MONDRIAN: Multi-omics Integrative Modelling for Stereotactic Body Radiotherapy in Early-stage Non-small Cell Lung Cancer

The overall planned duration of the study is 60 months.

The project is structured into five main tasks, as following:

• Task 1_Clinical study: Routine clinical data will be collected from eligible patients, including demographics, disease and treatment characteristics, pulmonary function, and blood count parameters.
• Task 2_Radiomics study: Quantitative imaging analysis will be performed on both diagnostic computed tomographies (CTs) and positron emission tomographies (PETs). A software compliant with the Imaging Biomarker Standardization Initiative (IBSI) will be used to extract radiomic features at baseline and at the first clinical progression (or at the last available follow-up, whichever comes first). Clinically established qualitative and semiquantitative metrics (e.g. pleural effusion in CT images and standardized uptake value in PET images) will be retrieved, as well.
• Task 3_Gene expression study: Gene expression profiling will be performed at baseline in both cohorts to investigate the presence of a prognostic pretreatment signature. The analysis will be performed on formalin-fixed paraffin-embedded (FFPE) tissue samples (namely, resected tumor in the surgical cohort and bioptic sample- whenever available- in the SBRT group). A next-generation sequencing (NGS)-based assay which targets 395 immune-related genes and named as Oncomine™ Immune Response Research Assay, will be employed.
• Task 4_Proteomics study: Proteomics will integrate the study of RNA expression performed in Task 3. As in Task 3, the analysis will be performed on FFPE samples, from both patients' cohorts.
• Task 5_Integrative model building: Integration of all previous features in a unique classification model able to determine a precise estimate of individuals with high vs low risk of disease progression and, at the same time, identify relevant predictive factors. The proposed classification model will integrate data from different sources using a multi-layer deep neural network. Federated learning techniques will also be used to reduce the bias due to the choice of a specific trial or data source.

Inclusion criteria

• Histologically proven NSCLC; all histologies will be considered eligible (e.g. squamous, adenocarcinoma, large cell carcinoma);
• American Joint Committee on Cancer Tumor Node Metastasis (AJCC TNM) classification (8th Edition) clinical stage I-II;
• Eastern Cooperative Oncology Group (ECOG) performance status 0-1;
• No absolute contraindications to either surgery (e.g. severe comorbidities) or SBRT (e.g. active connective diseases, severe pulmonary fibrosis);
• Ability and willingness to sign a written informed consent for treatment and study participation.

Exclusion criteria

• Prior diagnosis of invasive cancer (within 3 years before the diagnosis of ES-NSCLC);
• Mental diseases or psychiatric disorders preventing the acquisition of a valid informed consent.