Activity Prediction Modeling Based on a Combination of Growth Differentiation Factor 15 and Serum Biomarker Levels in Dermatomyositis and Polymyositis

Abstract

Aims

Growth differentiation factor 15 (GDF15) plays an important role in multiple inflammatory disorders. We aimed to analyze serum GDF15 levels in adult patients with idiopathic inflammatory myopathies (IIMs).

Methods

Serum GDF15 levels were measured in 179 adult patients with IIMs and 76 healthy controls (HCs). The association between GDF15 levels and disease variables was analyzed using Spearman's rank correlation. Receiver operating characteristic (ROC) curve analysis was performed to evaluate the discriminatory ability of GDF15 and the GDF15-to-lymphocyte ratio (GLR). Machine learning methods were applied to build predictive models.

Results

GDF15 levels and GLR were significantly elevated in patients with adult IIMs than in HCs. Compared with patients in remission, both GDF15 and GLR were significantly higher in myositis patients in an active phase. GDF15 levels correlated positively with myositis disease activity indices and negatively correlated with lymphocyte and platelet counts. ROC curve analysis revealed that GDF15 levels and GLR outperformed muscle enzymes and distinguished well between patients with active disease and those in remission. Furthermore, even in the normal muscle enzyme group, GDF15 levels and GLR were also well-distinguished between patients with active disease and those in remission. Using machine learning, a logistic regression model of GDF15 combined with creatine kinase and lymphocyte count was constructed and had a reliable predictive value for disease activity.

Conclusions

GDF15, particularly GLR, was significantly correlated with disease activity in adult patients with IIMs. They could serve as useful biochemical markers for evaluating disease activity, monitoring disease progression, and guiding treatment in adult patients with IIMs.