Lung cancer is presently the leading cause of cancer-related mortality worldwide.1 Widespread efforts have been made to effectively detect and diagnose lung cancer.2–4 Recently, the wide implementation of lung cancer screening with low-dose chest computed tomography has considerably increased the identification of suspicious pulmonary lesions that require diagnostic evaluation.5,6 Diagnosing pulmonary lesions involves the use of various techniques, including bronchoscopy, transthoracic approach, and surgery. According to clinical practice guidelines, the least invasive approach with the highest yield is recommended for diagnosing pulmonary lesions suspicious for lung cancer.7–9 For peripheral lesions difficult to reach with conventional bronchoscopy, transthoracic needle aspiration (TTNA) has been commonly adopted. However, despite a good sensitivity of approximately 90%, complications are not uncommon with TTNA, with a reported pneumothorax rate of >18% of the biopsied cases.10,11

Recently, advances in image-assisted technologies using flexible bronchoscopy have improved diagnostic yield and safety. Leading bronchoscopy techniques include electromagnetic navigation bronchoscopy (ENB) and radial endobronchial ultrasound (R-EBUS).12 ENB and R-EBUS allow physicians to access peripheral lung lesions through a minimally invasive technique using a dedicated navigation software system (ENB) or a 20-MHz ultrasound probe (R-EBUS), respectively. According to current guidelines, ENB or R-EBUS is recommended for pulmonary lesions that cannot be reached with conventional flexible bronchoscopy alone.7,13 For ENB, one large multicenter prospective study,14 followed by a recent meta-analysis,15 has demonstrated that ENB-guided diagnosis has a diagnostic yield of 68–77%. Meanwhile, R-EBUS has been reported to have a diagnostic accuracy of approximately 70%.16,17 Both techniques provide an acceptable performance for diagnosing lung cancer, with markedly lower complication rates than TTNA.18

However, to date, information on patient-level data comparing ENB and R-EBUS, especially when performed under moderate sedation, is limited. Despite the theoretical advantages on localization and the higher diagnostic yield of ENB than R-EBUS demonstrated in separate pooled-analyses,15,17–19 a direct comparison of the two procedures are rarely available with uncontrolled lesion characteristics and sedation methods.12,18 Therefore, we aimed to evaluate and compare the diagnostic performance and complication rates of ENB and R-EBUS under moderate sedation for diagnosing pulmonary lesions suspicious for lung cancer. We performed propensity score matching between patients who underwent ENB and those who underwent R-EBUS including lesion characteristics to control for pre-procedural factors and address selection bias.

To the best of our knowledge, this study represents the first patient-level evaluation to compare sole ENB-guided and sole R-EBUS-guided biopsy under moderate sedation for diagnosing pulmonary lesions. In a 1:1 propensity score-matched cohort constructed to achieve balance for pre-procedural factors, our study revealed a significantly higher diagnostic yield and sensitivity for malignancy for ENB than for R-EBUS. Sole ENB was superior to sole R-EBUS under moderate sedation for lesions>20mm in size, radiologically solid lesions, and lesions with a class 2 bronchus sign (an airway directly aligned with the lesion on CT scan). In the multivariable analysis using the unmatched cohort and the analysis after propensity score matching, using ENB over R-EBUS was significantly associated with a higher diagnostic yield. The complication rate was generally low (4.8% for pneumothorax) in this matched cohort, with no procedure-related respiratory failure or mortality events. No significant difference in the safety profiles was observed between the two techniques. Our study provides important insight on the utility of ENB over R-EBUS by directly comparing both techniques in a least-invasive setting from an Asian population, of which current data are insufficient.

Currently, guidelines on the diagnosis of lung lesions recommend that the type of diagnostic biopsy should be based on the location, size, invasiveness of the available procedure, and the possible risk of complications.8,9 Recent advances in image-assisted bronchoscopy technologies, such as ENB and R-EBUS, which demonstrated improved diagnostic yield and safety, have made transbronchial approach more feasible.25,26 To date, most studies on the utility and safety of ENB or R-EBUS have reported a diagnostic yield between 38% and 90%.16,25–28 The most recently reported meta-analyses on the diagnostic performance of these advanced bronchoscopic techniques have presented that for ENB, a pooled diagnostic yield of 76.4% and a sensitivity of 70.7–77% was achieved.15,18 For R-EBUS, a pooled diagnosed yield of 72.4% and a sensitivity of 70.5–72% were reported.17,18 However, significant variations on the performance patterns exist within and between studies in terms of case selection, concurrent use of both modalities, mixed use of additional complementary techniques, and type of anesthesia. This hampers the relevant evaluation and comparison between both techniques performed in a minimally invasive setting.

