Exploring Immunological Determinants and Constructing a Predictive Model for for Diagnosis of Gout: Insights Into Immune Biomarkers

Huanhuan Zeng,* Liqiang Zheng,* Dan Wu, Xiang Yu, Guangjiang Zhang, Jinwan Du Department of Rheumatology and Immunology, People’s Hospital of Chongqing Liangjiang New Area, Chongqing, 401121, People’s Republic of China*These authors contributed equally to this workCorrespondence: Jinwan Du, Department of Rheumatology and Immunology, People’s Hospital of Chongqing Liangjiang New Area, No. 2, Jinkai Avenue, Yubei District, Chongqing, 401121, People’s Republic of China, Email [email protected]: Immune-inflammatory mechanisms play a pivotal role in gout pathogenesis, yet the specific immunological signatures and their predictive utility remain underexplored.Methods: This cross-sectional study enrolled 130 participants (65 gout patients, 65 controls) to compare clinical characteristics, inflammatory markers, and immune profiles. Key variables were selected via least absolute shrinkage and selection operator (LASSO) regression, and an extreme gradient boosting (XGBoost) prediction model was constructed. Model interpretability and robustness were assessed using SHapley Additive exPlanations (SHAP), principal component analysis (PCA), and external validation.Results: Baseline characteristics showed no significant intergroup differences, ensuring cohort comparability. Gout patients exhibited elevated levels of inflammatory mediators, including C-reactive protein (CRP), high-sensitivity CRP (hs-CRP), interleukin-1β (IL-1β), interleukin-6 (IL-6), and NLR family pyrin domain-containing 3 (NLRP3), alongside immune dysregulation marked by increased CD4+/CD8+ and Th17/regulatory T-cell (Treg) ratios and CD14+/CD16+ monocyte expansion, indicating systemic inflammatory activation and immune imbalance. Cluster analysis identified two immunological subtypes. The XGBoost model, incorporating seven LASSO-selected biomarkers, achieved perfect discrimination in internal validation (AUC = 1.0, accuracy = 100%) and high performance in an external cohort (AUC = 0.977, accuracy = 93.75%). PCA and random forest analyses confirmed hs-CRP and IL-1β as core predictors.Conclusion: Gout is characterized by distinct immune-inflammatory signatures. The machine learning model leveraging immunological biomarkers demonstrates exceptional classification accuracy and generalizability, offering potential for early screening and immunological subtyping in clinical practice, may support earlier diagnosis in patients with atypical or silent gout manifestations.Keywords: gout, immune biomarkers, inflammation, LASSO regression, XGBoost model, immunological subtyping, predictive model, principal component analysis, random forest