Diagnostic patterns after adoption of shape-sensing robotic-assisted bronchoscopy: a retrospective cohort study.

Introduction
In the US, there are nearly 235 000 new lung cancer diagnoses per year[1]. The lung cancer patient pathway is often complex, involving many different physician specialties and steps, including screening, biopsies, imaging, staging, treatments, and possible active surveillance periods required along the journey. A patient’s journey typically begins with the detection of a pulmonary nodule, either via lung cancer screening or an incidental finding, or after presentation with suspicious symptoms, like a persistent cough or hemoptysis. These scenarios can initiate a potentially complex journey and often require additional workup. While the next step of performing biopsy of a suspicious pulmonary nodule may confirm or rule-out a lung cancer diagnosis, unsuccessful sample collection can lead to delays in diagnoses.
On average, it takes two biopsy procedures of a lung nodule and/or staging to achieve a definitive diagnosis[2]. This may impact time to treatment, which may, in turn, impact cancer-related survival[3]. Delays in treatment that exceed 45 days have been associated with up to a 5–15% increase in mortality among early stage patients[4]. Approximately 40% of newly diagnosed lung cancers are stage IV, with an associated 5-year survival of 6%[5,6]. In contrast, the 5-year survival of stage I lung cancer ranges from 67 to 92%[5,6].
Improving diagnostic sensitivity has also been linked to positive economic outcomes. A cost-effectiveness analysis demonstrated that a 10% increase in sensitivity is linked to nearly $9000 in overall net monetary benefit to payors[7]. Similarly, diagnoses at later stages of disease are associated with greater overall healthcare costs. While per-patient per-month costs of stage I disease are approximately $7239, stage II and IIIa patients have been reported to incur an additional $2245 and $10 176, respectively[8]. These findings point to the potential for significant cost-effective improvements in patient outcomes if more lung cancer patients are diagnosed and treated in earlier stages.
Shape-sensing robotic-assisted bronchoscopy (ssRAB) is an advanced bronchoscopic procedure that utilizes robotic technology to navigate to the distal bronchi of the lungs. ssRAB can enhance the precision and control of the bronchoscope, allowing for better visualization and access to hard-to-reach areas within the lungs while preserving a strong safety profile.
Risk of pneumothoraces after ssRAB has been reported to range between 0 and 3.1%[9–11]. In contrast, a comprehensive systematic review and meta-analysis reported the overall pooled incidence for pneumothorax in Transthoracic Needle Aspiration (TTNA) procedures is 25.9%, with 6.9% of cases requiring chest tube insertion[10]. This data suggests that ssRAB is a safer alternative in terms of pneumothorax risk when compared to TTNA. Risk of pneumothorax with TTNA is higher in patients with chronic obstructive pulmonary disease and a lack of contact between the lesion and the pleura[12–14]. As such, ssRAB presents a viable alternative to TTNA, particularly amongst patients at high risk for pneumothorax.
This single-center study explores the impact of introducing ssRAB on the characteristics of lung cancer diagnosis. This cohort study has been reported in line with the STROCSS guidelines[15].

Methods

Participants
In a retrospective chart review cohort study, participants were seen at the interventional pulmonology department at a single center for diagnostic workup of a suspicious pulmonary nodule with an ssRAB from September 2020 – the month of ssRAB installation – through December 2023. Patients received an ssRAB biopsy from a single interventional pulmonologist following an abnormal lung imaging study. All patients were undergoing a diagnostic workup for a possible lung cancer diagnosis, and all patients evaluated during the study period were included. No patients were excluded. Patient medical records were reviewed for demographics, pathology findings, nodule size, complications and stage at diagnosis. The primary outcome was diagnostic stage. Secondary outcomes included histopathological findings, nodule size, sensitivity, and intraoperative complications. These research activities were approved by the Institutional Review Board (IRB #2020-010) and registered to the Open Science Framework (https://doi.org/10.17605/OSF.IO/JXGE9). A waiver of informed consent was granted for this retrospective chart review study under The Common Rule (45 CFR 46), given that the study represented minimal risk and could not be practicably carried out without the waiver.

Shape-sensing robotic-assisted bronchoscopy procedure
Before an ssRAB procedure, the patient is placed in a supine position and intubated, and the ssRAB system is docked. The pulmonologist uses pre-operative imaging to identify the target area and plan a bronchoscopic route to that location. During the ssRAB procedure, the pulmonologist navigates to the lesion location using the ssRAB platform. The location is further confirmed radiologically, both by radial endobronchial ultrasound (EBUS) and fixed cone beam CT. Tissue samples are collected using brush, needle, and forceps and sent for pathologic assessment, including rapid on-site evaluation. Cryobiopsy was not used, and bronchoscopic EBUS-Transbronchial needle aspiration (TBNA) lymph node staging was performed in all cases.
HIGHLIGHTS
Implementation of shape-sensing robotic-assisted bronchoscopy (ssRAB) was associated with an increased rate of early stage lung cancer diagnoses.

