Virtual vs. reality in cone-beam CT-guided robotic-assisted bronchoscopy

Abstract

Background

The diagnostic accuracy of shape-sensing robotic bronchoscopy (ssRAB) is impacted by computed tomography-to-body divergence (CTBD) due to a discrepancy between virtual and actual lesion locations. The prevalence, types, implication, and determinants of CTBD during ssRAB remain understudied. This study aimed to qualify and quantify CTBD during cone-beam computed tomography (CBCT)-guided ssRAB and identify contributing factors.

Methods

This was an analysis of a prospectively curated database of patients who underwent mobile CBCT-guided ssRAB for the sampling of parenchymal lung lesions between October 2021 and June 2022. The primary outcome was the distance and vector of target displacement between the virtual preoperative CT-based and the actual intraoperative CBCT-based relationship between the tip of the ssRAB catheter and the target lesion.

Results

A total of 81 lesions in 70 participants were included in the final analysis. Displacement was identified in 67% of targeted lesions. In 78% the ssRAB catheter was closer than expected (forward displacement) and an overshoot beyond the lesion occurred in 46% of displacement instances. In a multivariable logistic regression analysis, ≥7 traversed airway generations were independently associated with target lesion displacement. Despite a high level of displacement, there was no statistically significant difference in diagnostic yield rates between cases with and without displacement.

Conclusions

CTBD, i.e., displacement of the actual target lesion from the virtual location, is common during navigational bronchoscopy. When targeting peripheral lesions, the catheter is more likely to be closer to the lesion or beyond it than expected based on the virtual planning. Bronchoscopists should anticipate CTBD and integrate data from various intraoperative imaging tools to attenuate its impact on sampling and avoid potential complications.

Highlight box.

Key findings

• Among 81 lesions sampled via cone-beam computed tomography (CBCT)-guided robotic-assisted bronchoscopy, a discrepancy between the expected and the actual robotic catheter tip-to-lesion distance occurred in 67% of lesions.

• Forward displacement (catheter closer to the lesion than expected) occurred in 78% of displacement instances.

• Overshoot displacement (catheter beyond the lesion) occurred in 46% of displacement instances.

• The traversal of ≥7 airway generations was an independent predictor of any displacement, while lower tidal volume was an independent predictor of overshoot displacement.

What is known and what is new?

• Robotic-assisted bronchoscopy improves parenchymal lung lesion access and sampling but is limited by CT-to-body divergence.

• This study defines types of catheter-lesion displacements and quantifies the prevalence and degree of displacement during robotic-assisted bronchoscopy. This study highlights the importance of overshoot displacement as a clinically significant subtype of CT-to-body divergence.

What is the implication, and what should change now?

• Operators should anticipate forward and overshoot displacement when targeting lesions in the lung periphery or when applying lower mechanical ventilation tidal volumes.

• When targeting peripheral lesions, operators should consider positioning the catheter within a safety proximal distance of the virtual target to account for forward displacement.

• Future studies are needed to define and validate best practices regarding ventilation parameters that balance adequate oxygenation and ventilation with prevention of atelectasis and displacement.

Introduction

Robotic-assisted bronchoscopy (RAB) is an emerging iteration of navigational bronchoscopy for the sampling of parenchymal lung lesions (1). Virtual target marking and endobronchial navigation of the robotic catheter are based on a segmented airway “roadmap” derived from a preoperative computed tomography (CT) study (2). While the CT is typically performed under spontaneous respiration, navigational bronchoscopy is performed under general anesthesia, often with neuromuscular blockade, and with the application of a relatively high positive expiratory pressure (PEEP) as well as recruitment maneuvers in some instances. Fundamental differences in lung conformation and respiratory dynamics between the preoperative CT and intraoperative ventilation mechanics result in a discrepancy between the target lesion’s predefined virtual and actual locations. This mismatch is termed CT-to-body divergence (CTBD) (3).

Despite improved maneuverability into the lung periphery and stability throughout the tissue sampling process with RAB, CTBD is a major contributor to sampling errors and insufficient tissue acquisition (4). The incorporation of cone-beam CT (CBCT) imaging into navigational bronchoscopy augments navigation accuracy by allowing intraoperative correction for CTBD (5-7). However, determinants of CTBD during navigational bronchoscopy remain largely unknown.

