Intra-operative assessment of excised breast tumour marginsusingClearEdgeimaging device
J Michael Dixon1, OBE FRCS, Lorna Renshaw1, RGN, Oliver Young1, FRCS, Dhananjay Kulkarni1, FRCS, Talha Saleem1, FRCS, Moshe Sarfaty2, PhD, Ramaswamy Sreenivasan2, PhD, Catherine Kusnick2, MD, Jeremy Thomas1, FRC Path, Linda Williams3, PhD
Institutions:
1. Edinburgh Breast Unit, Western General Hospital, Edinburgh, Scotland
2. LS BioPath, Mountain View, California, United States
3. University of Edinburgh, Medical School
Address for correspondence:
Professor J Michael Dixon
Edinburgh Breast Unit
Western General Hospital
Edinburgh
EH4 2XU
Email:
Tel. +44131 537 2643
Abstract
Introduction
Breast conserving surgery (BCS) aims to remove a breast cancer completely andobtain clear margins. Complete excision is essential to reduce the risk of local recurrence. The ClearEdge™ (CE) imaging device examines margins of excised breast tissue intra-operatively. The aim of this study was to investigate the potential of the device in detecting margin involvement in patients having BCS.
Methods
In Phase-1 58 patients underwent BCS and had 334 margins assessed by the device. In Phase-2 the device was used in 63 patients having BCS and 335 margins were assessed. Patients with margins considered close or involved by the CE device were re-excised.
Results
The margin assessment accuracies in Phase-1 and Phase-2compared to permanent section pathology were very similar: sensitivity (84.3% and 87.3%), specificity (81.9% and 75.6%), positive predictive value (67.2% and 63.6%),and negative predictive value (92.2% and 92.4%). The false positive rate (18.1% and 24.4%) and false negative rate (15.7% and 12.7%) were low in both phases.In Phase-2 re-excision rate was 37%, but in the 54 where the CE device was used appropriately the re-excision rate was 17%. Had all surgeons interpreted all images appropriatelyand re-excised margins detected as abnormal by the device in Phase-2 then the re-excision rate would have been 7%.
Conclusion
This study shows that the CE device has potential to reduce re-excision after BCS and further randomized studies of its value are warranted.
Introduction
Breast conservation surgery (BCS) is the standard treatment of choice for the majority of women diagnosed with early stage breast cancer [1-3]. BCS involves the removal of malignant tissue with a surrounding margin of normal breast tissue. Having orientated the excised specimen, the surgeon sends it together with any additional margin shavings and following a thorough microscopic evaluation that takes several days, a report on the margin status is issued. The likelihood of local recurrence following BCS is linked to the status of the margins [4, 5]. Data from 2 meta-analyses show that clear margins of ≥1 mm reduce the risk of cancer recurrence [6, 7]. In the US about 250,000 patients undergo BCS annually. In BCS the range of re-excision to clear involved margins varies between 20-70% [8-13]. These second or even third additional surgical procedures add to the morbidity of BCS, adversely affect cosmesis [14], and add significant costs. There is a need for new technologiesthat detect involved margins during surgery, and allow the surgeon to remove all malignant tissue during one BCS procedure.
There are two possible approaches to Intra-Operative margin assessment. One option is to examine the surfaces of the resection cavity in vivo but this is extremely challenging for a variety of reasons. The second and more practical approach is to assess the margins of the resected specimen. Current intraoperative methods for margin assessment include gross evaluation of the tumor specimen [8], intraoperative ultrasonography, specimen slice radiography [15, 16], and pathologic evaluation either with touch preparation cytology or frozen section. The MarginProbe™ [17] uses radiofrequency reflection to locally assess margin involvement.
The ClearEdge™ (CE)is a handheld portable imaging device that differs from currently available devices in that it uses bio-impedance spectroscopy that is very sensitive to extracellular and intracellular variations of tissue dielectric properties. This new imaging modality enables surgeons to identify and localizeareas of abnormal tissue at a margin based on their dielectric properties. Tissue abnormalities detected by the device include DCIS, invasive cancer, and also includes abnormalities such as atypical ductal and lobular hyperplasia, lobular carcinoma in situ, and areas of increased cellularity associated with inflammation. The aim of the current study was to demonstrate the safety and accuracy of the CE device and determine its potential to reduce re-excision rates in patients undergoing BCS in a two phase clinical study.
Patients and Methods
TheCEimaging device used in this study is a battery operated, portable, hand-held imaging device equipped with a sterile single-patient head [Figure 1], and a docking base for battery charging. The device is docked in a charging cradle between operations. The single use head and drape wereprovided sterile and can be attached to the reusable handheld device in the sterile field of the operating room.A unique feature of the CE is that a baseline measurement is made on each patient’s normal breast tissue. Having baselined the device the surgeon uses the device to scan all margins of the excised tissue and any other cavity margin shavings. The CE head’s tissue penetration depth guard was set to a depth of 3mm in an effort to ensure that the margins of the excised tumor were free from any residual cancer to a depth of 3mm. A penetration depth of 3mm in fresh tissue typically corresponds to about 2 mm of fixed tissue as reported by the pathologist.Each scan produces a color-coded image on the device’s LCD display.[Figure 2]. Aftereachmargin of the excised tissue is scanned the surgeon can decide intraoperatively if re-excision of additional tissue from that margin is required. This study was performed in two phases.