The diagnostic yield of ssRAB was high at 94%, with a sensitivity of 97% for malignant disease.

Over the study period, there was a significant increase in the odds of an early stage diagnosis by 5.2% per quarter.

The ssRAB platform enabled successful navigation and biopsy of progressively smaller nodules, with a trend towards decreasing nodule size over time.

The study reported a low complication rate, with only two pneumothorax events (0.5%) and no bleeding events, indicating a strong safety profile for ssRAB.

The integration of ssRAB into lung cancer diagnostic pathways may enhance early diagnosis and improve clinical outcomes, potentially reducing the need for repeat biopsies and associated healthcare costs.

Statistical analysis
Procedures were considered to be diagnostic if a definitive malignant or benign result was determined based on pathologic findings from the ssRAB biopsy, aligning with recent American Thoracic Society definitions[16]. Specific malignant diagnoses included findings like small cell or non-small cell lung cancer. Specific benign diagnoses included pathologic findings like granulomas or infections like Aspergillus. Nonspecific benign encounters were defined as indeterminant pathologic findings like atypia or inflammation (either acute, chronic, or mixed). Nondiagnostic encounters included ssRAB procedures that ended due to a failed navigation to the nodule or if other nondiagnostic results occurred. These criteria were based on the aforementioned definitions of diagnostic and nondiagnostic biopsies, last updated in 2024[16]. Diagnostic yield was calculated using the intermediate definition. The numerator of intermediate definition is the sum of patients with malignant and specific-benign diagnoses, plus those with nonspecific benign diagnoses who were followed with imaging or subsequent biopsy to confirm a definitive benign diagnosis. The denominator of this ratio is all patients undergoing biopsy. Sensitivity was calculated as the probability that patients with true malignant disease had a malignant diagnosis resulting from their ssRAB biopsy. True malignant disease status was determined by chart review following the ssRAB biopsy, through any subsequent diagnostic workups, until a final diagnosis was documented. All chart reviews were performed by the same individual, and no patients had missing data on diagnostic results due to the nature of this chart review study.
The quarterly proportion of early stage cancer diagnoses was calculated as a fraction of the stage I or II diagnoses among all malignant non-small cell lung cancer diagnoses. Limited-stage small-cell lung cancer diagnoses were bundled with early stage diagnoses for these analyses. Quarterly median nodule size was calculated using a patient’s smallest nodule size based on the largest dimension of that nodule from the biopsy encounter. Statistical analyses were performed in R (Vienna, Austria) using generalized linear modeling to test for changes in nodule size and cancer stage over time and Chi-squared tests to compare diagnostic yield by subgroup with alpha = 0.05.

Results
399 patients underwent an ssRAB procedure, and most patients were female, smokers, had obesity (e.g., BMI > 24.9[17]), and had no history of lung cancer or COPD (Table 1). Two patients underwent preplanned, nondiagnostic dye markings and were excluded from the diagnostic yield calculations. Median nodule size at biopsy was 16 mm (IQR: 12–28 mm). Most of the nodules biopsied were between 10 and 20 mm (n = 202, 52.7%), with the remainder being 20–95 mm (n = 134, 35.0%) and 0.3–10 mm (n = 47, 12.3%).

Sixty-two percent of attempted biopsies (n = 247) were diagnosed with a malignant lung tumor, 12% (n = 48) were diagnosed with a specific benign condition, and 24% (n = 95) had nonspecific benign diagnoses. Seven patients had procedures where the nodule was not reached, insufficient tissue was sampled, and the procedure was terminated. Among these 95 nonspecific benign patients, 78 had follow-up imaging and/or biopsies that confirmed a benign diagnosis. Specific benign diagnoses primarily included granulomatous inflammation, organizing pneumonia, and findings related to Aspergillus. A full list of diagnoses following the initial biopsy are available in Supplemental Digital Content Table S1, available at: http://links.lww.com/MS9/B7. Two patients experienced a pneumothorax (0.5%), but neither case required a chest tube or other intervention, and no bleeding events occurred.
Malignant lung cancer diagnoses were primarily stage I (60.3%, n = 144), followed by stage II (10.1%, n = 25), stage III (8.9%, n = 22), and stage IV (6.1%, n = 15). There were five small-cell lung cancer diagnoses, all of whom were diagnosed at the limited stage. Thirty (14.6%) had metastatic disease originating from outside the lungs, and six patients had missing stage data. Among the 95 patients with nonspecific benign diagnoses, 78 were confirmed to have a benign diagnosis upon follow-up imaging and/or biopsy, while 8 patients were subsequently diagnosed with lung cancer. Including both cancers diagnosed following the ssRAB biopsy and after follow-up, there were 247 true positives and 8 false negatives, for an ssRAB biopsy sensitivity of 97.0% (95% CI: 93.9–98.6%). Based on model results, quarter-over-quarter, there was a 5.2% increase in odds of an early stage diagnosis (P = 0.0485) (Fig. 1) and 0.8 mm decrease in nodule size (P = 0.09) (Fig. 2).

