Abstract
Autoimmune disease associated autoantibodies have been implicated in both immune-related adverse events (irAEs) and chemoimmunotherapy responses; however, current biomarkers lack sufficient predictive power, especially for irAEs severity. Here, we developed an autoimmune disease (AID) autoantigen microarray (AID microarray) capable of detecting 125 autoantibodies associated with over 30 autoimmune diseases. The AID microarray demonstrated excellent reproducibility (intra-batch correlation: 0.99; inter-batch correlation: 0.97) and strong concordance with clinical chemiluminescence immunoassays (R(2) = 0.86). We analyzed baseline serum samples from 83 lung cancer patients who experienced varying severity of irAEs following immune checkpoint inhibitors (ICIs) therapy. Nine autoantibodies were identified as being positively correlated with irAEs severity (samr-nonparametric, p < 0.05). A predictive model incorporating these nine autoantibodies (9-panel) effectively distinguished patients at risk of irAEs (G0 vs. G1&G2&G3: AUC = 0.854) and severe irAEs (G0 vs. G3: AUC = 0.934). Additionally, an eight-autoantibody panel (8-panel) demonstrated robust performance in predicting immunotherapy efficacy, achieving an AUC of 0.855 in the training cohort and 0.746 in the validation cohort. Multivariate Cox regression analysis identified anti-NAP1L4 IgG and anti-Ku IgG as independent prognostic risk factors (hazard ratio [HR] > 1, p < 0.05), whereas anti-GLRA2 IgA and anti-KRT20 IgA exhibited protective effects (HR < 1, p < 0.05). These findings support the use of autoantibody profiling as a predictive tool for both treatment response and irAEs in NSCLC patients receiving ICIs. The AID microarray offers a high-throughput platform for identifying autoantibody biomarkers that may guide immunotherapy in cancer patients.