The adoption of low dose computed tomography (CT) for lung cancer screening has led to the identification of earlier stage lung cancers.1 In concert with increased health care utilization of chest imaging,2 a greater number of pulmonary nodules are now being identified. For nodules with intermediate to high risk for malignancy, expert consensus suggests the use of minimally invasive techniques such as, bronchoscopic biopsy or transthoracic needle biopsy (TTNB).3 TTNB is limited by its relatively high rate of pneumothorax.4 Electromagnetic navigational bronchoscopy (ENB) has been used to navigate to peripheral lung lesions and perform biopsies with lower rates of pneumothorax.5–7 However, with this technology, diagnostic yield continues to be inferior to TTNB.8,9 Potential limitations include the articulation and reach of the handheld video bronchoscopes used during ENB, as well as the inability of navigation probes to maintain a static position when tools are introduced and retracted from the working channel. Recently, robotic assisted bronchoscopy (RAB) (RAB is meant to be inclusive of shape sensing robotic assisted bronchoscopy AND electromagnetic robotic assisted bronchoscopy) has been developed to overcome some challenges encountered when using ENB. Shape sensing robotic bronchoscopes have the ability to articulate 180 degrees in 4 directions and have a longer length than traditional handheld bronchoscopes. This greater articulation along with more precise movements allow access into previously arduous airways.10 Despite these advances, high diagnostic yield during RAB continues to be elusive.11,12
CT to body divergence is thought to be a significant deterrent for successful navigation to target lesions in ENB and RAB.13 A myriad of factors contributes to CT to body divergence. Variations in lung volumes before versus during the procedure may be one of the most notable contributors. Lung volumes on the preprocedure chest CT obtained at full inspiration, vary drastically from intraprocedural mechanical ventilation providing volumes close to tidal breathing. Nodules have been demonstrated to vary in position by nearly 2 cm with these large differences in lung volumes.14 In some cases, the variation may be greater than the size of the lesion itself, leaving little margin for error and contributing to nondiagnostic biopsies. To overcome this divergence, many centers now supplement RAB or ENB with 3-dimentional (3D) imaging obtained during the procedure. One common modality is cone beam CT (CBCT) with augmented fluoroscopy. Studies with both ENB and RAB have demonstrated higher diagnostic yield when using CBCT for confirmation of tool-in-lesion.15,16 CBCT, however, can be expensive and may not be readily available at all institutions for use during pulmonary procedures.
O-arm CT is an alternative imaging modality which may be more accessible and affordable for pulmonary procedures at some centers. The O-arm CT is a mobile machine which functions much like a conventional, stationary CT scanner found in most hospitals. The O-arm CT has a gantry which can be opened into the shape of a “C”. This gantry shape is used when the machine is being transported or positioned around a patient. Once in position, the gantry is closed to encircle the patient, forming the “O” shape. Further adjustments can be made by moving the whole unit toward the head or foot of the bed. Rotating x-rays performed around the gantry then allow for reconstruction of images in the axial, sagittal, and coronal planes Traditionally, O-arm CT has been used in spinal, cranial, or orthopedic procedures, and is rarely seen in literature related to pulmonary nodules. However, it has been described as a useful tool for intraoperative location of nonpalpable nodules during video-assisted thoracoscopic surgery.17 In addition, O-arm CT has recently been described as a technically feasible option for use during ENB.18
The use of O-arm CT during RAB has not previously been described. Here we present a retrospective review of our experience using O-arm CT during RAB for secondary confirmation of successful navigation to lung lesions.
All consecutive procedures (N=79) in 75 patients from the initiation of our tertiary care center’s robotic bronchoscopy program in September 2020, to July 2021. Patients were selected for RAB at the discretion of the interventional pulmonologist for the workup of pulmonary nodules, masses, or parenchymal abnormalities felt to be inaccessible by conventional, flexible video bronchoscope. All patients provided informed consent before the procedure. The retrospective review was approved by the hospital Institutional Review Board.
