Objective To explore the value of combining chest muscle measurements, spleen density, and immunological serum indicators for predicting the occurrence of interstitial lung disease (ILD) in Sjögren’s syndrome (SS). Methods A retrospective study was performed by SS patients admitted to the First Affiliated Hospital of Henan Medical University from January 2018 to June 2025 and 196 cases were included. Propensity score matching (PSM) was used to balance baseline characteristics, resulting in the Sjögren’s syndrome without interstitial lung disease (SS-NILD, n = 59) and Sjögren’s syndrome-associated interstitial lung disease (SS-ILD, n = 32) groups. Spearman correlation analysis was performed to assess variable relationships, and variance inflation factor (VIF) and tolerance (TOL) to quantify multicollinearity severity. Binary logistic regression was used to build models. The area under the receiver operating characteristic curve (AUC) was used to determine discriminatory ability and DeLong’s test to compare different models. Finally, model performance was assessed through calibration curves, decision curve analysis (DCA), and Bootstrap internal validation, and the contribution of imaging indicators was analyzed. Results The occurrence of SS-ILD was associated with reduced spleen density (SD), decreased total pectoral muscle area (T-PMA), decreased right pectoral muscle area (R-PMA), lower CD4⁺T lymphocyte percentage (CD4+T%), lower CD4⁺/CD8⁺ ratio, and decreased complement C4 levels. There was significant collinearity between T-PMA and R-PMA, and potential collinearity between CD4⁺T% and CD4⁺/CD8⁺ ratio. Four models were constructed. Validation via calibration and decision curves confirmed that model 1 (T-PMA, SD, CD4⁺T%, and C4) had high predictive accuracy and clinical net benefit (AUC = 0.872, sensitivity = 0.813, specificity = 0.797, 95% CI: 0.800–0.945). Bootstrap internal validation indicated high stability for model 1 and analysis of model 1’s components showed enhanced predictive performance with imaging indicators. Conclusion Model 1 (T-PMA, SD, CD4⁺T%, and C4) demonstrates potential for the early prediction of SS-ILD and carries substantial clinical translational value.<Table Float="No" ID="Taba"> <tgroup cols="2"> <colspec align="left" colname="c1" colnum="1" /> <colspec align="left" colname="c2" colnum="2" /> <tbody> <row> <entry nameend="c2" namest="c1"> Key Points • Developed a SS-ILD predictive model using pectoral muscle area, spleen density, CD4⁺T cell percentage, and complement C4. • Built via correlation heatmap, collinearity diagnosis, and propensity score matching for bias control, the model includes more indicators than prior studies and has reliable internal validation. • Identified the association of pectoralis muscle cross-sectional area with SS-ILD, and potential links of systemic immune dysregulation, muscle atrophy, splenic involvement to SS-ILD pathogenesis. • Selected indicators are easily accessible, facilitating clinical application and validation. </entry> </row> </tbody> </tgroup> </Table>

Chest muscle area, spleen density, CD4 + T%, and C4 to predict the development of interstitial lung disease in patients with Sjögren’s syndrome: a clinical prediction model

期刊论文
发表日期：

2026年1月29日

查看全文

Wenpeng Ge,
Changhua Liang,
Zhixia Wang,
Chenhuan Fan,
Zhenqiang Li,
Chaonan Chen,
Limin Jing,
Zhenzhen Liang

摘要

Objective

To explore the value of combining chest muscle measurements, spleen density, and immunological serum indicators for predicting the occurrence of interstitial lung disease (ILD) in Sjögren’s syndrome (SS).

Methods

A retrospective study was performed by SS patients admitted to the First Affiliated Hospital of Henan Medical University from January 2018 to June 2025 and 196 cases were included. Propensity score matching (PSM) was used to balance baseline characteristics, resulting in the Sjögren’s syndrome without interstitial lung disease (SS-NILD, n = 59) and Sjögren’s syndrome-associated interstitial lung disease (SS-ILD, n = 32) groups. Spearman correlation analysis was performed to assess variable relationships, and variance inflation factor (VIF) and tolerance (TOL) to quantify multicollinearity severity. Binary logistic regression was used to build models. The area under the receiver operating characteristic curve (AUC) was used to determine discriminatory ability and DeLong’s test to compare different models. Finally, model performance was assessed through calibration curves, decision curve analysis (DCA), and Bootstrap internal validation, and the contribution of imaging indicators was analyzed.

Results

The occurrence of SS-ILD was associated with reduced spleen density (SD), decreased total pectoral muscle area (T-PMA), decreased right pectoral muscle area (R-PMA), lower CD4⁺T lymphocyte percentage (CD4⁺T%), lower CD4⁺/CD8⁺ ratio, and decreased complement C4 levels. There was significant collinearity between T-PMA and R-PMA, and potential collinearity between CD4⁺T% and CD4⁺/CD8⁺ ratio. Four models were constructed. Validation via calibration and decision curves confirmed that model 1 (T-PMA, SD, CD4⁺T%, and C4) had high predictive accuracy and clinical net benefit (AUC = 0.872, sensitivity = 0.813, specificity = 0.797, 95% CI: 0.800–0.945). Bootstrap internal validation indicated high stability for model 1 and analysis of model 1’s components showed enhanced predictive performance with imaging indicators.

Conclusion

Model 1 (T-PMA, SD, CD4⁺T%, and C4) demonstrates potential for the early prediction of SS-ILD and carries substantial clinical translational value.

Key Points

• Developed a SS-ILD predictive model using pectoral muscle area, spleen density, CD4⁺T cell percentage, and complement C4.

• Built via correlation heatmap, collinearity diagnosis, and propensity score matching for bias control, the model includes more indicators than prior studies and has reliable internal validation.

• Identified the association of pectoralis muscle cross-sectional area with SS-ILD, and potential links of systemic immune dysregulation, muscle atrophy, splenic involvement to SS-ILD pathogenesis.

• Selected indicators are easily accessible, facilitating clinical application and validation.