Seronegative antibody-mediated neurology after immune checkpoint inhibitors

Abstract

Checkpoint inhibitor medications have revolutionized oncology practice, but frequently induce immune-related adverse events. During autoimmune neurology practice over 20 months, we prospectively identified four patients with likely antibody-mediated neurological diseases after checkpoint inhibitors: longitudinally extensive transverse myelitis, Guillain-Barré syndrome, and myasthenia gravis. All patients shared three characteristics: symptoms commenced 4 weeks after drug administration, responses to conventional immunotherapies were excellent, and autoantibodies traditionally associated with their syndrome were absent. However, serum immunoglobulins from the myelitis and Guillain-Barré syndrome patients showed novel patterns of tissue reactivity. Vigilance is required for antibody-mediated neurology after checkpoint inhibitor administration. This phenomenon may inform the immunobiology of antibody-mediated diseases.

Figure 1

Radiological and immunological features of patient 1. (A) T2‐weighted STIR sequence with extensive medullary cord edema and swelling extending from the medulla oblongata into the low thoracic spinal cord. (B) The pons also shows T2‐weighted bilateral circumscribed areas of high signal on axial imaging. (C and D) After 4 months and corticosteroids, intravenous immunoglobulins and plasma exchange, both sagittal and axial images show almost complete resolution of swelling with no persistent gliosis or atrophy. (E) Flow cytometry plots show CD3⁺ CD4⁺ T cells gated on CD25 and CD127 to identify Tregs (CD25⁺ CD127^lo) in healthy control (HC) peripheral blood. (F) Enumeration revealed a reduction of Tregs in the blood of patient 1. (G) Tregs from patient 1 contained a subpopulation of CD25⁺⁺ CD62L⁺⁺ cells, all of which stained positively for human IgG, representing bound humanized pembrolizumab (blue contours). Red contours represent population which were not bound by human IgG. (H) This population was not bound by IgG in the healthy controls.

Figure 2

Novel autoantibody reactivities in patients with checkpoint inhibitors. (A) Aquaporin‐4 antibody positive (AQP4), healthy control (HC), and patient number 1 (Pt1) serum IgG binding to rodent sections of cerebellum (left column) and hippocampus (right column). Scale bar = 500 μm. Paraformaldehyde‐fixed brain sections were incubated for 1 hour with patient serum (1:200 dilution in PBS/0.1% Triton‐X100/5% bovine serum albumin), washed in PBS/0.1% Triton X100, and then incubated with anti‐human HRP‐conjugated secondary antibodies (1:750) for 45 min. Visualization with 3,3′‐diaminobenzidine and hydrogen peroxide. (B) Higher magnification showing HC and patient 1 (Pt1; C) IgG binding to cerebellar granule cells (G) more than molecular layer (M) or white matter (W). Scale bar = 100 μm. (D) Serum IgM (1:100 dilution for 1 h at 37°C) from the patient with Guillain–Barré syndrome bound live myelinating cocultures (from human‐induced pluripotent stem cell‐derived sensory neurons and rat primary Schwann cells). Subsequently, cultures were fixed in 1% paraformaldehyde and labeled with AlexaFluor‐488 anti‐human conjugated antibodies (green). The observed binding was to myelin blebs in particular. This was followed by permeabilization with ice‐cold methanol (30 min, on ice), and counter‐labeling anti‐neurofilament‐heavy (1:10,000, labeled blue, NF200) and anti‐myelin basic protein (1:500, labeled red, MBP) primary antibodies to visualize axonal processes and myelin internodes, respectively.