Analysis of Factors Associated With Interstitial Lung Disease in Treatment‐Naïve Rheumatoid Arthritis Patients and Development of a Classification Nomogram
Li Lu, Yongkang Chen, Ting Zhang, Lina Cao, Hua Ma, Yuanyuan Song, Dongmei ZhouABSTRACT
Background
Interstitial lung disease (ILD) is a serious extra‐articular manifestation of rheumatoid arthritis (RA) and contributes substantially to morbidity and mortality. Early identification of RA‐associated ILD (RA‐ILD) remains challenging, and reliable biomarkers for risk stratification are lacking.
Objective
This study aimed to identify cytokine‐related factors associated with RA‐ILD and to develop an exploratory classification model for assessing the presence of ILD in patients with newly diagnosed, treatment‐naïve RA.
Methods
A total of 66 newly diagnosed, treatment‐naïve RA patients were enrolled, including individuals with and without ILD confirmed by high‐resolution computed tomography (HRCT). All patients were disease‐modifying antirheumatic drug (DMARD)‐naïve and had not received glucocorticoids or biologics prior to enrollment. Plasma IL‐22 concentration was measured by ELISA; other cytokines (IL‐2, IL‐4, IL‐6, IL‐10, IL‐12p70, IL‐17A, IFN‐γ, and TNF‐α) were quantified using a multiplex bead‐based immunoassay. Clinical characteristics, laboratory parameters, and radiological findings were collected. Feature selection was performed using least absolute shrinkage and selection operator (LASSO) regression and Boruta algorithms, followed by multivariable logistic regression to identify independent factors associated with RA‐ILD. A nomogram was constructed based on selected predictors, and its performance was evaluated using receiver operating characteristic analysis and cross‐validation.
Results
Patients with RA‐ILD exhibited distinct cytokine profiles compared with those without ILD. Multivariable analysis identified age, smoking, lower mean corpuscular hemoglobin concentration (MCHC), higher plasma TNF‐α levels and lower plasma IL‐22 levels as independent factors associated with RA‐ILD. The nomogram exhibited good discriminative performance, with a fivefold cross‐validated AUC of 0.851 (95% CI: 0.755–0.947); the apparent AUC in the full sample was 0.87. Internal validation indicated acceptable model stability.
Conclusion
Age, smoking, MCHC, higher plasma TNF‐α levels and lower plasma IL‐22 levels were associated with the presence of ILD in newly diagnosed, treatment‐naïve RA patients. The proposed cytokine‐based nomogram may serve as a hypothesis‐generating tool for early identification of RA‐ILD. Further validation in larger, independent cohorts is warranted before clinical application.