Paraneoplastic Neuromyelitis Optica Spectrum Disorder Related to Glucose-regulated Protein 78 (GRP78) Autoantibodies in a Patient with Lynch Syndrome-associated Colorectal Cancer

Abstract

Neuromyelitis optica spectrum disorders have been previously reported in a paraneoplastic context, although there is no clear consensus on their pathogenesis. We herein report a case of aquaporin-4 antibody-positive neuromyelitis optica spectrum disorder in a 64-year-old woman with colorectal cancer. She underwent tumor resection, resulting in serum aquaporin-4 antibody titers subsequently becoming negative. Serum samples were also positive for glucose-regulated protein 78 antibody, which has recently been suggested to be a novel factor in the disruption of the blood-brain barrier. Serological and pathological investigations in this case highlight the role and involvement of aquaporin-4 and glucose-regulated protein 78 antibodies in paraneoplastic conditions.

Introduction

Neuromyelitis optica spectrum disorder (NMOSD) is an inflammatory and demyelinating disease that presents with optic neuritis and/or longitudinally extensive transverse myelitis (LETM) and is classified separately from multiple sclerosis as a disease with aquaporin-4 (AQP4) antibodies and an often severe clinical presentation (1,2). NMOSD is known an idiopathic disease that typically occurs in young women, and although its pathophysiology has been gradually elucidated, the effective prevention of its recurrence is still challenging.

Shimizu et al. recently revealed the presence of glucose-regulated protein 78 (GRP78) antibodies, which are involved in blood-brain barrier (BBB) disruption, in the sera of NMOSD patients and those with paraneoplastic cerebellar degeneration and Lambert-Eaton myasthenic syndrome (PCD-LEMS) (3,4). GRP78 autoantibodies bind to cell surface GRP78 on BBB-endothelial cells and increase BBB permeability, thus enabling AQP4 antibodies produced in the periphery to enter the central nervous system (CNS), subsequently leading to the onset of NMOSD. GRP78 autoantibodies have also been identified in patients with various cancers (5).

We herein report a case of paraneoplastic NMOSD caused by Lynch syndrome-associated colorectal cancer, in which GRP78 and AQP4 antibodies were confirmed in serum, with the AQP4 antibodies becoming negative following tumor resection.

Case Report

A 64-year-old woman was referred to our hospital because of thoracic dysesthesia and progressive paraplegia that had started 2 weeks earlier. She had a history of uterine and colorectal cancers associated with Lynch syndrome, diagnosed by the genetic examination of MSH2 exon 1-6 large deletion 16 years prior, and had undergone tumor resection. She had no history of either myelitis or optic neuritis.

On admission, she presented with total sensory loss below the T5 level, paraplegia with an inability to walk, and bladder and bowel dysfunction. Deep tendon reflexes were normal in the upper extremities but exaggerated in the patellar tendon, and pathological reflexes in the lower extremities were positive bilaterally.

Spinal cord magnetic resonance imaging (MRI) revealed LETM between the C7-T9 levels, with gadolinium enhancement of the T3-7 area (Fig. 1). An enzyme-linked immunosorbent assay for serum AQP4 antibodies was positive. A cerebrospinal fluid (CSF) examination showed a cell count of 5/mm³ (mononuclear cells: 100%), elevated proteins of 107.7 mg/dL and a CSF/serum quotient of albumin (Q_Alb) of 16.8, with cytology class I. Based on these findings, we diagnosed her with NMOSD and started her on methylprednisolone pulse therapy (1 g/day for 3 days) followed by plasma exchange therapy, with prednisolone 40 mg (1 mg/kg/day) and azathioprine 50 mg being given subsequently as maintenance therapy. She responded to immunotherapy with no remarkable clinical improvement, although the disease progression halted, and a follow-up evaluation on day 7 showed that the Q_Alb had improved to 3.2, which was within the normal limits, and the gadolinium-enhancing lesions had disappeared.

Figure 1.

Magnetic resonance imaging of the spinal cord. (A) T2-weighted images showed hyperintense areas at the C7-T9 level. (B) Gadolinium contrast-enhanced T1-weighted images showed enhancement at the T3-T7 level (arrowheads).

