Background
Guidelines for lung nodule evaluation recommend the least invasive approach given each patient’s clinical presentation [
1]. Utilization of electromagnetic navigation bronchoscopy (ENB) has increased over the past ten years as a minimally invasive approach to complement traditional bronchoscopy, endobronchial ultrasound (EBUS), and image-guided transthoracic biopsy. Selection of the most appropriate diagnostic modality based on patient comorbidities and lesion location is critical to provide the fastest, safest, and most complete diagnosis possible.
Seventeen published studies of ENB use have been summarized in three recent meta-analyses [
2‐
4]. Pneumothorax is the most common complication, occurring in approximately 3% of patients [
2], lower than the pooled 20% rate reported for transthoracic needle biopsy [
5]. However, published studies have typically been small, single-center, retrospective, and mostly conducted by expert users. The safety, usage profile, and clinical utility of ENB in a large, prospective, multicenter, generalizable population is unknown. The pragmatic design [
6] of NAVIGATE maximizes patient eligibility, usual care settings, flexibility of adherence, and a relevant primary outcome for clinical practice. The detailed prospective collection of data also minimizes retrospective bias and allows future multivariate analyses to provide more meaningful information on the variable utilization of this technology into real-world practice and its impact on measurable outcomes, such as diagnostic yield and risk. Furthermore, a heterogeneous dataset will be instructive for the design of potential comparative studies with respect to operator training, subject inclusion criteria, data to be collected, definitions, and expected complication rates.
The primary objectives of this protocol-specified 1,000-subject, 1-month interim analysis of the NAVIGATE study [
7] are to present the preliminary safety, clinical usage patterns, and performance of ENB in a large, unrestricted, generalizable population across diverse practice settings. The interim data will provide an early look at typical patient and lesion characteristics and procedural standard-of-care, generating questions for future NAVIGATE analyses and new clinical studies. Enrollment and continued follow-up are ongoing.
Methods
NAVIGATE is a prospective, multicenter, global, single-arm, cohort study in subjects undergoing ENB procedures. Enrollment of up to 1,500 subjects is planned at 37 sites in the United States and Europe. Subjects evaluations occur at baseline (within 30 days of the procedure), on the procedure day, and at 1 month, 12 months, and 24 months post-procedure. This manuscript describes the results of a prespecified 1-month interim analysis of the first 1,000 subjects enrolled at 29 sites in the United States and Europe. Enrollment and 12- and 24-month follow-up are ongoing. Brief methods are included below. A full list of study assessments and definitions is included in Additional files
1 and
2. The study design has been published [
7].
Inclusion criteria are intentionally broad to ensure external validity. All consecutive, consented adult patients, who are not pregnant or nursing, and who are candidates for an elective ENB procedure based on physician discretion per recommended guidelines and institutional standard-of-care, are eligible. A maximum of 75 subjects is allowed per site. All investigators must have prior ENB experience. Investigators without extensive experience may enroll a maximum of five “roll-in” cases, which are excluded from this interim analysis. Roll-in cases will be included in the 1-year and 2-year analyses of the full enrollment when a more complete evaluation of the impact of user experience on diagnostic yield and other outcomes can be conducted.
All ENB procedures use the superDimension™ navigation system [
8,
9] version 6.0 or higher (Medtronic, Minneapolis, MN) per product instructions and institutional standard practice. All complementary tools and procedures, including choice of catheter and biopsy tools, order of biopsy tool use, and strategy for staging and diagnostic bronchoscopy were performed at clinician discretion and were captured prospectively for data analysis.
The primary endpoint is pneumothorax related to the ENB index procedure rated Grade ≥2 according to the validated Common Terminology Criteria for Adverse Events (CTCAE) scale [
7,
10], as adjudicated by an independent medical monitor. Pneumothorax was protocol-specified as the primary endpoint because it is applicable to all ENB procedures, including lung lesion biopsy, lymph node biopsy, fiducial placement, and pleural dye marking. Major secondary endpoints were all ENB-related pneumothorax, bronchopulmonary hemorrhage, and respiratory failure. Other secondary endpoints reported at 1 month were subject self-reported satisfaction with the procedure; adequacy of samples for molecular testing and mutation type; accurate fiducial placement as assessed by follow-up imaging; and success rate of pleural dye marking demonstrated by surgical resection [
7].
