Abstract:
Purpose: To determine whether CT predictors of visceral pleural invasion (VPI) differ by tumor–pleural relationship in part-solid lung adenocarcinoma and to develop interpretable models for pleural-attached and pleural-tag nodules.
Methods: This retrospective study included 785 patients with surgically resected clinical stage IA part-solid lung adenocarcinoma (208 VPI-positive; 577 VPI-negative). Nodules were classified as pleural-attached (n=417) or pleural-tag (n=368). Common CT predictors were selected by LASSO for an overall clinical–CT model. Relationship-specific features were then incorporated into separate attached and tag models. Model performance and associations with adverse pathologic features and recurrence-free survival (RFS) were assessed.
Results: Maximum solid component diameter, tumor disappearance ratio, and pleural attachment were common predictors; the overall model achieved AUCs of 0.77 and 0.80 in training and testing cohorts, respectively. In attached nodules, pleural thickening, pleural contact depth (PCD), solid component–pleural abutment, and concomitant pleural tags independently predicted VPI (AUCs, 0.83 and 0.82). In tag nodules, pleural traction depth ratio (PTDR), triangular soft-tissue tags, and pleural traction alignment were independent predictors (AUCs, 0.86 and 0.85). Model-defined high-risk groups had more adverse pathologic features. The attached-specific score significantly stratified RFS, whereas the tag-specific score did not. Training-derived cutoffs of PCD >9.5 mm and PTDR >0.25 retained discriminatory value in the testing cohort (AUCs, 0.80 and 0.74, respectively).
Conclusion: VPI-associated CT features differ between pleural-attached and pleural-tag part-solid lung adenocarcinomas. Relationship-specific assessment of the solid component–pleural interface and pleural traction features may refine preoperative VPI risk stratification and provide complementary prognostic information, particularly for pleural-attached nodules.
Keywords: visceral pleural invasion; lung adenocarcinoma; part-solid nodule; computed tomography; pleural attachment; pleural tag; risk stratification
