A Novel Tool for Predicting Malignant Disease in Adult Patients with Dermatomyositis

Comprehensive Summary

This retrospective multicenter cohort study aimed to develop and validate a practical tool to identify adult patients with dermatomyositis (DM) who ar eat high risk for concomitant cancer, a major contributor to DM-related mortality. A total of 546 adults with DM or clinically amyopathic DM were included from two tertiary hospitals between 2015 and 2022. Using multivariate logistic regression and machine learning techniques, researchers identified five independent factors significantly associated with cancer: anti-transcriptional intermediary factor 1-γ (TIF1-γ) antibody positivity, interstitial lung disease (ILD), poikiloderma, DM subtype, and anemia. These variables were incorporated into the TIP-CA predictive model, a clinician-friendly scoring system. The model demonstrated strong discriminatory performance, achieving an area under the curve (AUC) of 0.809 in the derivation cohort and 0.808 in the validation cohort, indicating good predictive accuracy.

Outcomes and Implications

The TIP-CA model offers a practical, evidence-based tool for risk stratification and targeted cancer screening in patients with DM. High scores (4-5) were associated with a markedly elevated likelihood of cancer, allowing clinicians to prioritize intensive malignancy evaluation in high-risk individuals while potentially avoiding unnecessary extensive screening in low-risk patients. By incorporating routinely available clinical and serologic variables and validating performance across multidisciplinary cohorts, the model enhances generalizability and minimizes referral bias. This machine learning-informed, association-based approach supports earlier cancer detection, streamlined screening workflows, and improved survival outcomes in patients with DM, representing a meaningful step toward precision medicine in autoimmune-associated malignancy risk.