Feasibility and Safety of Percutaneous CT-Guided Bone Biopsies in Cancer Patients Using a Patient-Mounted Robotic System: A Retrospective Analysis of 40 Consecutive Biopsies

Abstract

This retrospective study evaluated the feasibility and safety of percutaneous computed tomography (CT)-guided bone biopsies in cancer patients using a patient-mounted robotic system with steering capabilities. The study included 39 patients (17 female, 22 male; median age 65.5 years, interquartile range: 54.8–71). Forty biopsies were performed in the pelvis, spine, ribs, shoulder, femur, and sternum. The technical success rate was 100%, and the median trajectory length was 55.9 mm (47.1–73.6). Intermediate checkpoints were used in eight biopsies. Median time from the first to final scan was 21 minutes (17–37). The overall procedure time was 30 minutes (23.5–36). The median dose length product and effective dose were 536.6 mGy.cm (396.2–837.7) and 7.1 mSv (4.7–10.8), respectively. No adverse events occurred. The diagnostic yield for cancer was 72.5%. Percutaneous robotic bone biopsies demonstrated high technical success, adequate diagnostic yield and safety profile.

Graphical Abstract

Introduction

Recent technological advancements have led to the development of robotic-assisted systems for percutaneous needle insertion (1–5). Navigation systems allow for real-time tracking of biopsy needles in 3D, and robotic navigation demonstrates enhanced accuracy while also reducing radiation exposure (6–8). However, limited evidence as yet exists for the feasibility and efficacy of robotic-assisted bone biopsies.

Accuracy, radiation dose, and procedure time of manual bone biopsy varies depending on several factors, such as location and operator experience (9,10). Accuracy rate can range from 60% to 90%, and some procedures have been reported to take up to an hour or more (11). Radiation exposure during the procedure can range from 0.5 mSv to 50 mSv, depending on the imaging technique and procedure duration (11).

This preliminary retrospective analysis aimed to assess the potential of robotic-guided bone biopsies that use a patient-mounted robotic system with steering capabilities by evaluating the technical success, diagnostic yield, radiation exposure, and potential complications in patients with cancer.

Materials and Methods

This retrospective observational cohort study received approval from the Institutional Review Board and was conducted in accordance with the principles of the Declaration of Helsinki. All participants provided written informed consent prior to undergoing percutaneous CT-guided bone biopsy. Measures were taken to ensure participant confidentiality and privacy, and the study was conducted in compliance with relevant regulations and guidelines that govern research involving human subjects.

Patients and lesions

This study included 40 consecutive bone biopsies performed between June 2022 and March 2023 in 39 outpatients, 17 of whom were female and 22 of whom were male (Table 1). The median patient age was 65.5 years (interquartile range [IQR]: 54.8–71). Median body mass index was 28.7 kg.m² (24.2–31.7). A total of eleven patients had a history of breast carcinoma; nine of prostate carcinoma; six of multiple myeloma; four of non-small cell lung cancer; two of sarcoma; two of squamous cell carcinoma; two of unknown primary at the time of biopsy; one of adenoid cystic carcinoma; one of ovarian carcinoma; one of bladder cancer; and one of non-Hodgkin lymphoma.

Table 1 :

Demographic Data and Results

Demographics
Number of patients	39
Gender	17 female / 22 male
Age (years)	65.5 (54.8–71)
Body Mass Index (kg.m 2 )	28.7 (24.2–31.7)
Size of lesion (mm)	26 (17–32)
Pattern	lytic (14/40; 35%) mixed (16/40; 40%) sclerotic (10/40; 25%)
Location	pelvis (n=19) spine (n=8) ribs (n=5) shoulder (n=5) femur (n=2) sternum (n=1)
Median trajectory length of needle insertion (mm)	55.9 (47.1–73.6)
Median time of needle insertion by the robot from skin to target (sec)	19 (15–31)
Median time from first to final scan (min)	21 (17–37)
Median procedure time (min)	30 (23.5–36)
Median Dose Length Product (mGy.cm)	536.6 (396.2–837.7)
Median effective dose (mSv)	7.1 (4.7–10.8)
Diagnostic yield for cancer	72.5% (29/40)

Note : Interquartile range is reported for each median value

Indications for bone biopsies were recommended to confirm the underlying condition of positron emission tomography (PET) uptake and guide appropriate treatment. Biopsies were performed in various sites, including the pelvis (n=19); spine (n=8); ribs (n=5); shoulder (n=5); femur (n=2); and sternum (n=1). The median lesion size was 26mm (17–32). The lesions biopsied were lytic (14/40; 35%); mixed (16/40; 40%); or sclerotic (10/40; 25%).

