The PRECIsE study: a prospective, multicenter study of shape-sensing robotic-assisted bronchoscopy with 2 years of follow-up.

🎯 Background & Rationale

Guidelines recommend nonsurgical biopsy for indeterminate pulmonary lesions >8 mm. While shape-sensing robotic-assisted bronchoscopy (ssRAB) has grown rapidly for sampling small peripheral nodules, existing literature has been largely limited to single-center studies or trials with short follow-up periods. The PRECIsE study addresses this gap by providing a multicenter, prospective evaluation of the first-generation ssRAB system across 2 years of follow-up, specifically assessing performance among nascent robotic users without the aid of cone-beam computed tomography (CBCT).

👥 Study Design & Population

• Study Type: Prospective, multicenter, observational cohort study.
• Population: 305 procedures performed across 6 centers (following an initial lead-in phase of 60 cases). Enrolled patients presented with pulmonary nodules measuring 10 to 30 mm located in or beyond the subsegmental airways.
• Key Baseline Metrics: Median nodule size was 17.0 mm (IQR, 14.0–23.0), median distance to the pleura was 11.0 mm (IQR, 1.0–23.0), and a CT bronchus sign was present in 37.4% of cases. Prevalence of malignancy in the study cohort was 77.1%.

📈 Methodology & Rigor

• Procedure & Follow-up: Patients underwent ssRAB under general anesthesia using the Ion Endoluminal System. The protocol mandated 2 years of follow-up to establish a reference standard for malignancy. CBCT guidance was prohibited during study biopsies, though 2D fluoroscopy and radial EBUS (rEBUS) were utilized.
• Statistical Rigor: Multilevel logistic regression models were constructed with center (and physician nested within center) modeled as random effects to control for confounding variables affecting diagnostic yield and sensitivity for malignancy. Akaike information criterion (AIC)–based stepwise regression was used for covariate selection.

🔬 Key Findings [or Planned Endpoints]

• Primary Outcomes (Sensitivity for Malignancy): Sensitivity through 1 year was 83.4% (95% CI, 77.9%–88.0%), which decreased slightly to 81.3% (95% CI, 75.7%–86.1%) at the 2-year mark due to additional cancers detected in the second year.
• Diagnostic Yield: Yield was 74.1% (95% CI, 68.8%–78.9%) according to strict ATS/ACCP criteria and 77.5% (95% CI, 72.4%–81.8%) using intermediate study criteria.
• Multilevel Predictors: Poorer sensitivity and yield were significantly associated with female sex, smaller nodule size, lower lobe location, semi-solid density, and higher body mass index (BMI). Increasing operator case numbers improved performance. Center-level effects did not reach statistical significance.
• Safety Outcomes: The pneumothorax rate was 4.6%, with only 1.6% requiring an intervention. Airway bleeding occurred in 1.0% of cases (two Nashville grade 2 events and one Nashville grade 3 event managed with a Fogarty balloon without selective intubation).

⚖️ Critical Appraisal

• Strengths: Prospective multicenter design, robust 2-year longitudinal follow-up, inclusion of proceduralists new to robotic bronchoscopy (enhancing external validity), and sophisticated multilevel modeling accounting for center-level variance.
• Limitations: Conducted in high-volume academic and community centers by bronchoscopists already experienced in navigational bronchoscopy, which may limit generalizability to entirely novice operators. Prolonged enrollment driven by the COVID-19 pandemic and a small loss-to-follow-up rate (2%) required sensitivity analyses, though worst-case scenarios confirmed stable performance bounds (~79.3%). Additionally, the restriction against CBCT means performance metrics do not reflect modern adjunct-heavy workflows.

💡 The Clinical Bottom Line

The PRECIsE study validates that shape-sensing robotic bronchoscopy provides high diagnostic sensitivity (~81.3% at 2 years) and an excellent safety profile (1.6% clinically significant pneumothorax rate) even among nascent robotic users. Clinicians must note that despite high technological sensitivity, nondiagnostic index biopsies carry a high posterior probability of malignancy (~59.5%), necessitating aggressive and vigilant clinical follow-up. Furthermore, the 2-year data confirm that standard 12-month follow-up windows slightly overestimate true diagnostic sensitivity.