Guillain-Barré Syndrome-Like Polyneuropathy Associated with Immune Checkpoint Inhibitors: A Systematic Review of 33 Cases

Immune checkpoint inhibitors (ICIs) have been increasingly used in the treatment of various types of tumors with favorable results. But these treatments also led to a variety of immune-related adverse events (irAEs). Neurological irAEs such as Guillain-Barré Syndrome are rare and may have serious consequences once they occur. A systematic literature search was performed in PubMed and Embase for all case reports of GBS associated with ICIs published in English reporting on human beings from 1990 up to date. A total of 30 case reports (total patients = 33) were used for final analysis. The included cases were from 11 countries, covering 10 tumor types, with melanoma accounting for the largest number. The mean age was 62.2 ± 11.1 years old, and males were dominant (male: 26 and female: 7). The median time of initial symptoms was 8.2 weeks after the 1st dose of ICIs. The most common manifestations of GBS associated with ICIs were weakness, hyporeflexia or areflexia, and paresthesia in order. The GBS subtypes suggested by electrophysiological results were acute inflammatory demyelinating polyneuropathy (AIDP), acute motor axonal neuropathy (AMAN), and Miller Fisher syndrome (MFS). The protein level of CSF in patients with GBS related to ICIs was 180.68 ± 152.51 mg/dl. Immediate termination of ICIs followed by intravenous immunoglobulin was the preferred treatment option. 72.7% of patients recovered or had residual mild dysfunction after treatment. Elderly male patients with melanoma were most likely to develop ICI-related GBS. The specific neurological symptoms, CSF analysis, and electrophysiological examination were important means of diagnosis.


Introduction
In the last decade, with a better understanding of the factors that promote or inhibit T cell response, great progress has been made on tumor immunotherapy. Immune checkpoint inhibitors (ICIs) have become a powerful clinical strategy for treating cancer, including an antibody targeting cytotoxic T lymphocyte-associated antigen-4 (CTLA-4, e.g., ipilimumab), antibodies directed against programmed cell death protein-1 (PD-1, e.g., nivolumab, pembrolizumab, and cemiplimab), and anti-PD-1 ligand (PD-L1, e.g., atezolizumab, durvalumab, and avelumab) [1]. These drugs can be used alone or in combination with other immunotherapy [2] or chemotherapy [3] to improve the survival of cancer patients. Currently, it has been used for the treatment of tumors of lung, kidney, liver, bladder and breast cancer, melanoma, and lymphomas [4][5][6]. ICIs improve the prognosis and qual-ity of life of patients, whereas the increase of use also brings various immune-related adverse events (irAEs). The dermatologic, gastrointestinal, pulmonary, hepatic, and endocrine systems were most frequently involved.
Cases of neurological irAEs are rare, accounting for less than 3% [7]. So far, the best-characterized central nervous system irAEs are encephalitis and meningitis. And neurologic irAEs known to be most relevant to the peripheral nervous system are peripheral neuropathies, GBS, myasthenia gravis, and myositis [8]. Once they occur, such as encephalitis, Guillain-Barré syndrome, or myasthenia gravis, they can develop into serious consequences or even death.
Epidemiology shows that nearly two-thirds of patients with GBS have a recent history of infection before the illness [9]. GBS with potentially life-threatening consequences occurs in approximately 0.1-0.2% of patients treated with ICIs [10]. To date, information on the incidence, characteristics, and outcomes of GBS associated to ICIs treatment is very limited. And the available information varies widely in diagnosis and treatment. Multidisciplinary treatment of tumors urgently requires neurologists and oncologists to accurately understand the clinical manifestations and treatment of ICI-related GBS.
This review summarized the published data on GBS or GBS-like disease occurring in patients after treatment with ICIs from 1990 up to date and analyzed their time patterns of occurrence, clinical presentation, diagnosis, treatment, and prognosis.

