Automated breast ultrasound (ABUS) for intraoperative margin control on surgical specimens in breast conserving surgery

As breast-conserving surgery (BCS) has become the standard for treatment of early breast cancer, the need for new technologies to improve intraoperative margin assessment has become clear. Close or positive margins during BCS lead to additional surgeries, treatment delay, additional stress for patients and increasing healthcare cost. Automated three-dimensional breast ultrasound (ABUS) systems are meant to overcome the shortcomings of hand-held ultrasound (HHUS). In this study, we investigate the feasibility of ABUS to conduct ultrasound on surgical specimens in breast conserving therapy. In this monocentric, non-interventional study, specimens of 40 women were examined via ABUS. A construction with isotonic saline solution, gel pads and ABUS membranes was invented by our team to produce images of breast cancer specimens using ABUS. Evaluation of the ABUS images was carried out by two independent physicians trained on ABUS evaluation. ABUS was conducted on 40 specimens. 90% of the generated images were of high quality. Measured tumor sizes with ABUS were bigger than measured tumor size with HHUS (mean tumor size 22.9 vs. 18.1 mm, CI 2.38–7.35, p < 0.05). The mean difference between the ABUS tumor size and the pathological tumor size was 1.8 mm (CI − 0.84–4.53, p = 0.17). The mean difference between the HHUS tumor size and the pathological tumor size was 3.2 mm (CI − 5.35 to − 1.03, p = 0.005). ABUS seems to be a suitable method to conduct specimen ultrasound. Further studies are required to evaluate the accuracy of ABUS for intraoperative margin assessment and possible implementation in clinical work routine.


Introduction
In the past decades, advances in early diagnosis of breast cancer have been achieved through widespread screening mammography, enhanced imaging techniques and increased patients' awareness [1]. These improvements have led to increased use of breast-conserving surgery (BCS), followed by radiation therapy, which has become the standard of treatment for early breast cancer [2,3]. Basic surgical requirement of BCS is the complete excision of the tumor with adequate safety margin. The surgical margin is defined as negative if no malignant cells are observed at the surface of the resected invasive cancer specimen. The term "No tumor on ink" is used to indicate that the inked borders are free from any detectable tumor tissue during the histopathological examination [1,4]. For ductal carcinoma in situ (DCIS), a clear margin of 2 mm is recommended [5]. In case of positive margins, reexcision or sometimes mastectomy are needed to achieve 1 3 clear margins. This second surgical procedure can lead to treatment delay of adjuvant therapy, additional stress for patients, and increasing healthcare cost [6]. Hence, one of the major goals of breast surgeons is to achieve tumor-free resection margins during the first surgery [3,7]. However, there is currently no established global standard for realtime and accurate intraoperative margin management in BCS. The gold standard for assessing surgical margins in BCS is microscopic pathologic evaluation of the excised tissue following formalin fixation, paraffin embedding, and hematoxylin and eosin (H&E) staining [5,8]. This is a labour-intensive and time-consuming process that cannot be completed intraoperatively. Several intraoperative assessment techniques have been described. Among these are frozen section analysis, imprint cytology, specimen radiography or hand held ultrasound of the surgical specimen, each of which has been used with the aim to reduce positive margin and re-excision rates [7,9]. Each method has its strengths and weaknesses and unique limitations. Imaging procedures are only feasible for image-correlated masses. Low sensitivity and specificity associated with conventional image correlated methods remain problematic [10]. Nonpalpable, but image-correlated invasive breast lesions are commonly marked for surgical removal using a wire which is inserted via ultrasound imaging. To improve the accuracy of the initial surgery an ex vivo ultrasonography of the excised specimen is carried out to confirm that the resected specimen resembles the candidate lesion targeted preoperatively and to assess the margins [11]. Most ductal carcinoma in situ (DCIS) are detected by clustered calcification in mammography and there is a difficulty in visualizing DCIS by ultrasound. Therefore, a preoperatively mammographically guided needle insertion followed by a specimen radiography of the removed breast tissue is carried out for evaluation of mammographic calcifications [11]. The described techniques have demonstrated practicability and cost-efficiency in clinical practice. These techniques in intraoperative assessment may reduce the rate of positive margins but they have not completely solved the problem. While using intraoperative imaging methods there are still rates of up to 20% for positive margins requiring for additional surgery [10,12]. The high number of positive margins justifies the goal to improve the rate of obtaining adequate margins at the initial surgery and the search for new methods for intraoperative margin assessment. The "Invenia Automated Breast Ultrasound Screening" (ABUS, General Electric Company, GE Healthcare GmbH, Germany) for automatic ultrasound examination is indicated as an adjunct to mammography for breast cancer screening in asymptomatic women (approved by the FDA), for whom screening mammography findings are normal or benign but have a high density of breast tissue [13]. ABUS reconstructs 3D data sets of the entire breast volume from automatically acquired B-mode images. The images of the ABUS are taken by specially trained medical assistants. The data are stored on a separate workstation to be analyzed by a medical professional. The stored volume data allow comfortable and time-efficient evaluation at any time [13]. The aim of this study is to develop a method for intraoperative ultrasound using the automated 3D-breast ultrasound (ABUS) for specimen ultrasound and margin control. This study describes the feasibility of using the ABUS in intraoperative specimen ultrasound for intraoperative breast margin assessment.