The identification of factors that promote CTBD can assist navigational bronchoscopists to anticipate, correct for, and potentially avoid CTBD, thus improving sampling accuracy. The current study aimed to identify patient, procedure, and target lesion factors associated with CTBD, as indicated by a discrepancy between the distance between the tip of the shape-sensing robotic bronchoscopy (ssRAB) catheter and the virtual target view and the actual distance, as quantified by intra-operative CBCT imaging. We further classified the catheter displacement in relation to the target based on the expected and actual catheter location. Secondary aims included diagnostic yield, diagnostic accuracy for malignancy, and interobserver and intraobserver agreement regarding the expected and actual intraoperative radial endobronchial ultrasound (EBUS) view. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1399/rc).

Methods

This was a retrospective analysis of a prospectively curated database (8) of shape-sensing RAB (ssRAB) procedures performed at Memorial Sloan Kettering Cancer Center (MSK) between October 2021 and June 2022 for the diagnostic sampling of peripheral lung lesions. Cases in which an orbital CBCT volumetric imaging (i.e., spin) was performed after the completion of the ssRAB navigation phase and prior to sampling tool deployment were included in the final analysis. Cases were excluded if no CBCT spin was performed, there was an inability to identify the target lesion on CBCT due to atelectasis or hemorrhage, if data indicating the virtual distances from the ssRAB catheter to the lesion were incomplete, or the first spin was performed after tool deployment. The latter exclusion criterion was due to concern that tool deployment could result in bleeding or other alterations that would make distance measurements inaccurate. Abstracted data included patient demographic and clinical characteristics, radiographic lesion characteristics, intraoperative mechanical ventilation settings and output measures, intraoperative imaging and sampling variables, tissue sampling results, and procedure-related complications. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of the Memorial Sloan Kettering Cancer Center (#20–166). Informed consent was waived due to the retrospective nature of the study design.

Preprocedural assessments and procedure performance

Preoperative CT imaging for navigational planning with 0.625–1.25-mm slices was obtained in all cases. Lesion size was defined as the largest dimension across the axial, sagittal, and coronal projection CT images. The number of traversed airway generations was defined as per Weibel (9) and determined from the schematic airway tree generated by the navigation planning software. Two experienced interventional pulmonologists [B.C.H. and O.K.D., both with >100 navigational bronchoscopies over the prior ≥4 years (10)] independently reviewed each CT to indicate the expected radial EBUS view as concentric, eccentric, or no view (11). ssRAB was performed using the Ion Robotic-Assisted Endoluminal Platform™ (Intuitive Surgical Inc., Sunnyvale, CA, USA). All procedures were performed under general anesthesia with neuromuscular blockade and via an endotracheal airway. At the time of the study, the institutional ssRAB ventilation protocol consisted of a PEEP of 8–10 cm H₂O, tidal volume (VT) of 8–10 mL/kg of ideal body weight (IBW), and fraction of inspired oxygen of 0.4, as tolerated by the patient. Recruitment maneuvers were not performed as part of the institutional ventilation protocol during the study period. Induction technique and the mode of anesthetic delivery were left to the discretion of the anesthesiologist. Neuromuscular paralysis was induced and maintained with rocuronium in all cases. Navigation was considered complete once the catheter was locked over the target lesion in preparation to confirm localization and commence sampling. At the completion of the navigation phase, radial EBUS (UM-S20-17S or UM-S20-20R, Olympus Corp., Tokyo, Japan) was deployed to identify the sonographic relationship of the airway and lesion as concentric, eccentric, or no view (11). Subsequently, a breath hold was initiated at peak inspiration with the adjustable pressure-limiting (APL) valve pressure set to a pressure that was 0-5 cm H₂O over the PEEP level. Once the airway pressure has plateaued, the expected distance between the tip of the catheter and the proximal border of the virtual target as indicated by the ssRAB system was documented. A volumetric mobile CBCT spin (Cios-Spin™ Mobile 3D C-Arm, Siemens Healthineers Inc., Erlangen, Germany) was performed simultaneously. The reconstructed volumetric images were then reviewed to identify the catheter-lesion relationship and document the actual distance between the tip of the catheter and the most proximal border of the target lesion. The distance was measured on a reconstructed CBCT hybrid image that was generated by aligning the imaging plane along the central axis of the robotic catheter. Sampling of the target lesion was then pursued with or without further catheter adjustments or CBCT spins as per the discretion of the operator. The performance of additional spins for tool-lesion relationship refinement or to confirm tool-in-lesion as well as the choice of sampling tools were left to the discretion of the operator. Rapid on-site cytology evaluation (ROSE) was available for all procedures. Case time was measured between docking and undocking of the ssRAB system (detailed definitions for key variables are provided in Appendix 1).