All cases were performed by a single interventional pulmonologist. Cases were done in the operating room (OR) under general anesthesia. Patients were intubated with an endotracheal tube at least 8 mm in diameter. The ventilator was set to deliver tidal volumes of 10 mL per kg of ideal body weight, and a positive end expiratory pressure of 10 cmH2O. After induction of anesthesia, the O-arm CT (Medtronic, Minneapolis, MN) was brought into position around the patient and OR table (SurgiGraphic® 6000, Steris, Mentor, OH) with verification of positioning achieved with a 2-dimensional fluoroscopic image. All staff in the OR followed standard precautions for radiation safety. Staff not wearing protective lead were given the opportunity to exit the room or position themselves behind radiation barriers before imaging. Once position of the O-arm was deemed satisfactory, an airway inspection with an ultraslim flexible video bronchoscope (BF-XP190, Olympus America Center Valley, PA) was performed. After airway inspection, the preplanned navigation route (Ion Planpoint Software, Intuitive Surgical, Sunnyvale, CA) was uploaded into the robotic bronchoscope control tower (Ion endoluminal robotic system) and navigation to the lesion was performed. After initial navigation, an aspirating needle was deployed (Ion Flexision Needle), then using the O-arm, a CT (Technical specifications previously described17), was performed while giving the patient a breath-hold at 20 cmH2O with the mechanical ventilator. Each breath-hold lasted ~30 seconds. After reviewing the CT in all 3 planes (axial, sagittal, and coronal), positional adjustments to the needle and subsequent O-arm CT runs were completed at the discretion of the interventional pulmonologist. Tissue acquisition was then carried out by any combination of transbronchial needle aspiration (TBNA) with 19 or 21-gauge needles, 1.9 mm forceps biopsy (Olympus EndoJaw), brushings (Olympus BC-202D-2020) and lavage with 20 mL of saline. Rapid onsite evaluation with a cytotechnician was available for all cases. All samples were sent for histopathologic evaluation. Additional samples for studies such as microbiology or flow cytometry were collected on a case-by-case basis.
After conclusion of tissue acquisition, the robotic bronchoscope was removed from the airway and the video bronchoscope reintroduced for final airway inspection, with suctioning of blood and secretions. In cases where Rapid onsite evaluation reported a preliminary diagnosis of cancer, a linear endobronchial ultrasound (EBUS) (Olympus BF-UC180F) was used for evaluation and staging of the mediastinum and hila before final airway inspection. All patients were recovered in the postanesthesia care unit.
Procedure time is defined as the time of first scope insertion for initial airway inspection before performing navigation with the robotic bronchoscope, until time of final video bronchoscope removal from the airway after RAB or EBUS. The procedure was considered diagnostic if the results yielded a specific malignant or benign process. Lymph node TBNA results were not included in the calculation of diagnostic yield.
Statistical analysis was performed with Microsoft Excel for Microsoft 365 (Microsoft, Redmond, WA). Median, interquartile range (IQR), and percentage are reported.
A total of 79 lesions in 75 individual patients are included in our analysis. Median age was 65 years old with an IQR of 57 to 74 years. Forty-eight percent of patients were female, and 56% were smokers. Median lesion size was 2.0 cm (IQR, 1.3 to 3.5 cm). Three cases involved parenchymal abnormality such as ground glass opacities or consolidation for which there was no discrete lesion to measure (Table 1). Definitive diagnosis was made in 61 of 79 cases (77%) (Fig. 1). Fifty-four percent of diagnostic cases involved malignancy (Table 2). Bronchus sign was present in 56% of cases. Overall median procedure time was 80 minutes (IQR, 59 to 103 min). Linear EBUS evaluation and sampling with lymph node TBNA was performed after RAB in 39% (31 of 79) cases. Cases involving linear EBUS had longer median times of 101 minutes (IQR, 72 to 127 min), whereas the 48 cases which did not involve linear EBUS had shorter median times of 70 minutes (IQR, 55 to 85 min). The shortest procedure was 32 minutes. The longest procedure was 157 minutes.
