Pulmonary nodule tracking in metastatic Ewing sarcoma enables personalized predictive models

Metastatic Ewing sarcoma (ES) is a rare, aggressive sarcoma subtype with poor outcomes, especially in the relapsed/refractory setting. Standard response evaluation, typically performed after two chemotherapy cycles, may be too delayed to effectively guide therapy for patients with rapidly progressing disease, limiting opportunities to switch to alternative regimens. Data-driven clinical tools are critically needed to provide earlier, patient-specific predictions of treatment efficacy. We performed a retrospective analysis of patients with metastatic ES with serial CT imaging. Tracking longitudinal measurements of individual pulmonary nodules within a modified RECIST framework, we calibrated patient-specific "digital twin” models based on ordinary differential equations and predictive simulation. Each digital twin estimates nodule-specific growth and response dynamics to various therapies, providing parameters to predict disease behavior and treatment response. We also evaluated response metrics, specifically the Disease Control Rate (DCR) compared to the Overall Response Rate (ORR), as predictors of Progression-Free Survival (PFS) and Overall Survival. Using the digital twin framework, we accurately recapitulated observed clinical trajectories, including periods of response, stability, and progression across multiple lines of therapy. Model-derived parameters, including patient-specific tumor growth rates and drug sensitivity rates, demonstrate potential as novel prognostic biomarkers for predicting time to progression and survival outcomes. Longitudinal surveillance in metastatic ES is a valuable yet underutilized clinical resource. Here, we demonstrate a digital twin model of pulmonary nodule growth as a platform for robust response classification and prognostic prediction. These patient-specific models can adaptively forecast individual treatment responses in silico, allowing prioritization of the most effective therapies. This approach offers a framework to optimize treatment sequencing, accelerate evaluation of novel agents, and ultimately maximize therapeutic opportunities for patients with this aggressive disease.