However, bloody stool and anemia were observed from day 9, and an endoscopic examination revealed transverse colon cancer. She underwent surgical resection of the tumor on day 64, with a subsequent improvement in anemia. Although her sensory impairment showed an improving trend from admission, paraplegia requiring a wheelchair persisted, and she was finally transferred to a rehabilitation hospital on day 99, by which time prednisolone had been tapered to 20 mg daily.

Surveillance of GRP78 antibodies, AQP4 and GRP78 expression in the tumor section, and a BBB permeability assay in vitro

At six months after treatment of NMOSD, we assessed the GRP78 and AQP4 antibody titers in serum samples that had been obtained at admission and at six months after treatment of NMOSD. A Western blot analysis for the detection of GRP78 autoantibodies was performed by previously described methods (6). The results showed that both GRP78 and AQP4 antibodies were positive at admission, while AQP4 antibody titers became negative, but GRP78 antibodies were still positive in samples obtained six months after treatment (Fig. 2A). Compared to control samples, the patient's serum both before and after treatment significantly increased the permeability of human brain microvascular endothelial cells (BMECs) to 10 kDa dextran (Fig. 2B). Immunostaining of tissue samples from the patient's colorectal cancer revealed a higher expression of GRP78 by tumor cells than by normal tissue (Fig. 3), but with no apparent expression of AQP4 by the tumor cells.

Figure 2.

(A) Detection of glucose-regulated protein 78 (GRP78) autoantibodies in the patient’s serum IgGs by a Western blot analysis. Human full-length GRP78 recombinant protein (1 μg, Abcam, Cambridge, UK) as the antigen was fractionated and transferred to a polyvinylidene difluoride (PVDF) membrane. IgG isolated (5 μg/mL) from this patient before (NMO-T) and after (NMO+T) treatment, NMO-IgG (positive control: PC), healthy control (HC)-IgG, and anti-GRP78 antibodies (dilution 1: 200; Abcam) were used as primary antibodies, and anti-human secondary HRP-conjugated antibody (dilution 1: 4,000) was used as the secondary antibody. Bands corresponding to the variant form of 60-kDa GRP78 (red arrowhead) were observed in IgGs from NMO-T, NMO+T and PC samples but not from HCs. Bands corresponding to 80-kDa GRP78 (black arrowhead) were observed in IgGs from the PC sample. The bands were visualized by chemiluminescence (ImmunoStar® LD; Wako, Osaka, Japan). (B) Permeability coefficient of 10-kDa dextran across human brain endothelial cell lines. Adult human brain microvascular endothelial cells were cultured in 24-well collagen-coated Transwell tissue culture inserts (0.4-mm pore size; Corning, Corning, USA). After exposure to IgG from the patient before (NMO-T) and after (NMO+T) treatment and from two healthy controls (HC1 and HC2), FITC-dextran fluorescence (10 kDa, final concentration 1 mg/mL; Sigma-Aldrich, St. Louis, USA) was added to the luminal insert, and 100 μL of medium was collected from the abluminal chamber over 40 minutes. Fluorescence signals were measured at a wavelength of 490/520 nm (absorption/emission) using a FlexStation 3 Multi-Mode microplate reader (Molecular Devices, San Jose, USA). Both NMO-T and NMO+T-IgGs significantly increased the 10-kDa dextran permeability compared to HC1 and HC2 (*p<0.05, **p<0.01).

Figure 3.

GRP78 expression in the tumor and normal tissue by immunohistochemistry. The pathology specimen from the colorectal cancer in this patient included tumor and normal tissue. Tumor (A) and normal tissue (B) from the colorectal specimen of this patient were stained with anti-GRP78 antibodies (dilution 1: 50; Abcam) as the primary antibodies and anti-rabbit secondary fluorescent antibodies (Alexa Fluor 488 goat anti-rabbit IgG; Invitrogen, Waltham, USA; dilution 1: 4,000) as the secondary antibodies. Greater cell surface GRP78 expression was observed in the tumor tissue than in the normal tissue. Hematoxylin and Eosin staining showed that the tumor was a well-differentiated tubular adenocarcinoma (C) with inflammatory cell infiltration in the surrounding normal tissue stroma (D). Scale bar, 50 μm.