TABLE 1 -
Baseline Characteristics of 75 Patients and 79 Pulmonary Lesions Reported as Median (IQR) or N (%)
|Body mass index (kg/m2)
| American Indian
| Pacific Islander
| Current or former
| 0-2 cm
| >2.0-3.0 cm
| >3.0 cm
| Parenchymal opacity
IQR indicates interquartile range; LLL, left lower lobe; LUL, left upper lobe; RLL, Right lower lobe; RML, Right middle lobe; RUL, Right upper lobe.
TABLE 2 -
Diagnoses of Positive Cases
|Adenocarcinoma of lung origin
|Squamous cell carcinoma of lung
|Poorly differentiated carcinoma
*Extrathoracic metastases included: renal cell carcinoma (n=2), head and neck adenocarcinoma, extracellular osteosarcoma, squamous cell carcinoma of the uterus, esophageal adenocarcinoma.
Table does not include 7 “suspicious” cases as noted on initial pathology, see Figure 2
IMT indicates inflammatory myofibroblastic tumor.
Median number of O-arm CT runs was 2 (IQR, 1 to 4 runs). The Median effective dose of radiation was 7.2 millisieverts (mSv (IQR, 3.9 to 14.3 mSv). Tool-in-lesion was confirmed by O-arm CT in 97% of cases (Table 3). Of the 2 cases with unconfirmed tool-in-lesion, one resulted in a positive diagnosis of metastatic renal cell carcinoma, and the other was nondiagnostic.
TABLE 3 -
Procedural Outcomes of the 79 Biopsies Reported as N (%) or Median (IQR)
|Positive diagnosis when including suspicious lesions*
|Linear EBUS performed after RAB
|Procedure time (min)
|O-arm CT runs overall
|O-arm CT runs final 20 cases
|Effective radiation dose (mSv)
|Effective radiation dose final 15 cases
*Lesions described as suspicious for malignancy on pathology report.
CT indicates computed tomography; EBUS, endobronchial ultrasound; IQR, interquartile range; mSV, millisieverts; RAB, robotic assisted bronchoscopy.
Major complications included 2 cases of pneumothorax (2.5%), 1 of which required tube thoracostomy. Both patients were discharged after 1 night in the hospital. There were no complications due to respiratory failure, or bleeding.
Two patients experienced brief intraprocedure cardiac arrest. The first had ventricular fibrillation arrest. Cardiopulmonary resuscitation was administered for 2 minutes with 1 shock for defibrillation before achieving return of spontaneous circulation (ROSC). This patient was eventually diagnosed with disseminated tuberculosis and was discharged on hospital day 14. The second patient experienced a 90 second arrest with pulseless electrical activity. ROSC was achieved after administering epinephrine. This patient was extubated after the procedure, monitored in the intensive care unit (ICU) afterward, and discharged the following day. Neither patient suffered any permanent sequelae from their arrest. One additional patient experienced postprocedural hypotension requiring overnight ICU admission for observation without administration of vasoactive agents and was discharged home the following morning. There were no other periprocedural complications.
The utility of O-arm CT has previously been demonstrated with use during ENB.18 For the first time, we have shown the feasibility of O-arm CT use during RAB. We found the O-arm CT to be convenient and consider the images highly valuable, resulting in the unit being made an integral part of our RAB program for the past year. During this time, we have been able to confirm tool-in-lesion in 97% of cases. Of our 79 cases, we had no episodes of bleeding and 2 instances of pneumothorax, with only 1 requiring intervention. This is slightly lower than previously described rates of pneumothorax during RAB.10–12,16
Despite our excellent rate of tool-in-lesion confirmation in 97% of cases, diagnostic yield was significantly lower at 77%. This discrepancy between tool-in-lesion and diagnostic yield has been reported previously.18–20 One potential explanation for this observation is inter-reader variability among pathologists, with some requiring more tissue than others before making a definitive diagnosis. In our series, we had 7 cases in which the final pathology report included terms such as “suspicious for malignancy” or “cannot rule out malignancy” because there was not enough tissue for a cell block. All 7 of these cases were eventually proven to be malignant after the lesions were surgically resected. Had these cases initially been interpreted as malignant, our diagnostic yield would improve to 86% (Fig. 2). This still leaves a nearly 10% discordance between tool-in-lesion confirmation and diagnostic yield. The remaining difference could be due to limitations in our ability to utilize technologies such as augmented fluoroscopy or other 3D imaging software to allow for real time tool adjustments, or the limitations in current biopsy tools including the lack of a Francine needle often used by interventional radiologists when performing TTNB.
