Prognostic factors and a nomogram for survival in anti-MDA5 antibody-positive dermatomyositis-associated interstitial lung disease
摘要
To characterize the clinical features and identify determinants of mortality in patients with dermatomyositis (DM)-associated interstitial lung disease (ILD) who are positive for anti-melanoma differentiation-associated gene 5 (MDA5) antibodies, and to develop a prognostic prediction model.
MethodsPatients with DM-ILD were retrospectively analyzed. Mortality and baseline features were compared between antibody-positive and antibody-negative groups. In the antibody-positive subgroup, Kaplan-Meier survival curves were generated, and clinical characteristics were compared between survivors and non-survivors. Prognostic factors were identified by elastic-net Cox regression, and a nomogram was constructed. Model performance was evaluated using C-index, calibration, and decision curve analysis (DCA).
ResultsAmong 147 patients with DM-ILD, the 6-month mortality was significantly higher in the anti-MDA5-positive group compared with the antibody-negative group (35.2% vs. 3.9%). In the antibody-positive subgroup, the mean follow-up time was 21.60 ± 14.93 days for non-survivors and 160.26 ± 52.91 days for survivors (P < 0.001). Kaplan-Meier analysis showed no significant difference in survival when stratified by year of diagnosis before and after 2020 (Log rank P = 0.298; HR, 0.653; 95% CI, 0.288–1.478). Elastic-net Cox regression identified rapidly progressive ILD (RP-ILD), serum albumin (ALB), C-reactive protein (CRP), ferritin, and neutrophil-to-lymphocyte ratio (NLR) as independent predictors of mortality in anti-MDA positive patients. A nomogram incorporating these variables was developed, and the final model demonstrated good discrimination (optimism-corrected concordance index 0.902) and calibration.
ConclusionAnti-MDA5 antibody positivity was strongly associated with higher short-term mortality in DM-ILD. The proposed nomogram, integrating RP-ILD, ALB, CRP, ferritin, and NLR, showed robust predictive accuracy and may aid individualized risk stratification. External validation is warranted.