OA02.08 Peritumoral and Intratumoral Radiomic Features Identify Aggressive Screen-Detected Early-Stage Lung Cancers

The National Lung Screening Trial (NLST) demonstrated that screening with low-dose computed tomography (LDCT) is associated with a 20% reduction in lung cancer mortality. Despite the mortality reduction benefit associated with lung cancer screening, there are many limitations of LDCT screening including overdiagnosis of slow growing, indolent cancers that may pose no threat if left untreated. Purpose: This study generated peritumoral (area surrounding the tumor) and intratumoral (within the tumor) radiomics to identify a reproducible parsimonious model that identifies a vulnerable subset of patients associated with poor survival outcomes. s: Incident lung cancer patients from the NLST were split into training (N=161) and test (N=73) cohorts and a cohort of non-screening detected adenocarcinoma lung cancers was used for further validation (N=61). Peritumoral (N=109) and intratumoral (N=155) radiomic features were extracted from LDCT images and correlated, non-stabile, non-reproducible features were removed. Overall survival (OS) and progression-free survival (PFS) were the endpoints. Following univariable analyses, the remaining significant variables were subjected to backward elimination to identify a single parsimonious model and Classification and Regression Tree (CART) was used to stratify patients into risk groups. Time-dependent Area under the Receiver Operating Characteristics (AUROC) was used to assess accuracy of the model at different time points. We identified two radiomics features (NGTDM Busyness and Statistical Root Mean Square [RMS]) which stratified patients into three risk-groups: low-risk, intermediate-risk, and high-risk. The final model was validated in the test cohort and further replicated in a cohort of non-screen detected adenocarcinoma patients. The model identified a vulnerable group early-stage patients with worse OS (HR= 9.91; 25% 2.5-year and 0% 5-year OS) versus the low-risk group (HR= 1.00; 93% 2.5-year and 78% 5-year OS). Based on the multivariable model, the OS time-dependent AUROC was 0.878 and 0.702 for 2 years in all patients and early-stage patients, respectively. Radiogenomics analysis found that statistical RMS correlated with LOC285043 and FOXF2. Though LOC285043 is an uncharacterized gene, decreased FOXF2 expression is a predictive factor for poor prognosis in early-stage NSCLC and is required for epithelial-to-mesenchymal transition. Peritumoral and intratumoral radiomics identified a subset of screen-detected lung cancers associated with very poor survival outcomes. These patients may require aggressive follow-up and/or adjuvant therapy to mitigate their poor outcomes.