Diagnostic yield was 94.0% using the intermediate definition. Diagnostic yield was consistent across lung centrality, lobe, and nodule size. Diagnostic yield was 89.4, 94.2, and 95.0% across the inner, middle, and outer thirds of the lung, respectively (Table 2). Similarly, the diagnostic yield ranged from 89.5 to 96.9% across all five lobes of the lung. Among the smallest nodule size group (<10 mm), diagnostic yield was 100%, while the middle size group (10–20 mm) was 92.6% and the large size group (>20 mm) was 94.0%. After a malignant diagnosis, most patients received either radiation (n = 73) or surgery (n = 76), with 30 patients receiving chemotherapy (Table 3). For patients with nodules 8 mm and smaller (n = 33, 8.6%), ssRAB achieved a diagnostic yield of 100% with 54.5% (n = 18) having a malignant diagnosis.

Discussion
In a single-center, single-arm study, the interventional pulmonology program experienced an increase in the proportion of early stage diagnoses (+22.7% per year) in the same community and referral catchment area after adoption of an ssRAB platform. Additionally, the diagnostic yield for this novel bronchoscopic technology was high, despite nodule size decreasing over time (−0.8 mm per quarter), even when utilizing the ATS criteria for diagnostic and nondiagnostic cases. This demonstrated the enablement of navigation to and biopsy of progressively smaller nodules by the ssRAB platform. To our knowledge, this is the first publication to detect a change in the proportion of early stage cancer diagnoses within a single practice in the United States.
Changing referral patterns may have been linked to the decreasing size of nodules at diagnoses, though the results show that diagnostic yield was not different by nodule size. With the advent of ssRAB, those patients with smaller nodules who fulfill the criteria for biopsy are no longer waiting for nodule growth to justify a safe biopsy. As diagnostic yield remained high across time, referring providers sent patients for ssRAB that may have otherwise been placed in a watchful waiting protocol. While watchful waiting protocols are applicable and can be cost-effective for low-risk patients, those patients at higher risk for malignancy benefit from a more thorough diagnostic workup[18]. Diagnostic delays can negatively impact disease prognosis and performance status[4,19,20], and these delays can be further associated with negative psychosocial impacts on cancer patients[21,22].
In addition to our findings on stage and nodule size, no difference in yield was detected between other patient characteristics like lobe location or centrality of lesion. This contrasts with previous studies of other bronchoscopic methods, which have found diagnostic yields ranging from 40 to 60% for harder-to-reach peripheral lesions[23–25]. High diagnostic yield can help lead to the avoidance of repeat biopsies. A repeated biopsy can increase the overall cost of a diagnostic workup for an individual patient by 40–80%[26]. In the United States, almost half of patients undergoing diagnosis for lung cancer end up requiring at least two total biopsy encounters[26]. Both the ssRAB’s high diagnostic yield and enablement of concomitant biopsy of lymph nodes for staging can reduce the potential for repeat biopsies and the downstream financial impacts to patients and payors.
A small segment of the cohort was biopsied at nodule sizes of ≤8 mm. These patients would typically have entered a watchful waiting protocol after their imaging, per current Fleischner Society guidelines last updated in 2017[27], with subsequent CT scans recommended 6–12 months after the suspicious finding instead of a tissue biopsy. Over half of these biopsied patients were diagnosed with malignant tumors from their ssRAB procedure, but their disease may have remained undiagnosed had their tissue biopsy been postponed following current guidelines. This scenario highlights the potential for disease progression in the current watchful waiting guidelines, despite having bronchoscopic technology that can achieve high diagnostic yields in small nodules.
There were two weaknesses in this study. All cases came from a single provider at a single center, which may limit generalizability. However, no other providers utilized the ssRAB at this institution, so there would be no variation in care. This was also a retrospective, single-arm study with no comparison group. Further comparative studies are needed to evaluate the impact of bronchoscopic modality in the diagnosis of lung cancer.
In conclusion, this analysis found that the adoption of the ssRAB platform was associated with an increased proportion of patients diagnosed with early stage lung cancer, which may have been related to underlying referral patterns.