Factors influencing the diagnostic accuracy of lung cancer using endobronchial ultrasound-guided transbronchial needle aspiration

Abstract

Objectives: Endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) is a key diagnostic modality for lung cancer, yet its accuracy varies based on several factors. This study aims to identify factors influencing the diagnostic accuracy of EBUS-TBNA for lung cancer detection.

Methods: A retrospective case-control study was conducted on lung cancer patients diagnosed at Gaozhou People's Hospital from October 2021 to September 2023. Patients with lung cancer confirmed by EBUS-TBNA, bronchoscopy with direct biopsy, or surgical intervention were re-evaluated using EBUS-TBNA. Based on diagnostic accuracy, they were classified into an accurate group (n = 204) and an inaccurate group (n = 41). An external validation cohort included 58 lung cancer patients. Data collection encompassed patient demographics and EBUS-TBNA findings. Logistic regression and receiver operating characteristic (ROC) curve analyses were performed to determine factors influencing detection accuracy. A generalized linear model incorporating independent influencing factors was developed to estimate the likelihood of inaccurate EBUS-TBNA detection of lung cancer.

Results: Smoking history [dds ratio (OR), 7.948; P < 0.001] and a diagnosis of small cell lung cancer (OR, 3.996; P = 0.007) were significantly associated with an increased risk of inaccurate detection. In contrast, a lesion diameter of ≥ 3 cm (OR, 0.343; P = 0.026) and linear filamentous changes in aspirate samples (OR, 0.106; P < 0.001) were strongly correlated with accurate detection. Larger lesion size and specific sample characteristics were also significant predictors in the external validation cohort (P < 0.05). Multivariate logistic regression confirmed these factors as independent predictors of diagnostic accuracy. The predictive model demonstrated robust performance [area under the curve (AUC) = 0.882], with external validation yielding a comparable AUC of 0.877.

Conclusion: Smoking history, pathologic subtype, lesion size and aspirate sample characteristics significantly affected the diagnostic accuracy of EBUS-TBNA in lung cancer detection. These insights underscore the importance of considering these factors in clinical practice to optimize EBUS-TBNA's diagnostic performance.

Keywords: Lung cancer; diagnostic accuracy; endobronchial ultrasound-guided transbronchial needle aspiration; lesion size; predictive model; smoking history.

Figure 1

EBUS-TBNA histologic examination with H&E staining. A. Normal tissue; B. Lung cancer tissue. EBUS-TBNA: Endobronchial ultrasound-guided transbronchial needle aspiration.

Figure 2

ROC analysis of independent factors influencing EBUS-TBNA lung cancer detection accuracy. A: Smoking history; B: Small cell lung cancer; C: Diameter of lesion or lymph node ≥ 3 cm; D: Presence of linear filamentous changes in aspirate samples. E: ROC curve of the performance of linear model incorporating established using above variables. ROC: Receiver operator characteristic; EBUS-TBNA: Endobronchial ultrasound-guided transbronchial needle aspiration.

Figure 3

ROC curve of the joint predictive model in the external validation cohort.