Diagnostic accuracy of imaging-guided biopsy of peripheral pulmonary lesions: a systematic review

The histologic definition of peripheral pulmonary lesion (PPL) is critical for a correct diagnosis and appropriate therapy. Non-invasive techniques for PPL biopsy are imaging-guided, using endobronchial ultrasound (EBUS), computed tomography (CT), and electromagnetic navigation bronchoscopy (ENB). To assess the diagnostic accuracy of PPL biopsy and provide a framework for reporting data for accuracy studies of PPL biopsy. A systematic review was conducted on PubMed, Scopus, and Web of Science to identify all the articles assessing the accuracy of EBUS, CT, and ENB between January 2000 and June 2023 basing search queries on keywords emerging from PICO question. Only studies investigating biopsy of PPL and reporting accuracy or necessary data to calculate it independently were included. Risk of bias was based on QUADAS-2 tool. In total, 81 studies were included. Median accuracy was 0.78 (range=0.51-0.94) in the EBUS group, 0.91 (range=0.73-0.97) in the CT group, 0.72 (range=0.59-0.97) in the ENB group, and 0.77 (range=0.61-0.92) in the combined group. Sensitivity and NPV ranges were 0.35-0.94 and 0.26-0.88 in the EBUS group, 0.71-0.97 and 0.46-1.00 in the CT group, 0.55-0.96 and 0.32-0.90 in the ENB group, and 0.70-0.90 and 0.28-0.79 in the combined group. Specificity and PPV were 1.00 in almost all studies. Overall complication rate was 3%, 30%, 8%, and 5% in the EBUS, CT, ENB, and combined groups. CT-guided biopsy was the most accurate technique, although with the highest complication rate. When calculating accuracy, indeterminate results must be considered false negatives according to the "intention-to-diagnose" principle.