Our RAB and O-arm CT protocol does come with limitations. First is the retrospective nature of the study. As such, there is no set inclusion or exclusion criteria. All cases were performed at the discretion of the interventional pulmonologist, and subject to selection bias.
Next, the O-arm CT carries a somewhat large spatial imprint, which may make its use technically difficult at some institutions. As part of our RAB protocol, we perform all cases in an OR suite which is much larger than a typical endoscopy room. We have found the OR size adequate to accommodate the patient, robotic and video bronchoscopy equipment, O-arm CT, and all staff members including, but not limited to, interventional pulmonologist, fellow, anesthesiologist, OR nurse, scrub technician, fluoroscopy technician, and cytopathology technician. Other than the occasional equipment maneuvering or repositioning, we have not found the size of the O-arm to be a barrier to its use with our room configuration (Fig. 3).
Another limitation is the static nature of obtaining and reviewing intraprocedural imaging. Once the needle is deployed, there is a delay before visualizing the image and confirming tool-in-lesion. This time includes allowing staff to leave the room in preparation for the CT run, the run itself, ~30 to 60 seconds of machine processing to recreate the 3D images, followed by the actual review of the images in the axial, coronal, and sagittal planes. If needle repositioning and subsequent CT runs are needed, the process is repeated, in some cases several times. This process can lead to longer procedure times as compared with CBCT and augmented fluoroscopy. Our median procedure time was 80 minutes. Previous studies with RAB and CBCT showed a lower mean procedure time of 64 minutes.16 We do note that our prolonged times may also be a function of how procedure time is defined. We defined procedure time as the time of first scope in, until scope removal after final airway inspection. In 31 of our cases (39%), this time also included performing linear EBUS with lymph node TBNA, which adds a significant amount of time to the procedure. Previous groups have defined procedure time as time of robotic bronchoscope insertion to robotic bronchoscope removal.11 This definition will significantly reduce perceived procedure time as it does not include the EBUS-TBNA procedure, or initial and final airway inspections. This variation highlights the need for a consistent definition of procedure time when discussing RAB.
Our use of the O-arm CT did produce a higher effective radiation dose when compared with CBCT. Previous groups have shown an average effective radiation dose of 3.0 mSv per case when using CBCT to assist ENB.15 This lower radiation exposure is partly due to the inherent differences in the machines themselves21 but may also highlight a functional advantage of CBCT over O-arm CT as previously discussed. The average number of CBCT runs per case in the previous study was only 1.5 per case, versus our average of 2.9. However, with continued use of the O-arm CT, our number of runs per case are trending down. Over the course of our first 59 cases, the median number of CT runs was 3 per case. In the final 20 cases, the median was 1 run per case. Similarly, the effective radiation dose can be refined. The O-arm CT has 3 settings: high, standard, and low dose modes. Each mode has a different effective radiation dose of 5.3, 3.5, and 3.3 mSv per CT respectively. Initially, our protocol did not explicitly state which mode should be used. During our preliminary data collection, we recognized the high and standard modes were being used most often. In the wake of this discovery, we clarified our protocol to set the low dose mode as the default for all cases. We have not found the low dose mode to prevent visualization of the target lesion or hinder confirmation of tool-in-lesion (Fig. 4). A subgroup analysis of the final 15 cases revealed low dose mode was utilized in all 15 cases, resulting in a median effective radiation dose of only 5.3 mSv per case. This is a significant reduction compared the median of 10.7 mSv from our first 64 cases and overall median of 7.2 mSv for the 79 total cases in this analysis. We believe ongoing, diligent utilization of the low dose mode will continue to narrow the radiation discrepancy between O-arm CT and CBCT.
