免疫检查点抑制剂（ICIs）被认为是晚期非小细胞肺癌（aNSCLC）最有前景的治疗方法。不幸的是，尚未确定统一准确的生物标志物和系统模型来预测和评估免疫相关不良事件（irAEs）。我们的目标是发现新的生物标志物，开发一种公开可用的方法，识别可能从ICIs中受益的患者。本回顾性研究招募了138名接受ICIs治疗的aNSCLC患者。进展无病生存期（PFS）和严重的irAEs是终点。收集了ICIs治疗前和1或2个周期后的人口统计学特征、严重的irAEs以及外周血炎症营养和免疫指标的数据。利用最小绝对收缩和选择算子（LASSO）与多元分析结合选择独立因素，并将其纳入诺模格构建中。通过应用曲线下面积（AUC）、校准图和决策曲线进行内部验证。本研究构建了三个具有良好预测准确性和差异度的诺模格。其中，基于组合炎症营养生物标志物的两个诺模格被构建用于预测PFS（1年-PFS和2年-PFS）和严重的irAEs，一个基于免疫指标的诺模格被构建用于预测1年-PFS。ESCLL诺模格（基于ECOG PS，preSII，changeCAR，changeLYM和postLDH）被构建用于评估PFS（1、2年-AUC = 0.893 [95％ CI 0.837-0.950]、0.828 [95％ CI 0.721-0.935]）。AdNLA诺模格（基于年龄、change-dNLR、changeLMR和postALI）被构建用于预测严重的irAEs的风险（AUC = 0.762 [95％ CI 0.670-0.854]）。NKT-B诺模格（基于change-CD3+CD56+CD16+NKT-like cells和change-B cells）被构建用于评估PFS（1年-AUC = 0.872 [95％ CI 0.764-0.965]）。虽然由于数据有限，免疫指标无法用于预测严重的irAEs，但我们是第一个发现CD3+CD56+CD16+NKT-like cells不仅与PFS相关，也与严重的irAEs相关，这在aNSCLC-ICIs研究中尚未报告。此外，我们的研究还发现较高的change-CD4+/CD8+比例与严重的irAEs显着相关。这三个新的基于非侵入性和简单的外周血数据构建的诺模格可能有助于决策。CD3+CD56+CD16+NKT-like cells被首次发现是治疗和严重的irAEs的重要生物标志物，并在区分ICIs治疗反应和严重毒性中发挥重要作用。©2023. 作者（们）.

Immune checkpoint inhibitors (ICIs) are regarded as the most promising treatment for advanced-stage non-small cell lung cancer (aNSCLC). Unfortunately, there has been no unified accuracy biomarkers and systematic model specifically identified for prognostic and severe immune-related adverse events (irAEs). Our goal was to discover new biomarkers and develop a publicly accessible method of identifying patients who may maximize benefit from ICIs.This retrospective study enrolled 138 aNSCLC patients receiving ICIs treatment. Progression-free survival (PFS) and severe irAEs were end-points. Data of demographic features, severe irAEs, and peripheral blood inflammatory-nutritional and immune indices before and after 1 or 2 cycles of ICIs were collected. Independent factors were selected by least absolute shrinkage and selection operator (LASSO) combined with multivariate analysis, and incorporated into nomogram construction. Internal validation was performed by applying area under curve (AUC), calibration plots, and decision curve.Three nomograms with great predictive accuracy and discriminatory power were constructed in this study. Among them, two nomograms based on combined inflammatory-nutritional biomarkers were constructed for PFS (1 year-PFS and 2 year-PFS) and severe irAEs respectively, and one nomogram was constructed for 1 year-PFS based on immune indices. ESCLL nomogram (based on ECOG PS, preSII, changeCAR, changeLYM and postLDH) was constructed to assess PFS (1-, 2-year-AUC = 0.893 [95% CI 0.837-0.950], 0.828 [95% CI 0.721-0.935]). AdNLA nomogram (based on age, change-dNLR, changeLMR and postALI) was constructed to predict the risk of severe irAEs (AUC = 0.762 [95% CI 0.670-0.854]). NKT-B nomogram (based on change-CD3+CD56+CD16+NKT-like cells and change-B cells) was constructed to assess PFS (1-year-AUC = 0.872 [95% CI 0.764-0.965]). Although immune indices could not be modeled for severe irAEs prediction due to limited data, we were the first to find CD3+CD56+CD16+NKT-like cells were not only correlated with PFS but also associated with severe irAEs, which have not been reported in the study of aNSCLC-ICIs. Furthermore, our study also discovered higher change-CD4+/CD8+ ratio was significantly associated with severe irAEs.These three new nomograms proceeded from non-invasive and straightforward peripheral blood data may be useful for decisions-making. CD3+CD56+CD16+NKT-like cells were first discovered to be an important biomarker for treatment and severe irAEs, and play a vital role in distinguishing the therapy response and serious toxicity of ICIs.© 2023. The Author(s).