Methods
This study is a monocentric, non-interventional feasibility study. 40 women with sonographically visible early-stage (cT1c-cT2, cN0-cN1) invasive breast cancer who were scheduled to undergo primarily BCS in our center were included. Diagnosis of invasive breast cancer was made by preoperative core biopsy. Specimens of patients who had received neoadjuvant chemotherapy, mastectomy, specimens of re-excision surgery, status post-radiation therapy of the affected breast or preoperatively diagnosed DCIS were excluded. We did not exclude patients with postoperatively diagnosed DCIS accompanying the preoperatively known invasive breast cancer. Therapy planning was according to national guidelines [14]. No changes on the standard surgery or therapy procedure were conducted. The tumor was marked preoperatively by sonographic fine needle insertion. After excision of the tumor the specimen was fixed on a KliniTray™ (KLINIKA-Medical GmbH, Germany). Intraoperative specimen ultrasound was performed using hand-held ultrasound (Voluson™ S8, GE Healthcare GmbH, Linear-Array-Sonde ML6-15, 4.0-15.0 MHz), so additional tissue could be excised if the tumor excision appeared incomplete. This procedure corresponds to the standard of care in our breast cancer center for sonographically detectable tumor masses that are primarily treated surgically. Axillary surgery was undertaken in the same session, according to national guidelines. A multidisciplinary team reviewed all cases preoperatively and postoperatively, and adjuvant radiotherapy or systemic therapy was administered according to institutional and national guidelines. The surgery was not interrupted or prolonged by the following additional specimen examination. After the hand-held specimen sonography, specimen sonography was carried out using ABUS. For this, the specimen stayed fixed on the KliniTray™ and was put in a corresponding plastic box. Pieces of base lead were placed on the exposed part of the KliniTray™ to weigh the tray down, without changing the fixation of the preparation. The plastic box was completely filled with isotonic saline solution. The use of this solution should prevent any possible influence on the cells. An ABUS membrane was then placed on the box and a plastic bag filled with ultrasound gel was placed on the ABUS membrane. This bag was additionally coated with ultrasound gel for lubrication to ease the movement of the transducer and the ABUS transducer was attached. The construction is shown in Fig. 1. The ABUS recording of the specimen was then carried out by one of the trained medical assistants according to the standard ABUS procedure. The transducer was placed with low compression. To avoid the risk of damage to the ABUS system, the ABUS transducer had no contact with the isotonic water solution in this arrangement (see Fig. 2). Evaluation of the created images was carried out by two independent physicians trained on ABUS evaluation. The first evaluation consisted of the quality of the images. The images were classified as "easy to evaluate" or "difficult to evaluate". High image contrast and quality which enable a clear differentiation of the tumor from the surrounding area were necessary to classify an image as "easy to evaluate". Further assessment was conducted as follows exclusively for the images which had been classified as "easy to evaluate": the size of the tumor was measured in three dimensions (horizontally, vertically and ventro-dorsally). The safety margins, i.e. the distance between the tumor and the edge of the specimen, were evaluated in six directions: ventrally, dorsally, caudally, cranially, laterally and medially. The findings were compared to the report of the HHUS and to the histopathological report, where the size of the tumor and the radial distances from the edge of the tumor to the edge of the specimen in six directions were described. Furthermore, the histopathological report included hormone status, HER2 status, proliferation index Ki 67, the type of breast cancer (such as lobular carcinoma and non-specific type) and the need for re-excision. To ensure correct correlation with the histopathologic examination of the margins and their directions, the orientation of the specimen was preserved on the KliniTray™ throughout every step of the procedures described above. ABUS measurements and HHUS measurements of the specimen were documented in millimeter in the specified dimensions. Image interpretations of ABUS were carried out without knowledge of the corresponding HHUS and without knowledge of pathological results.