Study outcomes

The primary outcome of the study was target lesion displacement as indicated by a discrepancy between the expected and actual tip-of-catheter to target lesion measured distance. The study definitions are illustrated in Figure 1. The expected value was obtained from the ssRAB system at the completion of the navigation phase, while the actual value was obtained from the CBCT spin images in the most appropriate projection. This distance was quantified in millimeters with a positive value indicating catheter position proximal to the nearest border of the target and a negative value indicating catheter position beyond the nearest border of the target. Overall displacement distance was calculated by subtracting the expected distance from the actual distance. Displacement was defined as forward if CBCT showed the catheter tip to be closer to the nearest aspect of the lesion than expected and backward if CBCT showed the catheter tip to be further away from the nearest aspect of the lesion than expected. Overshoot displacement was a subtype of forward displacement in which CBCT showed the catheter tip to be beyond the proximal border of the lesion.

Figure 1.

Study definitions for catheter-target expected vs. actual distance. Distance is measured in millimeters and indicates the shortest distance between the tip of the catheter and the nearest aspect of the target lesion. The expected distance is documented at the completion of the navigation phase. The actual distance is the shortest measurement on cone-beam CT imaging in the most appropriate projection. A positive distance value indicates the catheter to be located proximal to target lesion, while a negative distance value indicates that the catheter is located beyond the nearest aspect of the target lesion. The displacement distance was calculated by subtracting the expected distance from the actual distance. CBCT, cone-beam computed tomography.

Secondary outcomes included (I) diagnostic yield by the conservative (12) and intermediate (13) definitions, (II) diagnostic accuracy for malignancy, (III) preprocedural interobserver agreement regarding the expected radial EBUS view, and (IV) intraobserver agreement between the expected and the actual radial EBUS view.

Statistical analysis

Categorial variables are presented as counts and percentages while continuous variables are presented as medians and interquartile range (IQR). Comparisons of patient- and lesion-level characteristics across displaced and nondisplaced cases were performed using the Wilcoxon rank-sum, the chi-squared, the Fisher’s Exact, and the Cochran-Armitage Trend tests as appropriate. Mechanical ventilation parameters, including PEEP and VT, were dichotomized into low and high groups based on the median value across the study cohort. A multivariable logistic regression was then constructed to explore factors associated with any displacement and “overshoot” displacement using factors that had a P value of <0.05 in the univariable analysis and adjusted odds ratios were reported.

Diagnostic yield and accuracy were calculated using standardized definitions (12). Inter- and intra-observer agreement regarding anticipated and actual radial EBUS views were calculated using the McNemar’s Test and weighted Kappa statistics.

A P value of <0.05 was considered statistically significant. SAS Software version 9.4 (SAS Institute Inc., Cary, NC, USA) and R version 4.3.1 (R Foundation for Statistical Computing) were utilized for all analyses.

Results

During the study period, CBCT-guided ssRAB was performed on a total of 91 lung lesions in 75 patients. Ten lesions were excluded due to target lesion obscuration on CBCT by atelectasis or hemorrhage (n=5) and lack of intraoperative distance measurements (n=5). The final analysis included 81 lesions in 70 patients (Table 1). The median age was 69 years (IQR, 58–76), 60% were female, and 89% had a known history of cancer at the time of the procedure (Table S1). The median lesion size was 20.0 mm (IQR, 14.0–27.2), 54% were in the upper lobe, 59% had a solid CT consistency, 44% were subpleural, and 41% had a positive CT bronchus sign (Table 1). The median case time was 52 minutes (IQR, 41–65). The median PEEP was 8.5 (IQR, 8.0–10.0) and the median VT was 9.6 mL/kg of IBW (IQR, 8.7–10.4). A median of 2 CBCT spins (IQR, 1–2) were performed and more than one lesion was sampled in 14% of cases. The radial EBUS view was concentric, eccentric, and no view in 32%, 40%, and 28% of sampled lesions, respectively. Sampling was performed by needle aspiration, forceps biopsy, cryobiopsy, and bronchoalveolar lavage in 98%, 69%, 25%, and 41% of the lesions, respectively (Table 1). There were no instances of pneumothorax or grade ≥3 bleeding (14). One patient was admitted post-procedure for grade 1 decompensated heart failure.

