MIRACLE

Early-stage non small cell lung cancer (ES-NSCLC) represents 20-30% of all NSCLC and is characterized by a high survival rate after surgery. However, there is variability in clinical outcomes among patients sharing the same disease stage, suggesting that other factors could determine the risk of relapse. Accurate and validated tools to stratify patients according to their risk of relapse are still lacking. Hypothesis: We hypothesize that multiple factors could influence the prognosis of resected ES-NSCLC patients. In particular, tumor tissue and microenvironment (TME) characteristics, liquid biopsy, radiomics features and clinical-pathological factors could all be involved. Aims: Primary: Development of a machine learning (ML) algorithm acting as a clinical decision support tool for disease free survival (DFS) prediction and patient stratification based on joint analysis of biological, clinical and radiologic features on a training cohort of resected ESNSCLC. Secondary: Validation of the developed algorithm on an independent cohort. Methods: A previously prospectively collected cohort of 220 ES-NSCLC patients will be considered as a training set. Tumor tissue and TME characteristics will be analysed using DNA and RNA sequencing approaches; liquid biopsy will be used to assess free circulating DNA and extracellular vesicles; radiomics parameters will be retrieved from computed tomography images. All these features, together with clinico-pathological factors, will be integrated in a model that will enable personalized patient treatment. The developed algorithm will be validated in a prospective cohort enrolled during MIRACLE. Expected results and potential impact: We expect to develop and validate a practical solution for an algorithm for DFS prediction to identify resected ES-NSCLC patients with different risk of relapse. This algorithm could be useful to improve patient management and establish more efficient and ethical therapeutic strategies.