Methods
A systematic literature search was performed in PubMed and Embase for all case reports of GBS associated with ICIs published in English reporting on human beings from 1990 up to date. For the case reports search, the keywords used were as follows: ["Guillain-Barré Syndrome" OR "acute inflammatory demyelinating polyradiculoneuropathy" OR "Miller Fisher Syndrome" OR "acute motor axonal neuropathy" OR "acute motor-sensory axonal neuropathy"] AND ["Immune Checkpoint Inhibitors" OR "Immune Checkpoint Blockers" OR "PD-L1 Inhibitors" OR "Programmed Death-Ligand 1 Inhibitors" OR "CTLA-4 Inhibitors" OR "Cytotoxic T-Lymphocyte-Associated Protein 4 Inhibitors" OR "PD-1 Inhibitors" OR "Programmed Cell Death Protein 1 Inhibitor"]. Abstracts of medical conference were excluded. For each case, we extracted data on demographics and clinical manifestations and adjuvant examinations (imaging, cerebrospinal fluid, and electrophysiology). If gender, age, GBS clinical variation [11], electrophysiological subtype [12], or results of the relevant examination were not explicitly reported in the article, this case could not be considered for analysis. The search was conducted by Yan Li and Xiuchun Zhang. The selection of the articles should be agreed upon by the above two persons.
GraphPad Prism 8 was used for statistical analysis, and continuous data were expressed in the form of mean ± standard deviation or median. P < 0:05 was considered statistically significant. Since the proportion of males and females in the included cases was significantly different, unpaired t -test was used to analyze the ages of males and females.
Because the patient's personal information was provided in the original case report, authorization from the Ethics Committee was not required for this study.

Results
Using the search terminology, 38 case reports were identified from our database search, covering the period from January 2008 to February 2021. Three patients were excluded due to lack of age and CSF protein concentration, respectively. One case report was excluded because it was written in Japanese. Two case reports were not included because the diagnosis of GBS was ambiguous due to a disease progression similar to acute-onset CIDP. In addition, two patients with a history of GBS had no serious toxicities or deterioration of the previous autoimmune disorders after ICI therapy.
The two patients were also excluded. A total of 30 case reports (total patients = 33) were used for final analysis . According to the diagnostic criteria of GBS in NINDS, all the above 33 cases were consistent with the features of GBS. The clinical data and diagnostic details of all included patients were summarized in Tables 1-3. 3.1. Demographic Characteristics. GBS cases (n = 33) were from the United State (n = 13), United Kingdom (n = 3), Japan (n = 2), Italy (n = 2), Belgium (n = 2), Australia (n = 2 ), China (n = 1), Greece (n = 1), Netherlands (n = 4), France (n = 2), and German (n = 1) ( Table 1). Of the 33 cases, twenty-six of these cases were male and seven were female, with an average of 62:2 ± 11:1 years (median: 65 years and range: 37-81 years). There was a male preponderance in 33 ICI-associated GBS patients we collected, with 3.7 times as many cases as female (26 vs. 7 cases: 78.8% vs. 21.2%). There was a significant difference between male and female ages at onset (mean: 64:4 ± 10:3 vs. 54:1 ± 11:1 years, P = 0:0278). The reports of comorbidities were variable, and no epidemics of specific diseases had been observed, so we did not analyze the comorbidities.
MRI was performed in 54.5% (18/33) of cases, of which both brain and spinal cord were examined in 8 cases and only spinal cord was 10 cases. MRI results showed nerve root involvement in 4 cases [16,27,29,31], severe spinal stenosis in 1 case [19], cranial nerve involvement in 1 case, and normal or no metastatic signs in the rest [29].

Treatment and Prognosis of GBS.
Twenty-seven cases (81.8%) were treated with intravenous immunoglobulin (IVIG), sixteen of the 27 patients with a combination of steroid therapy and one with PLEX. Seven of the 27 patients were treated with all three treatments. Five cases were treated with steroid therapy alone, and one case was treated with steroid combined with PLEX. No patients were treated with PLEX alone.