Phase-1:This phase of the study was designed to validate the safety and accuracy of the CEimaging device when used ex-vivo to image the margins of excised wide local excision specimens. In Phase-158 patients were enrolled and the clinical and pathological features of these patients and their cancers are summarized in Table 1a[Table 1a].BCS was performed by nine different surgeons in the Edinburgh Breast Unit.
The excised specimen was first subjected to intraoperative X-ray imaging. The Faxitron® produces an image of the entire excised specimen to determine first the mammographic lesion has been excised and second does not extend to any imaged radial margin. Additional margin shavings were removed in 34 patients (59%) based on examination of the specimenX-Ray. Assessment of the margins based on an X-ray image was documented on a clinical research form. After surgery had been completed the CE device was used to image each margin of the excised tissue specimen. All marginswere scanned completely using the device generating multiple images per margin. Assessment of each margin based on CE images was performed and documented by the operating surgeon on a specific clinical research form.Any location identified as abnormal by the device was marked on the specimen using orange ink so the pathologist could assess this area in detail. A margin was deemed abnormal based on the CEimage when at least one of the images taken from that margin was assessed as abnormal. Six margins (anterior, posterior, superior, inferior, medial and lateral) were assessed in each patient. In Phase-1, as the margins were assessed by the CEdevice after the operation had been completed, no patient with a margin that was abnormal on CE imaging had re-excision based on the result.
The accuracy of both CE and X-Ray imagesassessments were compared to permanent section pathology. The potential ofCEto reduce re-operation rate was based on the surgeon’s assessment of the CEimage of each marginand correlating this with the need for re-excision based onpermanent section pathology. If the pathologist identified DCIS or invasive cancer within 1mm of the final radial marginthen re-excision at a second operative procedure was performed. It was assumedthat had the surgeon acted on the basis of an abnormal CEimage and re-excised the abnormal margin, that later re-excision may have been avoided. The potential reduction in the rate of re-operations based on CEimaging was compared to theactual re-operation rate.
Phase-2: In this phase, theCE device was used intraoperatively by each surgeon in an attempt to reduce the reoperation rate. In this phase surgeons used the CE device on excised and oriented specimens in real time and documented their assessment and actions on a clinical research form. Surgeons also used X-ray images and documented their assessment of the specimen radiographs on a separate from. Unlike in Phase-1,the surgeon was asked to interpret the CEimages and to perform re-excision of any margin in which the imaging was abnormal. 63 patients were enrolled in this study [Table 1b] and ten surgeons used the device and performed further margin excisions based on the results obtained with either CE or X-ray. Of 63 patients enrolled in the study in only 54 caseswas the CE device used as per protocol.
Complete excision was defined as achieving clear radial margins of both invasive cancer and DCIS by greater or equal to 1mm. The actual reduction in the reoperation rate by theCE device was calculated by first determining the number of patients in which further margins were excised based on the CE and this cleared a margin that would otherwise have necessitated re-excision. Images were reviewed blind by JMD. It was clear that in 9 patients the device was not used appropriately asthe images were incomplete; these patients were classified as protocol violations and were removed from the calculation of reoperation rate. It was also noted that on review insomeof these 9 patients abnormalities were present at one or more margin but the surgeon failed to re-excise. These nine patients were removed from calculations of re-operation rates given that the device was not used as per protocol. In a further group of 5 patients it was evident that the surgeon had not taken an appropriate baseline image.
Results
Table 2 summarizes the accuracy results of Phase-1(58 patients) and Phase-2(63 patients) using CE and X-ray imaging. In Phase-1and Phase-2,334 and 335 margins were measured with the CE device, and 351 and 338 margins were assessed by X-Ray, respectively. [Table 2a][Table 2b]The reason that not all margins were assessed by the device was that in Phase-1 some patients surgeons reported some of the margins were disrupted, or the margin was too small to maintain proper contact between the device and the breast tissue and in Phase-2 in nine patients imaging of the margins were incomplete. The 351 margins assessed by X-Ray included a number of re-excisions performed based on specimen radiography. The results demonstrate that CEhas a higher sensitivity in Phase-1 than specimen X-Ray.
Re-operation rates achieved using only X-Ray images are compared to those potentially achievable based on the CE images in the 2 phases of the study are shown in Table 3 [Table 3]. Potential re-operation ratesare provided as in Phase-1 the device was used after surgery was completed. 17 of 58 patients had re-operation for involved margins in Phase-1: a rate of 29%. Had the surgeonexcised margins that were identified as involved based on theCE image in Phase-1then only 6 out of 58 patients would potentially have required re-excision, a re-operation rate of 10%. The results obtained by the CE were deemed satisfactory to proceed to Phase-2 of the study.