Diagnostic yield of the ENB index procedure will be calculated at the 12- and 24-month follow-up as the proportion of subjects with a definitive diagnosis (final pathology of the ENB-aided sample). One-month follow-up in this interim analysis is not sufficient to calculate the true negative rate or diagnostic yield. All lung nodules evaluated during the ENB index procedure will be followed for confirmation. Sensitivity, specificity, negative predictive value, and positive predictive value will be published beginning with the 12-month follow-up.
No sample size calculations were conducted for this single-arm, observational study. Analyses were performed using SAS® Version 9.4 (SAS Inc., Cary, NC). Data are summarized by descriptive statistics, including frequency distributions and cross-tabulations for discrete variables and mean, standard deviation, median, minimum, and maximum values for continuous variables. At least 10% of the data are verified against source files by the sponsor using risk-based monitoring.
Discussion
Lung cancer causes one quarter of all cancer deaths, representing a significant public health problem [
12]. While the incidence has declined in concert with decreased smoking prevalence, survival rates have improved little over the past 50–60 years, largely due to a high proportion of late-stage diagnoses with a 5-year survival rate of only 4% [
12]. Earlier-stage diagnoses will lead to more meaningful improvements in survival and will require modalities that allow the accurate sampling of smaller, more peripheral lung lesions. The National Lung Screening Trial [
13] and screening coverage in select high-risk patients [
14] has been projected to increase low-dose CT utilization by over ten million procedures annually [
15]. However, an extremely high percentage (96%) of false positive screening results [
13,
16] and the risk of unnecessary procedures requires the judicious use of minimally invasive options and a careful balance of the risk-to-benefit ratio for further diagnosis and management [
17].
Several current technologies can provide minimally invasive diagnostic evaluation in appropriately selected patients, although each has limitations. PET-CT is often considered the second-line diagnostic option for nodules detected on CT [
18], but is typically not reimbursed for screening and does not provide tissue diagnosis. Conventional bronchoscopy is safe but is limited to proximal lesions and has a high non-diagnostic rate, potentially leading to unnecessary invasive procedures in 20–25% of patients, including the use of thoracoscopy for diagnostic wedge resection [
19‐
21]. Image-guided transthoracic biopsy provides high diagnostic accuracy but at the cost of pneumothorax rates averaging 20% (range 4 to 62%) [
5] and the need for additional procedures to diagnose and stage mediastinal lymph nodes. ENB provides a minimally invasive platform for peripheral lung lesion sampling, concurrent lymph node staging with linear EBUS, preparation for treatment via fiducial placement or localization via pleural dye marking in a single procedure.
The primary objective of this interim NAVIGATE analysis was to evaluate ENB safety. While published pneumothorax rates are low (3.1% [range 0–13%]) [
2], most prior studies are single-center with fewer than 100 subjects [
4,
7]. The current analysis demonstrates low pneumothorax, bronchopulmonary hemorrhage, and respiratory failure rates in the context of a large, diverse study cohort and a wide range of user experience levels, confirming the safety of advanced bronchoscopy for the access and sampling of all pulmonary nodules. Pneumothorax was also infrequent following fiducial placement (3.8%), in contrast to rates ranging from 22 to 67% following percutaneous fiducial marking [
22], Despite the advanced stage of some of the enrolled subjects, there were only 23 deaths within the 1-month follow-up timeframe, further substantiating the safety of the procedure. Only one death was considered related to the ENB index procedure, due to general anesthesia in a patient with multiple comorbidities, and none were related to the ENB device or associated tools. These results suggest a highly favorable risk-to-benefit ratio for the use of ENB to aid in lung lesion biopsies, particularly given the risk profile of the patients included, with approximately 45% COPD incidence and a relatively high rate of Stage III-IV disease.