Statistical analysis
The ability of ABUS and HHUS examination to predict tumor size and safety margins was assessed by comparing the safety margins and tumor size measurements as predicted by the imaging modalities with the histopathologic report. Microsoft Office Excel 365 (Microsoft Corporation, Redmond, WA) was used for data collection. Statistical analysis and graphical representation were performed using IBM SPSS Statistic 25 (IBM Corporation, Armonk, NY) and validated by the authors. Concordance between the tumor sizes measured with ABUS and HHUS was calculated with Excel 365. A paired t test was used for to assess the differences between the tumor sizes measured with HHUS and ABUS, respectively. These measured tumor sizes were also compared with final pathologic tumor size. Statistical significance was assumed for a P value of less than 0.05 for all tests.

Ethical and legal aspects
This study was carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki). The study was registered by the local ethical committee (reference No 21-2309-104). All members of the research team committed themselves to the confidentiality of the information provided as well as to data protection and are subject to medical confidentiality. The data are not accessible to anyone outside the research team.

Results
ABUS was conducted on 40 specimens. The median age of the women from whom the specimens were obtained was 59.8 years. Immunohistochemistry and FISH analysis (where required) revealed hormone receptor positive tumors without HER2 protein overexpression in 37 cases (92.5%). There were two tumors (5.0%) with HER2 overexpression and one (2.5%) triple negative tumor. The majority of the generated images were of high quality. 90% and 87.5% of the images, respectively, were classified as good for evaluation by the two independent examiners. In 4 (10%) and 5 (12.5%) cases, the quality of the ABUS images was significantly affected by artefacts and the images were considered as "difficult to evaluate" by the two different examiners. In these cases, the ABUS images of the specimen were not further evaluated. The most frequent problem was air pockets in the gel pad that caused artefacts and lead to a poor image quality. When the quality was classified as good, the tumor and safety margins were measured as described above. Figure 3 shows an example of a specimen ultrasound conducted by ABUS with the described method. Figure 4 also shows a specimen ultrasound conducted by ABUS but with positive resection margin to the ventral direction. An intraoperative re-excision in the ventral direction should be conducted in this case. Table 1 shows the average safety margins assessed with ABUS and HHUS, as well as final histopathological distances. Measured tumor sizes with ABUS were bigger than measured tumor size with HHUS (mean tumor size 22.9 vs. 18.1 mm, Confidence interval (CI) for the difference 2.38-7.35, p < 0.05). The radial distances from the edge of the tumor to the edge of the specimen measured by ABUS and HHUS were comparable (see Table 1). There was a notable difference between ABUS, HHUS and final pathological safety margin in the dorsal direction. The mean distance from the edge of tumor to the edge of the specimen in the dorsal direction was 4.7 mm (0-24.4 mm) measured with ABUS and 2.3 mm (0-7.5 mm) measured with HHUS. Histopathologically the distance from the edge of tumor to the edge of the specimen in the dorsal direction was 8.5 mm (0-30.0 mm). The measurements with ABUS were more precise in predicting the pathological distance from the edge of the tumor to the edge of the specimen in the cranial, dorsal, ventral, medial and lateral direction (see Table 1). In the caudal direction measurements with HHUS were more precise to predict the pathological distance from the edge of the tumor to the edge of the specimen. Intraoperative specimen HHUS could not prevent re-surgery in 9 (22.5%) cases because of histopathological positive margin. In 1 case specimen HHUS detected a positive margin which was not visible on ABUS. This positive margin was confirmed histopathologically. The measured tumor sizes with ABUS and with HHUS were each compared to final pathological tumor size. The measured tumor size by ABUS was comparable to the final pathological tumor size. The mean difference between the ABUS tumor size and the pathological tumor size was 1.8 mm (CI − 0.84-4.53, p = 0.17). The tumor size measurements with ABUS were numerically bigger than the final pathological tumor size, but there was no statistically significant difference (mean 23.3 mm vs. 21.4 mm). In contrast, there was a significant difference between the measured tumor size via HHUS and pathological tumor size. The mean difference between the HHUS tumor size and the pathological tumor size was 3.2 mm (CI − 5.35 to − 1.03, p = 0.005). The tumor size measurements with HHUS were smaller than the final pathological tumor size (mean 18.3 mm vs. 21.4 mm) ( Table 2).

