacy of CPI therapy since the immune system was properly stimulated. However, in our case series, only 1/5 patients had partial response to CPI on restaging scans, 1/5 had stable disease, and 3/5 had progressive disease [Table 1].
The search for immunotherapy response or toxicity biomarkers is ongoing but remains challenging. Various tissue and serum biomarkers have been proposed and are already being used to guide therapy. Examples include PD-L1 tumor expression level, tumor-infiltrating lymphocytes level, degree of tumor's microsatellite instability and/or mutational load, serum interleukin 6 levels, and to a lesser extent circulating tumor cells or DNA. However, their use has yet to be clinically validated across different malignancies, reproduced among different immunostimulating therapies, in the context of a dynamic immune system with incompletely understood regulatory pathways. There are no known biomarkers associated with neurological irAEs. Elevated CK levels can be seen with CPI use [4,6,7] and in the proper context may represent CPI-induced myositis, but it is unclear if its level correlates with the severity of myositis. Anti-Ach-R antibody elevation has been described before, but it can also be present subclinically before CPI therapy. Anti-SM antibody elevation, seen in two patients in our case series, is not a common finding but is not typically checked either. It has only been described once before in a patient who had pre-CPI treatment elevation that increased after immunotherapy; therefore, more data are needed before its utility as a CPI-induced neurotoxicity biomarker can be verified.
Early recognition of neurological irAEs is challenging due to their vague presentation. Any new reported weakness by the patients receiving CPIs, especially when accompanied by cranial nerve findings (ptosis and facial weakness) and bulbar symptoms (swallowing and speech changes), should be promptly evaluated, ideally by the neurology service. Neuroimaging (brain and spine), spinal fluid analysis, blood work (for appropriate antibodies), and EMG/NCS studies may all be needed. Differential diagnosis of tumor infiltration, carcinomatosis meningitis, and infections should be promptly excluded. Immunomodulatory treatments should be initiated with steroids being the main stay. Close monitoring of clinical and pulmonary status in the intensive care unit may be warranted for patients with bulbar weakness and respiratory compromise. It may be helpful to check the serum levels of CK and MG antibody panel including anti-SM antibody before initiation of CPIs and during therapy if clinically indicated, but their cost-effectiveness is not fully clear at the moment.
Standardized management guidelines of neurological irAEs are lacking although the recently published American Society of Clinical Oncology guidelines nicely summarized the various strategies that can be employed when dealing with such toxicities in general. Similar to systemic irAE management, it relies primarily on discontinuation of the offending agent with or without accompanying steroids and with or without steroid-sparing immunosuppressive therapy depending on the severity of symptoms.[57–60] The challenge is that immunosuppressive therapy counteracts the anticancer effects of CPIs and at times is associated with its own set of neurotoxicities (e.g., steroid-induced myopathy), which is why it is advised to use it sparingly when irAEs occur. However, the duration of therapy is unknown and treatment algorithms detailing in what order immunosuppressive agents should be tried are lacking. In addition, there are some therapeutic challenges unique to CPI-induced neurotoxicity as seen in some of our cases. Acuteness of the clinical presentation, cranial nerve findings, bulbar weakness, and respiratory muscle weakness require rapid evaluation and quick initiation of the fast-acting immunomodulatory therapies. Management of CPI-induced myositis and MG is not much different than standard therapy for these conditions. According to the recently published consensus-based guidelines for MG management, steroids and immunosuppressive therapy should be used in all patients who do not improve on pyridostigmine. Other nonsteroidal immunosuppressants include azathioprine and cyclosporine, for which randomized clinical trial data exists, as well as mycophenolate, methotrexate, and tacrolimus. Rituximab and cyclophosphamide, on the other hand, are reserved to patients with refractory MG. Finally, the guidelines support the role of IVIg or PEx in treating MG patients with respiratory insufficiency or dysphagia and are thus recommended for patients who manifest these symptoms after starting CPI therapy. There are no standardized treatment guidelines for myositis given that they represent a heterogeneous group of disorders, but similar immunosuppressive-based therapies are typically used in managing them. IVIg, steroids, and nonsteroidal immunosuppressants are widely acceptable therapeutic modalities.[62–65] However, IVIg with its multiple mechanisms of action that effects B- and T-cells and neutralizes cytokines complement and autoantibodies without significant long-term immunosuppression might offer an advantage compared to other modalities.[66,67] Neuropathy manifestations were seen in three patients in our case series. Demyelinating neuropathies, such as Guillain–Barre syndrome (GBS) or its variant Miller Fisher seen in case 5, are the more commonly described CPI neurotoxicities, and their management is focused on preventing respiratory failure, its most serious complication, with PEx (5 cycles of 50 mL/kg over 1–2 weeks) and IVIg (0.4 g/kg daily for 5 days).[69,70] Steroids are still used in GBS, but Cochrane systemic review did not support their role. The axonal polyneuropathy described in case 4 is likely an incidental subclinical manifestation of age and prior chemotherapy use. However, axonal neuropathies have been reported following CPI therapy and can be steroid-responsive in that setting. Acquired axonal neuropathies, such as demyelinating ones, can have immune pathogenesis. Steroids can also be used to manage radiculoneuritis – a nonspecific terminology describing nerve root inflammation – seen in case 1, but only in the acute setting and for cases where other etiologies of inflammation (e.g., infection) have been ruled out.