Table 1. Baseline characteristics of patients, target lesions, and procedures.

Variable	Value
Patient level (n=70)
Age, years	69 (58–76)
Female sex	42 (60)
IBW, kg	57.9 (52.4–67.4)
BMI, kg/m2	24.2 (21.4–29.0)
<30.0	57 (81)
≥30.0	13 (19)
Smoking status
Never	32 (45)
Ever	38 (55)
Former	33 (87)
Current	5 (13)
ASA Score
1	2 (3)
2	5 (7)
3	59 (84)
4	4 (6)
Known history of cancer
None	8 (11)
Any†	62 (89)
FEV1, L	2.19 (1.90–3.51)
FEV1, % predicted	95 (80–105)
DLco, % predicted	76 (60–92)
Preoperative radiographic lesion level (n=81)
Lesion size, mm	20.0 (14.0–27.2)
<10.0	6 (7)
10.0–19.9	33 (41)
≥20.0	42 (52)
Radiographic consistency
Solid	48 (59)
Parenchymal infiltrate	12 (15)
Part-solid	10 (12)
Pure ground-glass opacity	6 (8)
Cavitary	5 (6)
Lobar location
Right upper	22 (27)
Right middle	8 (10)
Right lower	14 (17)
Left upper and lingula	22 (27)
Left lower	13 (16)
Mediastinal lymph node	2 (3)
Traversed airway generations	7 (6–8)
Lung centrality
Inner 2/3	38 (47)
Outer 1/3	43 (53)
Subpleural‡	36 (44)
CT bronchus sign
Positive	33 (41)
Negative	48 (59)
Expected radial EBUS view by Observer 1
Concentric	28 (38)
Eccentric	37 (51)
No view	8 (11)
Expected radial EBUS view by Observer 2
Concentric	34 (45)
Eccentric	27 (36)
No view	14 (19)
Procedure-level (n=70)
Total procedure time, minutes	52 (41–65)
Number of sampled lesions
1	60 (86)
2	9 (13)
3	1 (1)
Number of CBCT spins	2 (1–2)
1	38 (47)
2	24 (30)
3	13 (16)
4	5 (6)
5	1 (1)
PEEP, cmH2O	8.5 (8–10)
Tidal volume, mL/kg IBW	9.6 (8.7–10.4)
Intraoperative lesion level (n=81)
Actual radial EBUS view
Concentric	26 (32)
Eccentric	32 (40)
No view	23 (28)
Sampling tools
TBNA	79 (98)
TBFB	56 (69)
TBCB	20 (25)
BAL	33 (41)

Data are presented as median (interquartile range) or n (%). ^†, details in Table S1. ^‡, defined as ≤1.0 cm from the nearest pleural surface. ASA, American Association of Anesthesiologists; BAL, bronchoalveolar lavage; BMI, body mass index; CBCT, cone-beam computed tomography; CT, computed tomography; DL_co, carbon monoxide diffusion capacity; EBUS, endobronchial ultrasound; FEV1, forced expiratory volume in one second; IBW, ideal body weight; IQR, interquartile range; PEEP, positive end-expiratory pressure; TBCB, transbronchial cryobiopsy; TBFB, transbronchial forceps biopsy; TBNA, transbronchial needle aspiration.

Primary outcome: target lesion displacement

Target lesion displacement (i.e., discrepancy between the expected and actual tip of catheter to target lesion measured distance) was identified in 67% of cases (n=54). Of these, 78% (n=42) were forward displacement (i.e., catheter located closer to the lesion than expected), 22% (n=12) were backward displacement (i.e., catheter located more distant from the lesion than expected), and 46% (n=25) were an overshoot displacement (i.e., catheter beyond the closest border of the lesion). At the time of the first CBCT spin, the median expected distance was 1.0 mm (range, 0 to +20 mm) while the median actual distance was 0.0 mm (range, −10 to +11; P<0.001) with a median displacement distance of −1.0 mm (range, −20 to +10). The median displacement distance was −5.5 mm (range, −1 to −21 mm), 2.0 mm (range, 1 to 9 mm), and −7.0 mm (range, −21 to −2 mm) for forward, backward, and overshoot displacement cases, respectively. As illustrated in Figure 2, the degree of discrepancy between the expected and actual distance between the tip of the ssRAB catheter and the target lesion decreased across three sequential CBCT spins.