The NAVIGATE study, which is the largest multi-center study on ENB to date, has reported that among the 1388 participants who underwent ENB, 50.6% received concurrent R-EBUS guidance. Other complementary techniques included fluoroscopy in 85.0% and ROSE in 61.7% of the cases, and 78.2% received the procedure under general anesthesia.14 Similarly, a substantial proportion of published studies on the performance of R-EBUS presented results with additional complementary tools such as fluoroscopy, virtual bronchoscopy, and ROSE,29–32 without strict control for the type of sedation.33,34 The difficulty to control for and compare different bronchoscopy techniques in various settings is also noticeable from the results of the AQuIRE registry, which revealed significant variations on practice and sedation patterns for diagnostic bronchoscopy.26 The only available randomized controlled trial to compare the performance of sole ENB and R-EBUS at a patient level to date was reported by Eberhardt et al., which revealed an insignificant difference between the diagnostic yield of ENB (59%) and R-EBUS (69%). However, this study was not designed to strictly control for lesion characteristics and type of sedation.12 In this aspect, the design and results of our study are particularly relevant because there is no study comparing the diagnostic yield of sole ENB and R-EBUS matched for pre-procedural characteristics including radiologic findings. Moreover, in our study cohort, owing to the minimalized practice standards on guided bronchoscopy in the institute where the study was conducted, the use of additional technologies and the type of sedation was completely controlled. Our study also provides relevant data on the performance and safety of the sole procedures under moderate sedation without the use of complementary tools. Regarding data that suggest that general anesthesia can be associated with a greater need for post-procedure escalation of care in bronchoscopic procedures,35,36 sole procedure under moderate sedation would be the least-invasive setting for ENB and R-EBUS. Accordingly, our data would serve as a good reference across numerous clinical settings with no access to general anesthesia or other complementary techniques.

In our study, the diagnostic yield achieved by sole ENB was higher than the generally reported results from previous studies.18,27 Meanwhile, the yield of R-EBUS may be marginally lower.16,18,28 The relatively low diagnostic yield achieved by R-EBUS in our study could have been contributed by the infeasibility to use an aspiration needle for tissue sampling and complementary ROSE. Although transbronchial forceps biopsy remains the mainstream modality for tissue sampling for both ENB and R-EBUS,26,37,38 a recent updated meta-analysis on the diagnostic performance of R-EBUS has reported that the use of ROSE was associated with a higher sensitivity.17 However, ROSE may not solely impact the diagnostic value of R-EBUS as it is more likely to be used and associated with a higher diagnostic yield when transbronchial needle aspiration (TBNA) is performed;39 TBNA can also be a driving factor for increased performance of R-EBUS.26,40 By contrast, a recent updated meta-analysis on the diagnostic performance of ENB has revealed no significant difference with using complementary ROSE or using different type of biopsy tools.15 Likewise, our study revealed a similar diagnostic yield achieved by ENB-guided TBNA and transbronchial forceps biopsy.

In our study, after adjusting for potential factors (e.g. lesion size, location, and presence and the type of the bronchus sign) reported by single studies to affect the diagnostic yield of guided bronchoscopy,21 the use of ENB over R-EBUS was associated with a higher overall diagnostic yield and higher sensitivity for malignancy, without difference in the complication rates. This emphasizes the superior utility of ENB over R-EBUS when performed for similar pre-procedural settings. In most clinical settings, R-EBUS is more widely accessible than ENB since the expense of ENB consumables required per case is higher than R-EBUS. The rationale for using ENB over R-EBUS would be that the former may offer a higher diagnostic yield, as shown in our study.19 In addition, the higher proportion of those who underwent sequential linear EBUS for nodal evaluation in the ENB group than the R-EBUS group with similar total procedure time may indicate an additional benefit of shortening the total procedural time using ENB over R-EBUS. When selecting a sole modality under moderate sedation, as long as the financial cost for the health care facility is acceptable, using ENB would provide a better yield for diagnosing malignant lung lesions. Particularly, ENB would be recommended for solid lesions with a class 2 bronchus sign.

This study had some limitations. First, this was a retrospective study of patients from a single center, Therefore, the results may not be generalizable to other clinical settings, emphasizing the need for a well-designed prospective randomized controlled trial to corroborate our findings and apply them into future guidelines. Although we performed propensity score matching, an uncontrollable selection bias might have remained (e.g. the attending physician's preference for a particular situation). Second, our results may not validate the utility of sole ENB or R-EBUS under moderate sedation over other performance patterns in different clinical settings. Although the combination of ENB and R-EBUS and the combination of various tissue sampling techniques (e.g. transbronchial biopsy and TBNA) may improve the diagnostic yield, the current data on the efficacy of combining technologies and biopsy tools remain inconsistent. Moreover, concerns regarding financial burden increase with the combined use of various techniques, resulting in clinical settings with limited accessible modalities. Third, the number of lesions ≤20mm included in our study was relatively small. Therefore, the power to detect a statistical significance for the difference in the diagnostic performance between ENB and R-EBUS for smaller lesions may have been insufficient. In addition, a substantial proportion of lesions larger>30mm was included in our study. The relatively larger size of the included lesions could have possibly contributed to the better diagnostic performance of ENB. Nevertheless, our study provides valuable information on the superiority of sole ENB over R-EBUS in a least-invasive setting, and data on the mainstream modality with the least combination of complementary techniques. This would be the clinical setting that a maximal number of physicians would be accessible.

In conclusion, from a propensity score-matched analysis controlled for pre-procedural factors, sole ENB demonstrated a higher diagnostic yield than sole R-EBUS under moderate sedation for diagnosing pulmonary lesions suspicious for lung cancer, with similar and generally low complication rates. Our data support the utility of ENB over R-EBUS in a least-invasive setting and serve as a good reference for the choice of modality in similar clinical settings which would be accessible to a maximum number of physicians performing guided-bronchoscopy.

YWK had full access to all of the study data and take responsibility for the integrity of the data and the accuracy of the data analysis, including and especially any adverse effects. YWK, HJK, and SHY contributed to the concept and design of the study, data collection, data analysis, data interpretation, manuscript preparation, revision, and final approval of the manuscript. MJS, BSK, SYL, YJL, JSP, YJC, JHL, and CTL contributed to the data collection, data analysis, revision, and final approval of the manuscript.

The authors declare no conflict of interest.