We propose the following algorithm [Figure 1] based on awareness of the possible neurotoxicities and early consultation of neurological services. Broad diagnostic and treatment goals were used although any such generalization should be patient specific given the significant variation of neuromuscular weakness both in severity and presentation. Early diagnosis, optimizing vitamin D status, and early initiation of immunosuppressive therapy are recommended. The potential role of low vitamin D status to cause muscle weakness especially in cancer patients and possible positive immunomodulatory effects were considered[74,75] for the above recommendations. IVIg induction dose of 2 g/kg divided over 2–5 days and maintenance 1 g/kg, with a maximum of 60 g given every 2–3 weeks depending to the clinical response, can be considered. Initiation of PEx early for patients with marked weakness, respiratory compromise, and swallowing difficulty is preferred. While high-dose steroids are generally contraindicated in severe MG patients, the accompanying myositis and acute triggered inflammatory cascade of events might differ from typical MG where such high CK levels are generally not seen. Combination approach such as steroids and IVIg might be worth considering. IVIg could also give the immunological buffer for tapering steroids. We recommend steroid taper over 3–4 weeks depending on the clinical improvement and reduction of elevated CK levels assuming this is a possible monophasic illness, with or without few doses of IVIg maintenance. Monitoring for resolution of elevated paraneoplastic, MG antibodies including anti-striational antibodies is reasonable. Infliximab has been reported to cause GBS in rare case reports, and while it is described for other organ CPI-induced toxicities, it is recommended not to be used in neuromuscular weakness until further research. Rituximab and tacrolimus are some of the options for patients with ongoing significant weakness,[76,77] given relatively faster onset of these agents in patients with rapid progression of symptoms. However, research into upfront use with few doses of these agents in selected patients with very profound rapid progression of symptoms is needed. Prompt cardiac assessment with telemetry monitoring, 2-D echocardiogram, and serum brain natriuretic peptide and troponin levels in patients with elevated CK and respiratory symptoms is recommended to rule out concomitant myocardial immunotoxicity.[78,79] Thyroid function tests are also recommended to check for additional immunotoxicities as thyroid dysfunction can cause neuromuscular weakness.
We start to understand the spectrum of neurotoxicities related to CPIs, but many questions remain unanswered. We still do not know if CPIs unmask an existing subclinical neurologic pathology rather than cause it, and in the latter scenario, we do not know why certain individuals are at higher risk than others. We do not know whether the underlying malignancy plays a role in the type of neurotoxicities or toxicities that might develop upon CPI use. We also do not know with certainty if CPI-induced neurotoxicity is self-resolving (monophasic illness) or one that requires long-term immunosuppressive regimens as the ones used in nonmalignant settings. It is also unclear if these toxicities are reversible with current evidence, suggesting an irreversible component. More research is needed to address these issues especially with the increased use of immunotherapeutics for various malignancies in the upfront and recurrent disease settings. Avoidance of long-term immunosuppressive regimens is also important to prevent infections and other known complications. Until then, a high level of suspicion is crucial for prompt recognition and early intervention to best control the potentially substantial morbidities associated with CPI-induced neurotoxicities.
Financial support and sponsorship
The authors disclosed no funding related to this article.
Conflicts of interest
The authors disclosed no conflicts of interest related to this article.
For reprints contact:email@example.com