Discussion

We encountered a case of paraneoplastic NMOSD with Lynch syndrome-associated colorectal cancer. In this case, the tumor cells showed a high expression of GRP78, which might have promoted the production of GRP78 antibodies as a result of an autoimmune response triggered by the tumor.

Cases of paraneoplastic NMOSD have been reported increasingly frequently in several cancers since Mueller et al. first reported a case of AQP4 antibody-positive myelitis associated with breast cancer (7). Paraneoplastic neurological syndromes are, in general, triggered by autoimmune activation against the tumor, and some case reports of paraneoplastic NMOSD have shown the expression of AQP4 by the tumor cells (8-12). Thus, it is reasonable to consider that AQP4 autoantibodies might be produced peripherally as a result of an autoimmune response to AQP4 protein on the tumor cells in some patients with cancer. Subsequent disruption of the BBB is an important step in the onset of NMOSD, since serum AQP4 antibodies cannot enter the CNS without breakdown of the BBB.

Recently, Shimizu et al. identified GRP78 autoantibodies that bind strongly to GRP78 on BBB endothelial cells, which leads to subsequent BBB disruption in NMO patients (3). GRP78 autoantibodies decrease the number of tight junction proteins and increase BBB permeability. LETM was shown to have a greater association with the presence of GRP78 antibodies than optic neuritis (6). The expression of GRP78, a member of the heat shock protein 70 family that functions as an endoplasmic reticulum chaperone (13), is up-regulated on the surface of tumor cells compared to normal cells (14,15). Importantly, GRP78 autoantibodies have been identified in patients with various tumors, such as esophageal, gastric, and colon cancers (5), and Xing et al. showed that the staining intensity of GRP78 in immunohistochemistry was significantly higher in tumor tissue than in normal tissue in cases of colorectal cancer (16). Shimizu et al. also reported that GRP78 autoantibodies were detected in patients with PCD-LEMS, which is also a paraneoplastic syndrome, and that these antibodies act on BMECs, resulting in an increase in BBB permeability (4).

Based on previous findings, we hypothesized that the high expression of AQP4 and GRP78 by tumor cells might stimulate the production of AQP4 and GRP78 antibodies, causing BBB disruption and NMOSD development. Immunostaining of tumor tissues revealed the high expression of GRP78, and both pre- and post-treatment sera were positive for GRP78 antibodies to the 60-kDa splicing variant of GRP78 that was previously identified from PCD-LEMS (4). These results suggest that autoimmune reactivity to GRP78 expressed by the tumor cells promoted the production of GRP78 autoantibodies. In addition, serum AQP4 antibodies were positive before treatment but negative at six months after immunotherapy (four months after tumor resection) in our patient. Immunosuppressive drugs and plasma exchange are known to decrease AQP4 antibody titers, although not often to the extent that they become negative (17,18). In contrast, negative changes in anti-AQP4 antibodies have been reported in several cases of paraneoplastic NMOSD in combination with tumor therapy and immunotherapy (10,19). Considering the above, although no apparent AQP4 expression in the tumor was observed in our case, the seronegative conversion of AQP4 autoantibodies after tumor resection may support the presence of a paraneoplastic mechanism for NMOSD, although immunotherapy itself may have contributed to the postsurgical AQP4 seronegativity to some extent. Interestingly, both pre- and post-treatment IgG antibodies in the present case had the ability to increase the permeability of the BBB in an in vitro assay, and GRP78 autoantibodies were still positive after immunotherapy and tumor resection. We speculate that seronegative conversion of AQP4 autoantibodies, which have a pathogenic role in NMOSD, may have contributed to the maintenance of remission despite the increased BBB permeability in this case.

In conclusion, we reported a case of paraneoplastic NMOSD with Lynch syndrome-associated colorectal cancer that was positive for both GRP78 and AQP4 antibodies. Our case highlights the notion that paraneoplastic production of GRP78 autoantibodies induces BBB disruption, consequently contributing to the development of NMOSD. GRP78 autoantibodies might similarly be an important humoral factor in the pathogenesis of certain other paraneoplastic CNS diseases as well.

Written informed consent was obtained from the patient for publishing this case study.

The authors state that they have no Conflict of Interest (COI).