Role of autoimmune reactivity in neurotoxicities (N-Tox) in melanoma patients treated with immune-checkpoint inhibitors (ICI).

2517 Background: N-Tox is a grossly understudied immune-related adverse event (irAE), despite its association with mortality (e.g. encephalitis) and morbidities (e.g. peripheral neuropathy). We reported that pre-treatment sera autoantibodies (auto-Abs) are implicated in the pathogenesis of irAEs (Johannet et al. CCR 2022). We here examined the rate and patterns of N-Tox in melanoma patients who received ICI in the adjuvant setting and whether baseline specific serum auto-Abs are associated with N-Tox. Methods: We examined clinicopathological features and baseline auto-Abs of 965melanoma patients (551 male and 414 female) enrolled in two phase III clinical trials: Checkmate 238 and Checkmate 915 (797 resected stage III, 166 resected stage IV, and 2 unknown). Patients received ipilimumab (n = 423), nivolumab (n = 347), or both (n = 195). We compared pre-treatment serum auto-Ab profiles using the HuProt Human Proteome Microarray v4.0 (CDI Laboratories, Mayaguez, PR) that has 21,000+ individually purified full-length human proteins and protein isoforms in duplicate, in patients who developed at least a single incidence of N-Tox grade ≥2 to those who developed only other types of irAEs grade ≥2. We used a threshold of logFC > 0.3 and false discovery rate (FDR) adjusted P value < 0.05 to determine differentially expressed auto-Abs in patients with N-Tox. Results: 329/965 (34%) patients developed N-Tox. There were 426 total incidences of N-tox (grade 1 n = 258, grade 2 n = 132, grade 3 n = 34, grade 4 n = 2). 97/329 patients developed more than one grade of N-Tox. Patients who received ipilimumab were more likely to experience any grade N-Tox (P = 0.002) in a multivariate model. Any grade N-tox was also associated with a lower recurrence rate (P = 0.004). Gender and melanoma stage were not associated with N-Tox (P > 0.05). A signature of 160 auto-Abs, including those targeting mitochondrial proteins (ATP5PO, COX6C, NDUFA3, NDUFB6), calcineurin (PPP3CC, PPP3R1), cellular architecture (RAC1), and inflammation/apoptosis (TRAF2) were significantly overexpressed in the N-Tox grade ≥2 cohort (n = 143) compared to non-N-Tox grade ≥2 (n = 569). Pathway analysis revealed these auto-Abs were enriched in several pathways involved in neuroinflammation and neurodegeneration, including TNF-a signaling, B cell receptor signaling, interleukin-2 production, natural killer cell mediated cytotoxicity, and cellular senescence. Conclusions: Our data demonstrate that the incidence of N-Tox is higher than previously reported, possibly due to stringent assessment and follow up in clinical trial settings. The multiplicity of pathways involved, some of them directly involved in neurodegeneration and neuroinflammation, suggests a complex N-tox pathogenesis that requires further clinical and pre-clinical investigations.