Discussion
Although surgical margin analysis has been acknowledged as a critical component of BCS, the best assessment technique has yet to be determined [11,15]. The aim of the presented 1 3 method is to reduce the risk of re-excision. Avoidance of this further treatment potentially reduces adverse effects on cosmesis and health care costs as well as psychological distress. Since the Cobalt trial demonstrated the benefits of an ultrasonography-based procedure over palpation-based approach, this has become a commonly used method in BCS [16]. HHUS examination is easily available and cost effective. Nevertheless, high re-excision rates justify the search  for new methods to improve the rate of clear resection margins. Other methods used for intraoperative margin assessment in BCS include frozen section analysis and imprint cytology, both of which are traditional pathologic methods for real-time intraoperative margin assessment in BCS [17,18]. These methods have a very high diagnostic accuracy in terms of both sensitivity and selectivity and are currently recognized as the most accurate methods for lowering the rates of positive margins during BCS [9,19,20]. However, the limitations of routine intraoperative frozen section for margin status include time resource allocations, labour intensity, technical challenges, and cost considerations [21]. Another limiting factor for routine frozen section analysis as part of intraoperative margin assessment in BCS is the worldwide shortage of trained pathologists [11]. Considering all the available methods, the optimal cost-and time effective technique with high diagnostic accuracy to assess resection margins in BCS during surgery reoperation is yet to be determined. To our knowledge, this is the first attempt to evaluate the feasibility of ABUS and its potential role to assess intraoperative margins on breast tissue specimen. This study was a feasibility study which assesses the performance of the ABUS technique in the intraoperative monitoring of breast cancer specimen. The results show that it is possible to generate high-quality ABUS images of breast surgery. In this study, the images achieved with ABUS are promising in terms of optimizing intraoperative tumor imaging. This new method is of interest for intraoperative margin control because it presents various benefits compared to specimen ultrasound with HHUS. One advantage of the method is the lack of compression of the specimen. HHUS requires a certain compression of the specimen which may lead to over-or underestimation of the margins. In the ABUS construction developed by our team, the specimen lays in a saline water solution without compression. Therefore, it is unlikely to falsify the margins due to tumor compression. Another advantage is the three-dimensional depiction of the complete tumor. In HHUS, the examiner works with vertical and transversal section of the tumor and normally measures the tumor in one vertical, one transversal and one horizontal section. In specimen ultrasound conducted by ABUS, the whole tumor can be screened at the workstation in a three-dimensional reconstruction. The coronal view in ABUS provides an additional view of the tumor. This leads to a precise imaging. Irregular tumor extensions, which may be overlooked in HHUS, can be visualized. The sample preparation for ABUS is time-efficient, standardized and can be conducted by a physician assistant. A general advantage of ABUS is the reproducibility and standardization of the ultrasound examination. This advantage can also be useful in specimen ultrasound. Conducting an ABUS of the specimen allows a time-and location-independent evaluation of the acquired 3D volume by different physicians. The method offers the possibility to install a workstation directly in the operating theater. This enables the surgeon himself to see the images of the tumor intraoperatively and he may decide directly and more precisely whether a resection in a certain direction is necessary-provided the surgeon to be qualified for ultrasound. In contrast, performance of HHUS is highly dependent on the examiner's experience and often requires relocation of the specimen to obtain highquality images. Some problems concerning the quality of the acquired ABUS data arise from the gel pad which may cause air pockets which decrease the quality of the images. A further development of the method consists in using a water bag instead of the gel pad to ensure that there is no air at any level of the construction. Another limitation of our study is the small number of included patients. For this reason, we were not able to obtain statistically significant results. However, we did reach our main objective which was the development of a suitable method for using ABUS on breast specimens intraoperatively.
This study suggests that ABUS in specimen ultrasound could be a valuable tool for intraoperative margin control and could even provide more precise information on margins than HHUS. The development of a method to use ABUS in specimen ultrasound provides the opportunity to conduct further studies regarding the accuracy of ABUS in specimen ultrasound and to evaluate whether the generated images will be of advantage in perioperative sonography. We are planning to confirm this newly developed technique in a further prospective clinical study. Objective of this continuative study will be the assessment of specificity, sensitivity and accuracy of ABUS on breast cancer specimen and the comparison to HHUS in a larger number of patients. Major goal will remain the reduction of positive margins and reexcision rates to improve the quality of life for our patients.
Author contributions MEH: protocol development, data collection and management, data analysis, and manuscript writing. DH: data collection, data analysis, and manuscript editing. ET: data management and data analysis. AS: pathological analysis. MF-P: data collection. MH: data collection. BRR: data collection. OO: manuscript editing and supervision. SS: protocol development, manuscript editing, and supervision. The average distances in the respective direction as well as the minimum and maximum distances between tumor and margin/minimum and maximum tumor size in the examined collective are indicated