RESUMEN
BACKGROUND: Neurological immune-related adverse events associated with immune checkpoint inhibitors can have several clinical manifestations, but the syndromes and prognostic factors are still not well known. We aimed to characterise and group the clinical features, with a special focus in patients presenting with encephalopathy, and to identify predictors of response to therapy and survival. METHODS: This retrospective observational study included patients with neurological immune-related adverse events from 20 hospitals in Spain whose clinical information, serum samples, and CSF samples were studied at Hospital Clinic de Barcelona, Barcelona, Spain. Patients with pre-existing paraneoplastic syndromes or evidence of alternative causes for their neurological symptoms were excluded. We reviewed the clinical information, classified their clinical features, and determined the presence of neural antibodies. Neurological status was assessed by the treating physician one month after adverse event onset (as improvement vs no improvement) and at the last evaluation (complete recovery or modified Rankin Scale score decrease of at least 2 points, indicating good outcome, vs all other modified Rankin Scale scores, indicating poor outcome); if the participant had died, the date and cause of death were recorded. We used Fisher's exact tests and Mann-Whitney U tests to analyse clinical features, and multivariable logistic regression to analyse prognostic factors. FINDINGS: From Jan 1, 2018, until Feb 1, 2023, 83 patients with suspected neurological immune-related adverse events after use of immune checkpoint inhibitors were identified, of whom 64 patients were included. These patients had a median age of 67 years (IQR 59-74); 42 (66%) were male and 22 (34%) were female. The predominant tumours were lung cancer (30 [47%] patients), melanoma (13 [21%] patients), and renal cell carcinoma (seven [11%] patients). Neural antibodies were detected in 14 (22%) patients; 52 (81%) patients had CNS involvement and 12 (19%) had peripheral nervous system involvement. Encephalopathy occurred in 45 (70%) patients, 12 (27%) of whom had antibodies or well defined syndromes consistent with definite paraneoplastic or autoimmune encephalitis, 24 (53%) of whom had encephalitis without antibodies or clinical features characteristic of a defined syndrome, and nine (20%) of whom had encephalopathy without antibodies or inflammatory changes in CSF or brain MRI. Nine (14%) of 64 patients had combined myasthenia and myositis, five of them with myocarditis. Even though 58 (91%) of 64 patients received steroids and 31 (48%) of 64 received additional therapies, 18 (28%) did not improve during the first month after adverse event onset, and 11 of these 18 people died. At the last follow-up for the 53 remaining patients (median 6 months, IQR 3-13), 20 (38%) had a poor outcome (16 deaths, one related to a neurological immune-related adverse event). Mortality risk was increased in patients with lung cancer (vs those with other cancers: HR 2·5, 95% CI 1·1-6·0) and in patients with encephalopathy without evidence of CNS inflammation or combined myocarditis, myasthenia, and myositis (vs those with the remaining syndromes: HR 5·0, 1·4-17·8 and HR 6·6, 1·4-31·0, respectively). INTERPRETATION: Most neurological immune-related adverse events involved the CNS and were antibody negative. The presence of myocarditis, myasthenia, and myositis, of encephalopathy without inflammatory changes, or of lung cancer were independent predictors of death. Most deaths occurred during the first month of symptom onset. If our findings are replicated in additional cohorts, they could confirm that these patients need early and intensive treatment. FUNDING: The Instituto de Salud Carlos III and the European Union.