Predicting Pneumothorax and Hemorrhage After CT-Guided Lung Biopsy: Role of Lesion Size, Depth and Their Interaction

Background/Objectives: CT-guided transthoracic lung biopsy is essential for diagnosing pulmonary lesions but poses risks, chiefly pulmonary hemorrhage and pneumothorax. Both complications can prolong hospitalization or require chest drainage. Emerging evidence suggests that hemorrhage along the biopsy tract may influence pneumothorax risk. This study aimed to evaluate lesion characteristics, tract hemorrhage, and their interaction as predictors of complications, as well as to explore predictive modeling approaches. Methods: In this retrospective single-center study, 118 patients (median age 69 years; 55.9% male) underwent CT-TTLB between January 2020 and April 2025. Multivariable logistic regression with Firth correction was used to identify predictors of hemorrhage and pneumothorax, including a lesion depth-hemorrhage interaction. Model discrimination was assessed via bootstrap-corrected ROC analysis. CART analysis defined lesion size thresholds associated with hemorrhage risk. Random Forest and XGBoost models were applied for exploratory comparison. Results: Hemorrhage occurred in 29.7% and pneumothorax in 22.0% of cases, with overlap in seven. Larger lesions were less likely to bleed (OR 0.96 per mm, p = 0.002), while deeper lesions increased pneumothorax risk (OR 1.06 per mm, p = 0.007). The depth-hemorrhage interaction suggested a possible mitigating effect of bleeding in deeper lesions; however, this was an exploratory observation and did not reach conventional statistical significance (p = 0.078). Conclusions: Limited tract hemorrhage may partially reduce pneumothorax risk in deeper lesions, but this finding is exploratory and requires prospective validation. Lesion size remains the primary determinant of bleeding, and CART-derived thresholds may inform individualized procedural planning and risk stratification.