Discussion
The effects of checkpoint inhibition affect a wide range of system and trigger a wide range of autoimmune toxicity. The exact mechanism of neurological irAEs in ICI-treated patients is unknown [44]. The existence of shared antigens between the tumor and itself may be one possible mechanism, such as gangliosides found in both melanoma and Schwann cells [45]. ICI-related GBS is rare as a type of neurological irAEs, but it can develop into life-threatening consequence once occurred. Diagnosis of ICI-related GBS should be made in the shortest possible time so as not to delay the administration of immune-regulation therapy.
There was a male preponderance in 33 ICI-related GBS patients we collected, with 3.71 times as many cases as female (male 26 and female 7). However, among GBS triggered by infection, males were affected only 1.5 times more frequently than females [11]. An epidemiological study on the risk factors of melanoma showed that the incidence of melanoma of men was almost three times that of by women by the age of 75 [12]. The incidence of lung tumor in men has historically been higher than in women, although the incidence of lung tumor in women has risen since the 1960s, especially in younger women [46][47][48]. Perhaps the higher incidence of cancer in men, resulting in more opportunities for men to use ICIs, was one of the reasons for the higher incidence of ICI-related GBS. Besides, ICIs were more effective for male cancer patients than female patients [49]. And the development of irAEs was related to the beneficial effects of immunotherapy in malignant tumors, especially in advanced-stage melanoma, advanced and metastatic NSCLC, and advanced renal cell carcinoma (RCC) [50]. The high effectiveness of ICI treatment in male patients may be also responsible for the male preponderance in ICI-related GBS.
In this study, melanoma had the largest number of tumor type that caused GBS after ICI treatment. Neurological irAEs were rare, with an incidence of less than 3% [8], whereas the overall incidence of severe nerve injury in melanoma patients treated with nivolumab with or without ipilimumab reached 0.93% [7], nearly one-third of the total neurological irAEs. Well know, GBS is a multifactorial autoimmune disorder, cell-mediated immunity plays an important role in immunopathology of all types of GBS. The activation of T cell caused by bacterial and virus leads to the production of cytokines and the release of free radicals, thus resulting in segmental demyelination [51]. And the cross-reaction between B-cell autoantibodies and axon gangliosides leads to axonal degradation [51]. Melanocytes and Schwann cells originate from the neural crest and have many common epitopes in humoral and cellular immune responses [52]. T cells regulate the degree of initial response of T cells by upregulating CTLA-4, while PD-1 inhibits the response of T cells in peripheral 13 BioMed Research International tissues and plays an important role in immune self-tolerance [53]. Due to the cross-reaction of molecular mimicry, T cellmediated autoimmunity against melanoma cell antigens may also have an effect on the myelin antigens on Schwann cell membranes. In addition, the response rate of melanoma to ICIs was higher than other tumors [54]. A FAERS database-based clinical study also showed that patients with melanoma or non-small-cell lung cancer maybe at higher risk of fatal neurologic AEs [55]. Therefore, the author speculated that melanoma patients were more likely to develop GBS after ICI treatment than other types of tumors.
Existing reports suggested that the overall incidence of neurological adverse events (nAEs) at all levels was 3.8% for anti-CTLA-4 and 6.1% for anti-PD-1/PD-L1 [56]. CTLA-4 and its ligands are only expressed on immune cells, while PD-1 and PD-L1 (ligands of PD-1) are expressed on both immune and nonimmune cells in peripheral tissues [57]. The difference in spatial distribution between the CTLA-4 and PD-1 pathways may explain the high incidence of GBS in patients treated with nivolumab.
Many literatures had reported the timing of irAEs onset after the initiation of ICI treatments. The skin manifestations appeared at 2-3 weeks, and immune-mediated colitis, hepatitis, pneumonitis, and nephritis appeared approximately 5-10 weeks, 12-16 weeks, 8-14 weeks, and 14-42 weeks, respectively. Endocrine dysfunctions appeared from 9 weeks [58]. At present, there is no literature on the time to onset of neurologic symptom related to GBS caused by ICIs. The initial time of symptoms in 33 patients in this article was analyzed, and the results suggested that the median time was 8.2 weeks after the initiation of ICI treatments. Compared with GBS caused by infectious triggers, neurologic symptom appeared much later. This conclusion was expected to be helpful for the rapid diagnosis of ICIs-GBS.
Of the 33 GBS related to ICI patients we collected, the initial symptoms were very similar to infection-triggered GBS. Lumbar puncture was performed, and CSF analysis revealed elevated protein levels and albuminocytologic dissociation. A study of 962 patients with infection-induced GBS showed that the average protein level was 113:8 ± 11:8 mg/dl (range: 18-450 mg/dl) [59]. In this study, the average protein level of CSF in patients with GBS related to ICIs was 180:68 ± 152:51 mg/dl (range: 37-680 mg/dl). From a numerical point of view, the levels of CSF protein level of ICI-related GBS seemed to be higher than those of GBS induced by infection. Although it was not possible to distinguish infection-induced GBS from ICI-induced GBS based on CSF protein levels, lumbar puncture and protein-cell-separation were valuable in differentiating a variety of diseases such as spinal cord compression, metabolic diseases, side effects of drugs, vasculitis, and chronic inflammatory demyelinating polyneuropathy. In addition to elevated protein levels, lymphocytosis may also occur [37]. A similar pattern was observed in this study, with 4 patients showing a mild lymphocytosis in the CSF. However, multiple cases had been reported of lymphocytosis in the CSF of patients with infection-induced GBS [60][61][62][63]. Obviously, lymphocytosis in the CSF was not a characteristic of ICI-related GBS.