Additional limitations and concerns which apply to our cases are the two instances of cardiac arrest. Although both arrests lasted ˂2 minutes and the patients did well afterward, these occurrences cannot be ignored. Patient 1 had a 90 second ventricular fibrillation arrest. The procedure was being done 9 days into an existing hospitalization. The patient was in her early 60s with limited pre-existing conditions, however, was diagnosed with a severe infectious disease (disseminated tuberculosis) after the procedure. She also had some minor electrolyte abnormalities on the day of the procedure, which may have contributed. Patient 2 had a 90 second PEA arrest which was preceded by hypotension and bradycardia. Compared with patient 1, this patient was more medically complex including increased age of 83, and multiple pre-existing cardiac conditions (atrial fibrillation, mitral regurgitation, heart failure with preserved ejection fraction). After ROSC, she was fully responsive and able to be extubated. She was observed overnight in the ICU. Neither case had any procedural complications, such as bleeding or pneumothorax, to explain the cause of the arrest. Inspiratory breath-hold performed although obtaining intraprocedural 3-D imaging is a theoretical culprit for predisposition to cardiac arrest as this increases intrathoracic pressure and temporarily decreases venous return to the heart. Patient 1 had 4 O-arm CT spins before the arrest, resulting in four inspiratory breath holds. This number is higher than the mean number of O-arm CT spins and breath holds found in the analysis; however, there were several patients who had many more breath holds and did not experience any issues. Patient 2 had only 1 breath-hold performed. The timing of cardiac arrest did not seem to correlate with the timing of the breath-hold in either case. We suspect the arrests were unique and related to independent factors such as electrolyte imbalance in patient 1, and chronic cardiac conditions in patient 2. It is worth mentioning that both patients had prolonged procedure and OR times. Patient 1 had procedure and OR times of 160 and 284 minutes respectively whereas patient 2 had times of 114 and 155 minutes. Noting that some of the prolonged OR time was a direct result of the arrest and subsequent coordination of ICU care, it is possible that increased time exposed to anesthetic medications contributed in part to the cardiac arrests. More information from future studies is required to determine the incidence of hemodynamic instability and cardiac arrest during these procedures and caution should be exercised when selecting patients to undergo RAB.
Finally, the cost of performing RAB with O-arm CT may be a limitation. although a cost analysis was not part of our retrospective review, it should be taken into consideration before considering this procedure and may be a topic for future study.
This is the first study evaluating the feasibility, and long-term use of an O-arm CT with RAB. We achieved tool-in-lesion confirmation in 97% of cases with a diagnostic yield of 77% to 86% and only 2 occurrences of pneumothorax. Although use of CBCT is currently more common for use with RAB, O-arm CT is a satisfactory substitute. Larger, prospective studies are needed for a direct comparison between cone beam and O-arm CT to better understand their impact on procedural characteristics and diagnostic yield during RAB.
1. National Lung Screening Trial Research Team. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med. 2011;5:395–409.
2. Gould M, Tang T, Liu I, et al. Recent trends in the identification of incidental pulmonary nodules. Am J Respir Crit Care Med. 2015;10:1208–1214.
3. Gould M, Donington J, Lynch W, et al. Evaluation of individuals with pulmonary nodules: when is it lung cancer? Diagnosis and management of lung cancer: American College of Chest Physicians evidence-based clinical practice guidelines. Chest. 2013;5:e93S–e120S.