Figure 2.

Expected vs. actual distance between the tip of the robotic catheter and the target lesion across three CBCT spins. The box-and-whiskers plot describes the distribution of the median documented expected and actual distance between the tip of the robotic catheter and the closest aspect of the target lesion. Expected distance refers to the distance in millimeters as provided by the navigation system at the end of the navigation phase. Actual distance refers to the shortest distance in millimeters as measured on volumetric CBCT imaging in the most appropriate projection. The data is presented across the first 3 CBCT orbital spins. Wilcoxon matched pairs signed rank test was used to compare each pair of expected and actual distances across the CBCT spins. CBCT, cone-beam computed tomography.

Comparison of cases in which target lesion displacement was present vs. absent is shown in Table S2. In a univariable model, peripheral lung location, subpleural location, and ≥7 traversed airway generations were associated with target lesion displacement. In a multivariable model, ≥7 traversed airway generations remained independently associated with target lesion displacement (Table 2).

Table 2. Factors associated with target lesion displacement.

Variable	Univariable logistic regression model			Multivariable logistic regression model
	OR (95% CI)	P value		OR (95% CI)	P value
Factors associated with any displacement†
Lung centrality		0.01			0.89
Inner 2/3	1.00			1.00
Outer 1/3	3.40 (1.28–8.98)			1.13 (0.29–4.41)
Subpleural location‡		0.02			0.34
No	1.00			1.00
Yes	3.31 (1.20–9.13)			1.81 (0.53–6.17)
Number of traversed airway generations		0.001			0.03
<7	1.00			1.00
≥7	5.36 (1.97–14.56)			3.93 (1.12–13.77)
Factors associated with “overshoot” displacement§
Subpleural location‡		0.02			0.15
No	1.00			1.00
Yes	3.20 (1.19–8.54)			2.25 (0.73–6.92)
Number of traversed airway generations		0.03			0.07
<7	1.00			1.00
≥7	3.22 (1.06–9.81)			3.26 (0.88–12.03)
Tidal volume, mL/kg IBW		0.03			0.01
≥9.6	1.00			1.00
<9.6	2.83 (1.05–7.64)			3.94 (1.31–11.80)

^†, comparison index value indicating no displacement. ^‡, defined as <1 cm from the nearest pleural surface. ^§, comparison index value indicating lack of “overshoot” displacement, including no displacement and non-“overshot” displacement. CI, confidence interval; IBW, ideal body weight; OR, odds ratio.

Comparison of cases in which “overshoot” displacement was present vs. absent is shown in Table S3. In a univariable model, subpleural location, ≥7 traversed airway generations, and lower tidal volumes were associated with target lesion “overshoot” displacement. In a multivariable model, lower tidal volumes remained independently associated with target lesion displacement (Table 2).

Diagnostic yield and accuracy

Tool-in-lesion was confirmed by CBCT in 94% of cases, of which a CBCT spin was performed after tool deployment. ROSE was deemed adequate for sufficient diagnostic tissue in 75% of target samplings. The overall diagnostic yield rate was 82.7% by the conservative definition (12) and 92.6% by the intermediate definition (13). There was no statistically significant difference in diagnostic yield rates between cases with and without displacement (Table S4). Of note, the median number of CBCT spins was 1 (range, 1–3) and 2 (range, 1–5) in the groups without and with displacement, respectively (P=0.04). The distribution of diagnoses is shown in Figure 3 and Table S5. The sensitivity, specificity, positive predictive value, and negative predictive value for malignancy were 86%, 100%, 100%, and 81%, respectively, with an accuracy rate of 91% (Table S6).

Figure 3.

Distribution of diagnoses. AL, amyloid light-chain.

Radial EBUS view prediction

As illustrated in Table S7, the degree of interobserver agreement regarding the anticipated radial EBUS view based on the preoperative CT was fair (weighted Kappa =0.20, 95% CI: 0.01–0.39).

