Central Tumor Location in Resected Lung Adenocarcinoma: Association with Recurrence and Survival Outcomes.

Prognostic implications of lung cancer location remain uncertain, in part relating to heterogeneous definitions for central tumor location. To assess associations of central tumor location, determined using an automated lung segmentation algorithm, with recurrence and survival outcomes in patients with resected lung adenocarcinoma. This retrospective study included patients who underwent lobectomy or pneumonectomy for invasive lung adenocarcinoma between July 2010 and December 2019. An automated algorithm was developed to generate segmentation masks dividing the lungs on CT into three concentric regions (https://github.com/provbs/L3P_rb). Radiologists reviewed preoperative chest CT examinations using these masks to classify each tumor as located in the lung's inner, middle, or outer third based on each of two reference points (tumor center, tumor medial margin). Each tumor was then classified as central versus peripheral in location according to four definitions: center within inner one-third (definition 1); center within inner two-thirds (definition 2); medial margin within inner one-third (definition 3); medial margin within inner two-thirds (definition 4). Associations of central location with locoregional and distant recurrences, recurrence-free survival (RFS), and overall survival (OS) were evaluated, adjusting for covariates influencing lung cancer survival including pathologic nodal category. Among 1796 patients (mean age, 62.7 years; 849 male, 947 female), 147 (8.2%), 916 (51.0%), 521 (29.0%), and 1427 (79.4%) were classified as having central tumors by definitions 1, 2, 3, and 4, respectively. Central location showed independent associations with increased risk of locoregional recurrence for definition 1 (subdistribution hazard ratio [sHR]=1.75; p=.004) and definition 3 (sHR=1.44; p=.01); no definition was independently associated with distant recurrence. Central location showed independent associations with worse RFS for definition 1 (HR=1.52; p<.001) and definition 3 (HR=1.28; p=.003) and with worse OS for definition 1 (HR=1.45; p=.02). All of these associations were also observed in a subset of 1172 patients with pathologic stage I disease. In patients with resected lung cancer, central location, defined as location of the tumor center within the lung's inner one-third, was independently associated with increased risk of locoregional recurrence and poorer RFS and OS. Patients with centrally located lung cancers may warrant closer postoperative surveillance.