Results of Phase-2 of thestudy are shown in in Table 3. The re-operation rate obtained with the CE device iscompared to that obtained by using X-Ray images. 63 patients’ specimens had specimen radiology and the reoperation rate based on this alone was 37%. The CE device was used correctly per protocol only in 54 patients and, 9 patients in this grouphada re-excision: a rate of 17%. In 5 of the 54 patients who had re-excisionthe surgeon did not baseline the CE properly on the patient’s specimen. This included 2 patients wherethe surgeon baselined the device on malignant tissue and 3 where the baseline was on fat tissue. Thiswas evident from review of the images stored on the device. When the baseline is set on a malignant or fibrous tissue,the images from all margins areuniformly deep green pixels with no other color pixels. When the baseline is set on a fat tissue then theimages of all margins show a majority of red pixels. This leaves 49 patients in whom the device was used as per protocoland baselined appropriately and in this group the re-excision rate was only 4/49 (8%). The CE device had no effect on margin assessment by the pathologist.
Discussion
The challenge of BCS is to remove the invasive cancer and or DCIS to clear margins while at the same time limiting the volume of excision to leave a satisfactory cosmetic result. The definition of what constitutes a clear margin has been the subject of 2 meta-analyses and a recent Consensus statement [7, 18, 19]. Studies have shown that close margins increase local recurrence rates (OR 1.74) and two meta-analyses [7, 19] have demonstrated that when looking at different thresholds for negative margins ≥1mm is as good as wider margins. The data on >0mm were insufficient in the second analysis as minimal data on this margin were included in the meta-analysis [7] The consensus panel decided to recommend no cancer on ink as a negative margin [18]. In the UK 1mm is the most widely used width for the definition of a clear margin as this is the margin width that is most consistent with the conclusion reached by the two meta-analyses. 1mm is the margin that was used in this study.
Historically, re‐excision rates have ranged from 31% to 46% [9, 20, 21] for DCIS alone and from 11% to 46% for invasive breast cancer with DCIS [22-26]. A variety of intraoperative technologies have been used to assess the surgical margin to allow targeted re-excision at the same operation, including frozen section analysis [27], touch preparation (imprint) cytology [28], near infrared (NIR) fluorescence optical imaging [29], x‐ray diffraction [30], high‐frequency ultrasound [31], micro‐CT [32], MRI of the specimen and the MarginProbe™ [33]. Optical breast imaging uses near‐infrared (NIR) light to assess optical properties of tissue and has a high reported sensitivity [34, 35]. Another promising approach is the use of high‐frequency (HF) ultrasound. [36, 37]. X‐ray diffraction (XRD) is based on alterations in tissues following invasion by cancer. A micro‐CT and mini MRI systems have also been developed to look at margins [30]. The MarginProbe™ measures electrical reflection properties of the breast tissue using radiofrequency (RF) spectroscopy. RF signals are delivered and acquired by the probe. It takes approximately 5 minutes to use the device on the excised tissue specimen. Results so far are promising but it requires further studies to demonstrate its cost-effectiveness. [17, 38-41]
TheCEimaging device was developed to examine the margins of excised breast tissue intra-operatively and differs from any currently available device in that it uses bio-impedance spectroscopy that is very sensitive to extracellular and intracellular variations of tissue dielectric properties. It has a coloureddisplay and each pixel in the image is coloured greens (fatty no cancer), yellows (fibrous no cancer) and reds (cancer or cellular abnormality close to margin). Green and yellow pixels are considered as characteristic of normal tissue and reds are abnormal tissue. It has shown in this study that it is safe and that surgeons can be trained in its use. In Phase-1 of the study nine surgeons used the device in 58 patients undergoing breast conserving surgery for invasive or in situ breast cancer. Each surgeon had training on only 1 or 2 cases prior to the study and all were able to use the device and get recordable images during the study. Based on surgeons’ assessments in Phase-1 of this study, we showed that the device has a high sensitivity for involved margins while at the same time having a low false positive rate. Based on these results, it was estimated the device had the potential to reduce the rate of re-excision below 10%. The results using theCE devicewere much better than those obtained with specimen radiography. This is not surprising as our recent study has shown that DCIS is frequently not visible on conventional imaging and that is the main reason for surgeons incompletely excising cancers [42]. As much of the DCIS is not visualized by mammography or ultrasound, it is unlikely that techniques that rely on X-rays or ultrasound including intraoperative ultrasound will decreasere-excision rates. That is why new techniques are required to increase the rates of complete excision.
A unique feature of the CE device is the capability to automatically adjust its baseline to the individual patient’s breast tissue, thus potentially enhancing its detection sensitivity for each individual patient. The patient’s specific baseline is generated by an initial reading taken on normal breast tissue away from the tumour site. Typically, a proper baseline will display a combination of green and yellow pixels at the initialimage. Failure to take an appropriate baseline is identified from consecutive images. If the images of the margins show only deep green pixels then most likely the device has been baselined on fibrotic or malignant tissue. If all the consecutive images on all margins show many red pixels then the device was baselined on fatty tissue. In such situationsthe surgeon needs to re-baseline the device on a different part of the specimen and repeat the measurement.