As illustrated in Table S8, the agreement between the anticipated and actual radial EBUS view was fair for Observer #1 (weighted Kappa =0.34, 95% CI: 0.15–0.53) and substantial for Observer #2 (weighted Kappa =0.65, 95% CI: 0.49–0.81).

Discussion

The widespread use of CT imaging for lung cancer screening and other indications has resulted in a surge in the detection of parenchymal lung lesions, which, in turn, has led to an increase in referrals for diagnostic sampling (15). CBCT-guided ssRAB is emerging as a reliable guided navigational bronchoscopy tool for the sampling of pulmonary lesions with diagnostic yield rates that are comparable to those of percutaneous transthoracic needle biopsy with a superb safety profile (16,17). To our knowledge, this is the first study to qualify and quantify the degree of intraoperative CTBD during CBCT-guided ssRAB for diagnostic sampling of parenchymal lung lesions. Our findings indicate that discrepancy between the actual target and the virtual target locations is common during ssRAB, occurring in 67% of cases. A majority of these were forward displacements, i.e., the catheter was located closer to the lesion than anticipated or beyond it.

Multiple studies reported a gap between bronchoscopic navigation success and diagnostic sampling rates (1,18,19). One factor contributing to this gap is CTBD. The virtual navigation roadmap is based on airway segmentation of a preoperative CT typically performed under spontaneous respiration with a deep inspiratory breath hold. Differences in respiratory mechanics between spontaneous breathing and mechanical ventilation result in a misalignment of anatomic structures between the intraoperative virtual view and actual catheter and target lesion positions (20). This phenomenon has been termed CTBD and applies to any guided navigational bronchoscopy procedure that relies on preoperative CT imaging for navigation planning. Factors contributing to CTBD include passive atelectasis, differences in patient positioning, non-physiologic tidal volumes, in electromagnetic navigation systems, metal interference, and possibly planning software limitations and inaccuracies (21-23).

Using a computerized algorithm, Reisenauer et al. reported a CTBD rate of 60% across 30 lesions targeted by CBCT-guided ssRAB (6). This rate is comparable to the one reported here and was more pronounced in the lower lobes compared with the upper lobes; however, the displacement vector was not reported. We have demonstrated that displacement was more common in peripheral lung lesions, particularly those requiring the traversal of ≥7 airway generations, independent of laterality or lobar location. Forward displacement was more common than backward displacement. This can be attributed to the higher propensity of peripheral lung tissue to undergo atelectasis as well as PEEP decay over airway generations (23,24). Additionally, inspiratory-expiratory oscillations in lung architecture tend to be more pronounced in the lung periphery compared with more central zones (25). Moreover, during navigation into smaller diameter airways, the robotic catheter tends to completely occupy the airway lumen thus impeding air flow and PEEP transmission. This can theoretically promote microatelectasis of the lung segment distal to the catheter and contribute to CTBD. Given these findings, when targeting peripheral lung lesions, we recommend that bronchoscopists regard the virtual target location with caution and anticipate forward displacement. Bronchoscopists should employ imaging tools that can assist with the confirmation of the catheter-target relationship prior to sampling tool deployment, such as radial EBUS and/or volumetric imaging.

In our study, “overshoot” displacement, i.e., the catheter is in reality beyond the lesion, accounted for 46% of displacement cases and 31% of all cases. While forward displacement without overshoot may mainly result in target obscurement due to metal artifact, overshoot displacement poses a risk of adverse outcomes. In subpleural lesions it may result in a pneumothorax by the mere advancement of the catheter or deployment of sampling tool via the visceral pleura. Similarly, it may result in diaphragmatic injury in the case of juxta diaphragmatic lesions. Catheter passage through a vascularized lesion may result in bleeding, which may obscure the lesion and prevent adequate sampling and cause additional hemorrhagic complications. “Overshoot” displacement was independently associated with VT <9.6 mL/kg IBW with a median difference of 7 mm between the expected and actual catheter tip-to-lesion distance. These findings suggest that when applying VT in the low range, especially in peripheral lesions, bronchoscopists should anticipate forward displacement of the catheter compared with the virtual view. Furthermore, bronchoscopists using ventilation parameters similar to those used during this study may wish to consider locking the catheter in a position that is at least 7 mm more proximal to the target than they would otherwise to account for a potential forward dislocation. This number is by no means absolute an would change based on the ventilation parameters implemented in each individual case.

