Abstract / Summary
Abstract Background: Respiratory motion and anesthesia-associated atelectasis may contribute to CT-to-body divergence and reduce the precision of peripheral pulmonary lesion sampling. Breath-hold ventilation has been proposed to stabilize the target and biopsy tool during robotic bronchoscopy, but prospective randomized evidence is limited. We evaluated whether sustained breath-hold during tissue acquisition improves diagnostic yield or tool-lesion positioning during imaging-integrated robotic bronchoscopy. Methods: The STABLE ( S tudy of T ool-lesion A lignment and B reath-hold Ventilation in Robotic Bronchoscopy for Peripheral Lung LE sions) Trial was a prospective randomized study of patients undergoing robotic bronchoscopy for peripheral pulmonary lesions using the Noah Galaxy system with integrated augmented fluoroscopy and C-arm-based tomosynthesis. Patients were randomized to standardized sustained breath-hold during tissue acquisition or conventional ventilation during tissue acquisition; both groups underwent brief breath-holds during AF/CABT image acquisition. The primary endpoint was strict diagnostic yield according to ATS/ACCP definitions, independently adjudicated by a pulmonologist blinded to ventilation assignment. Secondary endpoints included successful lesion localization, tool-in-lesion (TIL), tool-touching-lesion (TTL), no tool in lesion (NTL), radiation exposure, and complications. Standardized mean differences (SMDs) assessed baseline balance. Results: Among 117 randomized patients, 53 underwent breath-hold and 64 conventional ventilation. Mean age was 67.5±10.0 versus 66.3±10.4 years, and mean lesion diameter was 1.7±0.7 versus 1.9±0.8 cm, respectively. Most baseline characteristics were similar; the largest imbalances involved bronchus sign (34.0% vs 51.6%; absolute SMD 0.35) and left-upper-lobe location (28.3% vs 45.3%; SMD 0.35). Overall strict adjudicated diagnostic yield was 89.7% (105/117). Diagnostic yield was 86.8% with breath-hold and 92.2% with conventional ventilation (OR 0.56, 95% CI 0.17–1.87, p-value 0.34). With greater than 1 year of ongoing follow-up, overall diagnostic accuracy was 97.4% (114/117). Nearly all procedures achieved TIL or TTL (98.1% vs 96.9%). TIL occurred in 62.3% versus 64.1% (OR 0.93, 95% CI 0.44–1.97). Localization, tool-position distribution, sampling intensity, radiation exposure, and complications showed no clear between-group differences. Two pneumothoraces requiring chest-tube placement occurred (2/117; 1.7%), both in the conventional ventilation group. Across both groups, diagnostic yield increased from 0% with NTL to 85.0% with TTL and 95.9% with TIL. Conclusions: Sustained breath-hold during tissue acquisition did not improve diagnostic yield or tool-lesion positioning during imaging-integrated robotic bronchoscopy. Diagnostic yield varied according to imaging-confirmed tool-lesion position, supporting procedural strategies that prioritize accurate real-time positioning rather than breath-hold alone. Trial registration: International Standard Randomised Controlled Trial Number (ISRCTN) Registry, ISRCTN11802169; retrospectively registered (01/09/2026).