Techniques

Biopsies were all performed under sedation by a single interventional radiologist with 20 years’ experience. None were performed under general anesthesia. A patient-mounted robotic system that advanced and steered the needle on demand was used under CT guidance. The ACE^® robot (XACT Robotics, Caesarea, Israel) consists of several components and involves several steps to perform biopsies (2). The system works using conventional CT scanners only. Briefly, the patient was positioned on the CT scanner table and secured with a body strap to prevent movement during the procedure (Figure 1). The biopsy site was approximatively identified by the operator after reviewing previous CT or PET-CT imaging, and the skin was marked for the side. After scrubbing the skin, the robot was attached to the patient using a mounting bracket. The robot location was adjusted using the scout of the CT scan to reduce the navigation out-of-plan and to fit with the robot’s 6×6cm operative window. An initial CT scan, including the target and the robot, was obtained. CT images were transferred and automatically registered in the robot software within 30 seconds. The trajectory was planned using a computer interface connected to the robot, and by identifying the entry point and the target on two orthogonal views. This allowed for indifferent navigation both in-plan or out-of-plan of the CT scan. Local anesthesia with Lidocaine 2% was performed at the entry point. The 17G 20cm needle (Bard TruGuide, East Rutherford, New Jersey, USA), fixed on a plastic rack, was then attached to the robot arm.

Figure 1: Technical workflow for bone biopsies with a patient-mounted robotics system.

A- The system consists of a workstation connected to the CT scan and the robot. The robot is positioned, secured on the patient, and draped to remain sterile.

B- The trajectory is planned in-plan or out-of-plan by determining the entry point and the target. In the case shown, navigation was out-of-plan (arrow) along the neck of the femur.

C- The needle is fixed on a rack (arrow) that allows it to automatically advance into soft tissue until the tip reaches the periosteum.

D- The needle is exchanged over a K-wire (arrow) for an 11G needle, which is further introduced manually within the bone. The robotic arm served as holder. The 13G biopsy needle is introduced and further advanced as any bone biopsy, the core needle biopsy is performed, and the sample obtained.

Under operator supervision, the needle was moved mechanically toward the skin at the identified entry point where local anesthesia was obtained. It was then inserted into the patient’s body via the robot’s planned trajectory. Additional CT scans were performed on demand to further assess the needle location at each select checkpoint. The needle tip’s trajectory, automatically detected, can be adjusted on the computer interface as needed until the target is reached. The robot was able to steer the needle in-plan or out-of-plan when advancing it from one checkpoint to another by moving its robotic arm and the needle jointly. For mixed and sclerotic lesions requiring to drill bone, an exchange over a K-wire was needed. The K-wire was inserted through the 17G needle after the periosteum was reached. The robot served at that point only as a holder of the needles and did not further advance them. The 17G needle was then removed, and the 11G trocar needle threaded over the 20G K-wire (Kensington Bone Biopsy System, Merit Medical, South Jordan, Utah, USA). After removing the K-wire, the trocar needle was advanced to the target location, and the tissue sample was manually collected using a 13G core-needle biopsy system (Figure 2).

Figure 2: CT-guided biopsy of a spine lesion in a 75-year-old man with history of prostate cancer.

A- CT showing the lytic lesion of the right pedicle of L2 (arrow).

B- A CT scan is acquired to register the robot (arrow), and the planification software is used to plan the needle insertion from the entry point to the target in three dimensions. In this case, the trajectory was out of plan on this sagittal view. Artifacts originated from the robot.

C- The robot advances a 17G needle along the trajectory.

D- The needle is exchanged over a 20G K-wire for an 11G trocar. The bone biopsy is manually performed with the robot serving as holder. The pathology confirmed to be bone metastasis of prostate carcinoma.

A cytology was obtained on the first pass of cores on slides. The cores were collected in formalin and sent for surgical pathology both with and without decalcification. The bone biopsy device was then removed. After the procedure, the patient was scanned for complications. All biopsies were performed as outpatient and each patient followed after discharge by both the interventional radiology and the primary teams to document delayed complications.

Statistics

This study’s primary outcome measure was technical success, defined as the ability to insert the needle via the robot according to plan. Secondary outcome measures included diagnostic success, defined as the ability to obtain a diagnostic biopsy sample; diagnostic yield for cancer, defined as the proportion of biopsy samples that were found to be cancerous; procedure time, defined as the total time the patient spent on the CT table, including the planning/preparation time of the robot; biopsy time, defined as the time between the first CT scan for localization and the last CT scan performed after needle removal, including the uploading of imaging data to the robot software; radiation exposure by dose length product (DLP) measurement, or radiation dose for the CT scan in mGy.cm, and the effective dose; as well as adverse events. Safety was evaluated using the Society of Interventional Radiology Adverse Event Classification System (12).