Efforts to overcome CTBD are ongoing. Our group and others have demonstrated that the incorporation of intraoperative volumetric imaging, such as CBCT, into navigational bronchoscopy attenuates CTBD (5,21,26,27). Furthermore, advances in CBCT-guided ssRAB technology now allow the integration of the robotic system with mobile CBCT C-arms providing live updating of the virtual view based on intraoperatively-acquired CT images (7). The current study was performed before the incorporation of this integration into practice. The VESPA study indicated that certain mechanical ventilation parameters correlate with reduced atelectasis and potentially attenuate CTBD (28). In our study, lower VT was associated with “overshoot” displacement, while PEEP was not associated with displacement. It is noteworthy that our study was conducted before the implementation of a modified VESPA protocol in our institution and therefore PEEP values reported here were in the lower range of those implemented in VESPA (28).

Interestingly, displacement status did not affect our diagnostic yield rate in a statistically significant manner. This can potentially be explained by the higher number of CBCT spins in cases in which displacement was identified, suggesting the need for more refinement of catheter-lesion relationship to allow adequate sampling. It could be therefore postulated that without CBCT imaging diagnostic yield would have been lower among the displacement group and possibly more adverse events would have been observed. This finding further illustrates how intraoperative CBCT imaging allowed intraoperative refinement of catheter-tool-lesion relationship.

As a secondary outcome, this study demonstrated a fair interobserver agreement regarding the anticipated intraoperative radial EBUS view and significant variability in intraobserver agreement between anticipated and actual radial EBUS views. While this analysis may be biased by the individual experience of the observers included in this study, it may at the same time indicate that catheter access via a certain airway-lesion relationship plane anticipated by the preoperative CT is in many instances not feasible during the procedure. Hypothetically, this may result from an inability to access the target access airway due to anatomic constraints, extreme angulation, or limited visibility from secretions or blood. In those instances, the operator may resort to an alternative airway for navigation or employ a transparenchymal approach. Such alterations in the anticipated catheter-airway-lesion relationship may ultimately result in a different view than anticipated during deployment of the radial EBUS probe. This finding highlights a limitation of radial EBUS-based lesion localization that should be considered during navigational bronchoscopy. Accurate navigation is therefore ultimately based on the integration of data from multiple pre- and intra-operative sources, including thorough understanding of the pre-operative CT, visual bronchoscopic findings, radial EBUS view, and radiographic guidance from pulsed fluoroscopy, augmented fluoroscopy, or cross-sectional imaging modalities, as availability and expertise allow.

This study has several limitations. The retrospective, single center design and relatively small sample size increase the potential for selection bias and restrict the generalization of the findings. Displacement is a 3-dimensional phenomenon. In this study, we defined displacement by two dimensions—distance and forward/backward vector—but not by directionality in 3D space, i.e., cranial/caudal, lateral/medial, etc. For simplicity, we regarded any degree of distance misalignment as displacement; however, the intraoperative significance of distance displacement is a complex concept dictated also by the lesion size and catheter view angle. Finally, CBCT images were obtained during a peak inspiration breath hold with the APL valve set to 0–5 cm H₂O above the PEEP. Those parameters and the ventilation protocol implemented at the time of this study reflect our local practice and were not scientifically standardized. Procedures performed and CBCT images obtained under a different set of parameters may result in catheter-lesion relationships that are different from those described here.

Conclusions

A distance displacement between the virtual and actual target lesions occurred in 67% of ssRAB cases and is more likely to happen in peripheral lung lesions. “Overshoot” displacement is of particular importance due to the risk of disruption of pleural and vascular planes and is associated with lower VT. Future studies are needed to prospectively evaluate determinants of CTBD across multiple institutions and practices and among more diverse patient populations. Furthermore, a consistent, comprehensive, and graded definition of CTBD is needed to standardize definitions across institutions and scientific publications.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of the Memorial Sloan Kettering Cancer Center (#20–166). Informed consent was waived due to the retrospective nature of the study design.

Data Sharing Statement

Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1399/dss

jtd-17-10-9051-dss.pdf ^{(60.3KB, pdf)}

DOI: 10.21037/jtd-2025-1399