The electrophysiological results detailed in this study indicated that ICI-related GBS was a generalized, sensorimotor polyneuropathy characterized by mixed axonal/demyelination. An electrophysiological study of ICI-related peripheral neuropathy showed that immune-mediated neuropathy mainly manifested as demyelination of motor nerves, followed by length dependent axonal loss in sensory nerve [60]. The results of this study were in part consistent with the above conclusions. No matter what kind of electrophysiological changes were closely related to the autoimmune response of ICIs against peripheral nerve tissue, the onset of symptoms of neuropathy was closely related to ICI treatment, and corticosteroid or immune-regulatory therapy had good results.
In the treatment of all 33 patients, the immediate discontinuation of ICIs was an uncontroversial decision. Most patients received IVIG after discontinuation, with 16 patients receiving both IVIG and steroid therapy. Despite steroids were generally not recommended for treatment in infection-induced GBS, in ICI-related GBS, both American Society of Clinical Oncology (ASCO) Clinical Practice Guideline and National Comprehensive Cancer Network (NCCN) guidelines stated that trial of (methyl) prednisolone 1-2 mg/Kg was reasonable [64,65], especially when CSF pleocytosis was higher than being anticipated for GBS [10]. And if the symptoms deteriorated, plasma exchange or IVIG treatment could be considered [66].
Among the 33 ICI-related GBS patients, 5 patients developed tumor progression after discontinuation of ICI treatment. One of the four patients with progressed melanoma rechallenged different class of immunotherapy, with significant and sustained response nearly 1 year later and no recurrence of GBS-like neuropathy. Whether or not to retreat with ICIs and whether to retreat with the same or different ICIs are challenges for oncologists. There is also limited data on the clinical efficacy and safety of retreatment. The current guidelines suggested that corticoid therapy and temporary or permanent discontinuing ICIs were required for grade ≥ 2 irAEs and permanently discontinuing ICI treatment for grade 4 irAEs [65,67]. A study had shown that the recurrence rate of the same irAEs resulting in discontinuation of ICI treatment in cancer patients who rechallenged the same ICI was 28.6% (anti-PD-1 or anti-PD-L1 monotherapy), 47.4% (anti-CTLA-4 monotherapy), and 43.5% (combination therapy), respectively [66]. The recurrence rate of irAEs varied depending on the organ involved in the initial irAEs, with gastrointestinal irAEs having the highest recurrence rate [68]. And the variables associated with a higher recurrence rate of irAEs were anti-CTLA-4 regimen, age, colitis, hepatitis, and pneumonia, in order [68]. In addition, the duration from ICI discontinuation to rechallenge, and the severity of the initial irAEs did not predict whether irAEs would reappear after rechallenge of ICIs [69]. However, no relevant literature has been reported on whether discontinuation due to ICI-related GBS can rechallenge the same or different ICIs again.
In addition, paraneoplastic peripheral neuropathy should be excluded in the diagnosis of ICI-related GBS. Paraneoplastic peripheral neuropathy is a remote effect of the malignancy 14 BioMed Research International mediated by the immune system. It develops prior or during a cancer and is independent of tumor infiltration or cancer therapy [70]. Paraneoplastic sensory neuropathy is the most frequent in this group of disorders, and motor, autonomic, or central nervous systems are also involved [71]. Date on the treatment of paraneoplastic peripheral neuropathy is limited, and the combination of malignancy therapy with immunomodulatory therapies such as corticosteroids, IV immunoglobulin, or immunosuppressants may be effective [72]. However, ICI-related GBS occurred in malignant tumors after ICI treatment, and the immunomodulatory therapeutic effect was obvious. This is the most obvious difference between the two.
Here are also several limitations in this study. First, the number of the included cases was small which limited us to perform subgroup analysis. Second, patients with various cancers were included, which might have bias in the incidence of some adverse effects.

Conclusion
Elderly male patients with melanoma were most likely to develop ICI-related GBS. And the median duration was 8.2 weeks after the initial ICI treatment. In order to make a final diagnosis, physicians need to collect specific neurological symptoms and signs and combine them with CSF analysis and electrophysiological examination. In addition, imaging is required to exclude tumor metastasis. Immediate termination of ICIs followed by IVIG in combination with high-dose steroids therapy or PLEX and supportive treatment could lead to a better prognosis.