Descriptive statistics were used to report the median age and body mass index of the patient population, as well as the location and size of the biopsy sites. The success rate of the biopsy procedure, time to reach the target site, length of the trajectory, and number of checkpoints were also summarized using descriptive statistics, such as medians and IQRs, to report the central tendency and data spread. DLP was also reported using descriptive statistics. The effective dose was calculated as follows: effective dose (mSv) = k * DLP * w, where k is a conversion factor based on the patient’s age and gender, and w is a weighting factor for the specific organs scanned during CT. The tissue weighting factor (w) for bone is 0.015 mSv.mGy⁻¹.cm⁻¹ according to the International Commission on Radiological Protection (ICRP) (13).

Results

A total of 40 consecutive percutaneous CT-guided bone biopsies were performed with a robot in 39 outpatients with cancer. Tissue samples were obtained in all cases, resulting in a technical success rate of 100%. The median trajectory length of needle insertion was 55.9mm (IQR: 47.1–73.6). Intermediary checkpoints allowing for needle steering and needle trajectory correction were utilized in eight biopsies. Median time of needle insertion by the robot from skin to target was 19 seconds (IQR: 15–31), while the median time from first to final scan was 21 minutes (IQR: 17–37). Median procedure time was 30 minutes (IQR: 23.5–36). Median DLP was 536.6 mGy.cm (IQR: 396.2–837.7) and median effective dose was 7.1 mSv (IQR: 4.7–10.8 mSv). No adverse events were reported during or after the procedures. No significant differences were observed between lytic and mixed/blastic lesions for procedure time (24.5 min, IQR: 20.2–33, vs 31 min, IQR: 24.7–39.2, p=0.06) and DLP (571.7 mGy.cm, IQR: 317–795, vs 514, IQR: 396.2–720.2, p=0.7).

The diagnostic yield for cancer was 72.5% (29/40). Among these 11 lesions negative for cancer but with uptake on PET imaging, eight lesions were sclerotic or mixed. Only three were lytic. Osteonecrosis was confirmed in one patient, and two fractures were identified on subsequent follow-up. The remaining eight pathology results concluded for inflammation without any sign of malignancy on follow-up imaging (median of 6 months, IQR: 5–11 months), leading to a diagnostic success rate of 100%.

Discussion

In addition to accurate imaging and navigation software, robot assistance allowed for control and stability during needle insertion, which may have resulted in the adequate accuracy and low rate of complication observed in the study (1–3). Compared to traditional manual bone biopsies that are often performed under CT guidance in-plan, the ability of the robot to consistently advance and steer the needle in 3D may enable standardized access to a lesion without requiring advanced skill. The navigation system provides real-time visualization of the needle’s path from entry point to target, indifferently in-plan or out-of-plan, to optimize trajectory visualization. This combination may reduce the risk of damage to surrounding tissue while increasing the diagnostic yield (14).

Robot use did not extend the length of the procedure compared to the literature on manual biopsies (9,10). In this study, the median procedure time was 30 minutes with a duration of 21 minutes between the first and last CT scan and no significant differences were observed among the imaging patterns of lesions. Moreover, the median effective dose was found to be just 7.1 mSv (IQR: 4.7–10.8 mSv), lower than the recommendations of the ICRP report, which is 20 mSv for a single CT scan of an adult (13,15). This remains within the range of those observed in manual procedures. Although it was reported that effective dose can vary based on several factors, including biopsy location, patient size, and imaging protocol (11), the effective dose did not vary widely among the patients despite navigation either in-plan or out-of-plan. Furthermore, in the overweight study population, body mass index did not pose technical challenges with the robot as has been reported for manual biopsies in which narrowed space in the CT gantry or increased soft tissue thickness can make it difficult to accurately visualize and target the bone lesion. In terms of radiation exposure, some advantages of robot assistance may include remote operation and elimination of the need for lead aprons. Robot assistance also has the potential to reduce physical strain and discomfort experienced by medical personnel.

This study is limited by its retrospective nature. No direct comparison has been performed with manual biopsies. The sample size of the study is relatively small, and further studies are necessary to confirm these findings, as well as to perform any subanalyses based on lesion (e.g., depth, size) or trajectory (e.g., in-plane vs. out-of-plane). Median follow-up time for imaging is short for the pathology results negative for cancer, since two years of stable follow-up suggests benignity.

In conclusion, percutaneous CT-guided bone biopsies performed by a patient-mounted robotic system with steering capabilities are a safe and effective approach for diagnosing bone metastases in cancer patients.

Acronyms:

CT

Computed Tomography

DLP

Dose Length Product

ICRP

International Commission on Radiological Protection

IQR

Interquartile Range

NPV

Negative Predictive Value

PET

Positron Emission Tomography