The Diagnostic Accuracy and Sensitivity for Malignancy of Radial-Endobronchial Ultrasound and Electromagnetic Navigation Bronchoscopy for Sampling of Peripheral Pulmonary Lesions Systematic Review and Meta-analysis

Background: Lung cancer screening with computed tomography chest is identifying peripheral pulmonary lesions (PPLs) suspicious for early-stage lung cancer at increasing rates. Radial-endobronchial ultrasound (R-EBUS) and electromagnetic navigation bronchoscopy (ENB) are 2 methods to sample PPLs to diagnose and treat early lung cancer. ENB has a higher operating financial cost, however, the rationale for its use is possible higher diagnostic accuracy versus R-EBUS. Objective: The objective of this study was to determine the comparative diagnostic accuracy, sensitivity, and negative predictive value for R-EBUS and ENB in sampling PPLs. Methods: A systematic review and meta-analysis were conducted. The Ovid Medline database was queried for original research reporting a diagnostic yield of R-EBUS or ENB for PPLs identified on computed tomography chest suspicious for malignancy. The I-2 statistic assessed study heterogeneity. Random effects models produced pooled estimates of diagnostic accuracy and sensitivity for malignancy. Reasons for heterogeneity were explored with meta-regression. Publication bias and small study effects were assessed. Results: A total of 41 studies involved 2988 lung nodules (R-EBUS 2102, ENB 886) in 3204 patients (R-EBUS 2097, ENB 1107). Overall sensitivity to detect cancer was 70.7% [95% confidence interval (CI): 67.2-74.0]; R-EBUS 70.5% (95% CI: 66.1-74.8), ENB 70.7% (95% CI: 64.7-76.8). Pooled overall diagnostic accuracy was 74.2% (95% CI: 71.0-77.3); R-EBUS 72.4% (95% CI: 68.7-76.1), ENB 76.4% (95% CI: 70.8-82.0). The localization modalities had comparative safety profiles of Conclusion: Both technologies have a high proportion of successful PPL localization with similar sensitivity for malignancy and accuracy. As such, both reasonable options for health care authorities to employ diagnostic algorithms.