A second objective of this analysis was to explore the real-world usage patterns and clinical utility of ENB. The interim results elucidate the rates of general anesthesia use (79.7%), ROSE utilization (66.1%), and concurrent fluoroscopic (90.1%) and radial EBUS (54.3%) guidance. Of note, nearly half of ENB index procedures were conducted for multiple purposes, including 33.4% with lymph node staging (primarily EBUS-guided) and 21.0% with fiducial markers placed. Tissue adequacy for molecular genetic testing was also high (80.0%) and similar to prior studies [
23]. These results suggest that, in unrestricted practice settings, ENB is used to diagnose peripheral lung nodules and perform concurrent linear EBUS-guided mediastinal lymph node staging in a single anesthetic event, facilitating a multidisciplinary, comprehensive patient care approach.
A third objective of this interim analysis was to provide a preliminary look at ENB performance. From a patient perspective, all important follow-up cadence and treatment decisions are made within the 30-day window after the diagnostic procedure. At the 1-month time-point, ENB provided a preliminary malignant diagnosis in 45.8% of subjects, including 40.1% with lung cancer. The initial 45.8% malignancy rate in NAVIGATE is consistent with other recent ENB publications reporting malignancy rates of 35–60% [
24‐
27], and is expectedly higher than the 3.7% positive malignancy rate seen in the National Lung Screening Trial [
13].
One-month follow-up is not sufficient to calculate the true versus false negative rate or diagnostic yield, as the true prevalence of lung cancer in this population is unknown at this time. All non-malignant pathology findings require confirmation with longer-term follow-up or additional diagnostic procedures, depending upon the pretest probability of malignancy and in accordance with society guidelines [
1,
18]. All follow-up procedures and final diagnoses will be captured and reported. Early indicators of clinical stage in NAVIGATE subjects diagnosed with lung cancer also suggest a 64% rate of Stage I-II diagnoses, which are more amenable to surgical intervention for curative intent. In this observational study with consecutive enrollment, approximately 36% of NAVIGATE subjects had Stage III-IV lung cancer. Diagnostic testing of late-stage patients in NAVIGATE may reflect not only a lack of standardization for patient selection but also the changing landscape of personalized medicine and treatment options for Stage III-IV disease. Patient selection for ENB, as well as multivariate predictors of safety and effectiveness, will be explored in future NAVIGATE analyses of the full cohort. This will include an analysis of Stage III-IV cases to explore the patient, lesion, and operator/center factors leading to the inclusion of these cases in the study.
The final objective of this preliminary analysis was to generate questions for future NAVIGATE analyses and comparative studies. Unexpected observations included a high percentage of lesions without a CT bronchus sign (52%) and a relatively low proportion of subjects in whom genetic testing was attempted (28%). While current guidelines recommend genetic testing for only late-stage disease, there is extensive variation between institutions. Tissue requirements for comprehensive molecular testing and the practice of personalized medicine will continue to evolve. Future analyses will describe molecular genetic evaluation in the NAVIGATE cohort in more detail. Other future questions include multivariate predictors of safety and diagnostic yield, factors affecting the need for concurrent radial EBUS or other fluoroscopic guidance, usage patterns of fiducial and pleural dye marking, success rates of various biopsy tools, and cost effectiveness. In this way, NAVIGATE will help to set the benchmark for the ideal ENB patient, and define the procedural techniques contributing to enhanced performance. Whether ENB truly enables a shift to an earlier stage diagnosis, and the impact on long-term patient survival, healthcare utilization, and costs, will also be topics for future NAVIGATE analyses.
Limitations
This is a nonrandomized, single-arm analysis of 1-month interim results. Longer-term follow-up is required to determine the accuracy of ENB-aided diagnoses, and calculate diagnostic yield. Follow-up through 24 months is in progress. This analysis also evaluates only one navigational bronchoscopy system; other systems are currently available for clinical use.
Acknowledgements
The study is sponsored and funded by Medtronic (Minneapolis, MN). Medical writing support was provided by Kristin L. Hood PhD, a full-time employee of Medtronic and coauthor on this paper. The authors wish to thank Haiying Lin (Medtronic) for biostatistics support. The authors also wish to thank the investigators and staff of all participating clinical sites (see Additional file
3).