Integration of clinical variables and CT radiomics features in a nomogram for predicting interstitial lung disease in Sjögren’s syndrome

Background: Sjögren’s syndrome (SS) is a systemic disease characterized by a chronic autoimmune response, with complex and diverse clinical manifestations due to its multi-organ involvement, often complicated by interstitial lung disease (ILD), which severely impairs patients’ quality of life. Early identification of ILD in SS patients is of significant clinical importance, but challenges remain. This study aims to construct and validate a predictive model integrating clinical and computed tomography (CT) imaging features for individualized risk assessment of ILD occurrence in SS patients.

Methods: This retrospective study consecutively enrolled 57 patients diagnosed with SS who underwent thin-section chest CT, divided into Sjögren’s syndrome with interstitial lung disease (SS-ILD) and Sjögren’s syndrome without interstitial lung disease (SS-NILD) groups based on the presence of ILD. Demographic and serological data (including anti-Ro52 status) were collected, and a total of 107 imaging features were extracted from the lung parenchyma region of interest (ROI). Univariate comparisons were performed to screen candidate features, and least absolute shrinkage and selection operator (LASSO) regression was used for dimensionality reduction and selection of imaging features. Three progressive logistic regression models were constructed based on the selected imaging features and clinical variables: Model 1 (age only), Model 2 (age + anti-Ro52), Model 3 [age + anti-Ro52 + imaging feature mean absolute deviation (MAD)]. The discriminative ability of the models was assessed using the area under the receiver operating characteristic (ROC) curve (AUC), and clinical net benefit was evaluated using decision curve analysis (DCA), with the optimal model visualized as a nomogram.

Results: A total of 57 patients were included in this study, with clinical variable differences between groups including age (P=0.04) and anti-Ro52 status (P=0.01). LASSO regression identified MAD as a key imaging feature. The AUCs for Models 1, 2, and 3 in the training set were 0.660, 0.703, and 0.779, respectively. DCA showed that Model 3 had greater clinical net benefit than simpler models across a wide range of threshold probabilities. A nomogram was constructed based on Model 3 for individualized risk estimation.

Conclusions: The model integrating age, anti-Ro52 status, and a single CT imaging feature (MAD) outperformed models based solely on clinical variables in distinguishing whether SS patients have ILD. This nomogram aids in early identification of high-risk patients and can be used for risk stratification. However, further validation of the model’s generalizability is needed in prospective studies and larger external cohorts.