4. Wiener R, Schwartz L, Woloshin S, et al. Population-based risk for complications after transthoracic needle lung biopsy of a pulmonary nodule: an analysis of discharge records. Ann Intern Med. 2011;3:137–144.
5. Gex G, Pralong J, Combescure C, et al. Diagnostic yield and safety of electromagnetic navigation bronchoscopy for lung nodules: a systematic review and meta-analysis. Respiration. 2014;2:165–176.
6. Folch E, Labarca G, Ospina-Delgado D, et al. Sensitivity and safety of electromagnetic navigation bronchoscopy for lung cancer diagnosis: systematic review and meta-analysis. Chest. 2020;4:1753–1769.
7. Khandhar S, Bowling M, Flandes J, et al. Electromagnetic navigation bronchoscopy to access lung lesions in 1,000 subjects: first results of the prospective, multicenter NAVIGATE study. BMC Pulm Med. 2017;1:1–9.
8. Folch E, Pritchett M, Nead M, et al. Electromagnetic navigation bronchoscopy for peripheral pulmonary lesions: one-year results of the prospective, multicenter NAVIGATE study. J Thorac Oncol. 2019;3:445–458.
9. DiBardino D, Yarmus L, Semaan R. Transthoracic needle biopsy of the lung. J Thorac Dis. 2015;7(suppl 4):S304–S316.
10. Reisenauer J, Simoff M, Pritchett M, et al. Ion: technology and techniques for shape-sensing robotic-assisted bronchoscopy. Ann Thorac Surg. 2022;1:308–315.
11. Chen A, Pastis N Jr, Mahajan A, et al. Robotic bronchoscopy for peripheral pulmonary lesions: a multicenter pilot and feasibility study (BENEFIT). Chest. 2021;2:845–852.
12. Chaddha U, Kovacs S, Manley C, et al. Robot-assisted bronchoscopy for pulmonary lesion diagnosis: results from the initial multicenter experience. BMC Pulm Med. 2019;1:1–7.
13. Pritchett M, Bhadra K, Calcutt M, et al. Virtual or reality: divergence between preprocedural computed tomography scans and lung anatomy during guided bronchoscopy. J Thorac Dis. 2020;4:1595–1611.
14. Chen A, Pastis N, Furukawa B, et al. The effect of respiratory motion on pulmonary nodule location during electromagnetic navigation bronchoscopy. Chest. 2015;5:1275–1281.
15. Pritchett M, Schampaert S, de Groot J, et al. Cone-beam CT with augmented fluoroscopy combined with electromagnetic navigation bronchoscopy for biopsy of pulmonary nodules. J Bronchology Interv Pulmonol. 2018;4:274–282.
16. Benn B, Romero A, Lum M, et al. Robotic-assisted navigation bronchoscopy as a paradigm shift in peripheral lung access. Lung. 2021;2:177–186.
17. Ohtaka K, Takahashi Y, Kaga K, et al. Video-assisted thoracoscopic surgery using mobile computed tomography: new method for locating of small lung nodules. J Thorac Cardiovasc Surg. 2014;1:1–7.
18. Cho R, Senitko M, Wong J, et al. Feasibility of using the O-arm
imaging system during ENB-rEBUS–guided peripheral lung biopsy: a dual-center experience. J Bronchology Interv Pulmonol. 2021;4:248–254.
19. Verhoeven R, Fütterer J, Hoefsloot W, et al. Cone-beam CT image guidance with and without electromagnetic navigation bronchoscopy for biopsy of peripheral pulmonary lesions. J Bronchology Interv Pulmonol. 2021;1:60–69.
20. Kalchiem-Dekel O, Connolly J, Lin I, et al. Shape-sensing robotic-assisted bronchoscopy in the diagnosis of pulmonary parenchymal lesions. Chest. 2022;161:572–582.
21. Nachabe R, Strauss K, Schueler B, et al. Radiation dose and image quality comparison during spine surgery with two different, intraoperative 3D imaging navigation systems. J Appl Clin Med Phys. 2019;2:136–14.