IMPORTANCEEncephalitis is a severe immune-related adverse event secondary to treatment with immune checkpoint inhibitors (ICIs). The spectrum of ICI-induced encephalitis (ICI-iE) ranges from disease that resolves fully to lethal forms. Moreover, ICIs may unmask a paraneoplastic encephalitis. To our knowledge, the factors associated with ICI-iE prognosis are unknown.OBJECTIVES To evaluate the presentation of ICI-iE and to identify features helpful in assessing outcomes.EVIDENCE REVIEW This systematic review pooled case series from the published literature (n = 77) and medical records from 1 center (n = 5) to assess the association between the form of ICI-iE presentation and its prognosis. Eligibility criteria included references identified by searches of PubMed and Web of Knowledge databases in the English literature from June 2000 (first patient dose of ipilimumab) to April 17, 2020, that examined patients with encephalitis with presumed autoimmune etiologic features induced by ICIs. Information regarding clinical, cerebrospinal fluid, and neuroimaging (magnetic resonance imaging) features, as well as treatment given, were extracted. FINDINGS A total of 82 patients (52 men [63%]; median age, 61.0 years [interquartile range, 52.5-70.0 years]) were included. Most patients presented with focal syndromes (39 [48%])or meningoencephalitis (36 [44%]). Seven patients (9%) had nonclassifiable ICI-iE. Neuronal autoantibodies were detected in 23 patients with focal syndromes and 1 patient with nonclassifiable ICI-iE. Most autoantibodies were onconeuronal (17 of 24 [71%]), targeting intracellular antigens. Patients without a focal syndrome or with a negative-antibody focal syndrome had a good prognosis (49 of 55 [89%]). Among patients with autoantibodies, those with anti-glutamic acid decarboxylase or anticell surface responded to treatment and had a favorable prognosis (100%). However, patients with other autoantibodies had poor outcomes (17 of 24 [71%]). Antineuronal autoantibodies (13 of 24 [54%] vs 5 of 41 [12%]; P < .001), focal syndrome (16 of 39 [41%] vs 4 of 43 [9%]; P = .001), and abnormal magnetic resonance imaging findings (14 of 39 [36%] vs 4 of 32 [13%]; P = .02) were associated with poor outcomes. Conversely, fever (21 of 23 [91%] vs 41 of 59 [70%]; P = .04) and more inflammatory changes in cerebrospinal fluid (30 of 31 [97%] vs 21 of 33 [64%]; P = .001) were associated with a better prognosis.CONCLUSIONS AND RELEVANCE Immune checkpoint inhibitors may induce mainly 2 different encephalitic syndromes: a focal limbic or extralimbic encephalitis and a meningoencephalitis. Immune checkpoint inhibitor-induced encephalitis is associated with an overall favorable outcome, with a low rate of fatal events. An undetected preexisting paraneoplastic encephalitic syndrome may be triggered by ICIs, and this type of syndrome has the worst outcome among all the different types of ICI-induced encephalitis syndromes. Clinical presentation and systematic measurement of autoantibodies will be a helpful guide for the therapeutic st...