Neurological autoimmunity in melanoma patients: a comparison between those exposed and non-exposed to immune checkpoint inhibitors.

BACKGROUND: The clinical spectrum of melanoma-associated neurological autoimmunity, whether melanoma-associated paraneoplastic neurological syndromes (PNS) or induced by immune checkpoint inhibitors (ICI), is not well characterized. We aim to describe the clinical spectrum of melanoma-associated neurological autoimmunity.

METHODS: A systematic review of the literature combined with patients from French databases of paraneoplastic neurological syndromes was conducted. All melanoma patients with a possible immune-mediated neurologic syndrome were included and classified according to whether they had previously been exposed to ICI (ICI-neurotoxicity) or not (ICI-naïve) at first neurological symptoms.

RESULTS: Seventy ICI-naïve (literature: n = 61) and 241 ICI-neurotoxicity patients (literature: n = 180) were identified. Neuromuscular manifestations predominated in both groups, but peripheral neuropathies were more frequent in ICI-neurotoxicity patients (39.4% vs 21.4%, p = 0.005) whereas myositis was more frequent in ICI-naïve patients (42.9% vs 18.7%, p < 0.001). ICI-naïve patients had also more frequent central nervous system (CNS) involvement (35.7% vs 23.7%, p = 0.045), classical paraneoplastic syndrome (25.7% vs 5.8%, p < 0.001), and more frequently positive for anti-neuron antibodies (24/32, 75.0% vs 38/90, 42.2%, p = 0.001). Although more ICI-neurotoxicity patients died during the acute phase (22/202, 10.9% vs 1/51, 2.0%, p = 0.047), mostly myositis patients (14/22, 63.6%), mortality during follow-up was higher in ICI-naïve patients (58.5% vs 29.8%, p < 0.001). There was no significant difference in the frequency of life independence (mRS ≤ 2) in the surviving patients in both groups (95.5% vs 91.0%, p = 0.437).

CONCLUSIONS: Melanoma-associated PNS appear remarkably rare. The clinical similarities observed in neurological autoimmunity between ICI-treated and ICI-naïve patients, characterized predominantly by demyelinating polyradiculoneuropathy and myositis, suggest a potential prior immunization against melanoma antigens contributing to ICI-related neurotoxicity.