JONT Journal

Introduction

Worldwide, 2.3 million women were diagnosed with breast cancer in 2022, causing 670,000 deaths globally. Breast cancer is the most common cause of cancer in women in 157 countries, with only 0.5-1% of breast cancers occurring in men. High-income countries have dropped their age-standardised breast cancer mortality by 40% in the last 40 years, yet it remains the leading cause of cancerrelated deaths among women [1, 2].

Breast-conserving surgery (BCS) and adjuvant radiotherapy are the preferred surgical treatment for patients diagnosed with early-stage breast cancer [3-5]. BCS has a higher incidence of local recurrence compared to mastectomy; however, disease-specific survival rates are similar in both procedures. Negative surgical margins are essential to reduce local recurrence risk and disease-specific survival in BCS [6]. Further surgery due to positive margins varies according to institution, type of surgery undertaken and definition of a positive margin. Findings suggest that of the patients required to undergo re-excision, 65% have positive margins, with disease at the margins rather than close to the margins [7]. Centres across the UK show large variations in margin policy; however, one study’s mean national re-excision rates were 17.2%, with some units reporting rates up to 41% [7]. Only 4% of breast units follow the Society of Surgical Oncology and American Society for Radiation Oncology (SSO-ASTRO) guidelines. If these guidelines became standard practice, the UK re-excision rate would drop to 15.4% [7] . Some studies report margin re-excision as low as 4.3% amongst their cohorts when implementing SSO-ASTRO guidelines [8]. If the UK followed the 2015 ABS Consensus, re-excision rates would drop to 14.8% [7].

Further surgery can have far-reaching consequences, including psychological and social impacts, cosmetic outcomes, fat necrosis, mastectomy rates, theatre time and health economics [5, 7, 9].

Intra-operative specimen X-ray has become standard practice in the UK. These are obtained by sending specimens to the radiology department or by using purpose-built devices located within the operating rooms to capture intra-operative images. The Faxitron^® (Hologic^®, Arizona, USA) for example, provides standard 2 dimensional X-ray images of the specimen and or the Kubtec MOZART^® System (KUB Technologies^®, Stratford, Connecticut, USA), generates three-dimensional (3D images) using tomosynthesis, as well as a two-dimensional (2D images).

Common practice uses a conventional 2D imaging system, which may be inferior to 3D imaging. Previous studies have shown that assessment with 3D images can reduce the re-excision rate to 5% [10]. Our institution utilises the Kubtec MOZART^® System; however, most surgeons at our institution use the superimposed 2D image, not utilising the 3D specimen tomosynthesis function. At our institute, surgeons review the radiological specimen images; a radiologist is not involved in the assessment, and surgeons accordingly act on taking further tissue if required. Our breast unit has five consultant surgeons and eight senior registrars. All specimens had an anterior margin at the mastectomy plane and a posterior margin at the pectoral fascia. The primary outcome of this study was to evaluate if the use of the Kubtec MOZART^® System 3D specimen tomosynthesis function and radiological review would have reduced re-excision rates.

Methods

This study retrospectively reviewed specimen images that a consultant surgeon and senior registrar had intra-operatively examined and diagnosed to have radiological clear margins using the Kubtec MOZART^® System, and that final histopathology identified a positive margin. Two surgeons reviewed each image with negative findings at the initial review intra-operatively. As is our practice, most of the initial reviews were performed using only the 2D function. During the retrospective review, an additional senior surgeon and two senior consultant radiologists assessed the four margins using the 3D tomosynthesis function. Upon reviewing the images, all investigators were blinded to the histopathology results. The breast surgeon used a standard highdefinition computer, whilst the radiologists used high-resolution dedicated mammography reporting workstations. All reviewers used the 3D specimen tomosynthesis function and documented if the tumour was close or involving each radial margin and whether a cavity shave would have been recommended.

This study included all patients undergoing BCS who required a return to theatre (Figure 1) for further surgery from July 2021 to June 2023, including patients with primary in-situ or invasive disease and patients who underwent primary surgery or surgery following neo-adjuvant systemic therapy. Patients who did not have intra-operative specimen imaging or had a positive margin diagnosed intra-operatively and were acted upon at the time were excluded from the study. Patients whose specimens had suboptimal orientation were also excluded.

The Kappa coefficient was calculated to assess agreement between the surgeon and the radiologist interpretation of positive radial margins based on the 3D images. Sensitivity and specificity were calculated for both surgeon and radiologist assessment of margins as compared with the gold standard of histopathological assessment. Additionally, sensitivity and specificity were calculated for a ‘dual assessment’ of the images by both the surgeon and radiologist, where a margin was considered positive if either the surgeon or the radiologist interpreted it as positive. All statistical analyses were conducted in Stata (version 18.0, StataCorp LLC, College Station, TX).

Figure 1: Flowchart diagram of patient recruitment for the study.

Results

Fifty-four specimens from 54 patients with at least one positive radial margin were included in the analysis. The median age was 60 (IQR 55-70). Patient tumour characteristics are shown in Table 1, both invasive (44/54, 81.5%) and pure ductal carcinoma insitu (DCIS) (10/54,18.5%) patients were included. The individual tumour size and type and any associated DCIS are shown in Table 2.

Eighty-one radial margins out of 216 (37.5%) were positive on histopathology. Forty-four (44/81, 54.3%) were assessed accurately as having involved margins on imaging by the radiologists and 25 (30.9%) by the surgeon using the 3D tomosynthesis function, (Figure 2). Table 3 shows the sensitivity and specificity of the surgeon, radiologist and dual assessment as compared with the histopathological gold standard. The radiologist reported margins with better sensitivity (54.3% vs 30.9%). There was a fair agreement between surgeon and radiologist interpretation (Kappa agreement 78.3%, Kappa coefficient 0.26, p<0.0001).

The surgeon was more likely to correctly classify a margin on 3D imaging in invasive cancer specimens than in pure DCIS specimens (76.7% vs 62.5%, p=0.064). The radiologist was also more likely to correctly classify a margin in invasive cancer specimens than pure DCIS specimens (82.4% vs 70.0%, p=0.077).

For invasive cancers, neither the surgeon nor the radiologist’s proportion of correctly identified margins was dependent on the grade of tumour, presence of associated DCIS, or type of tumour. The radiologist correctly classified 100% of margins in mucinous tumours, compared to the surgeon correctly classifying 66.7% however, this only accounted for twelve margins in three specimens.

Both surgeon and radiologist were least likely to correctly classify a margin in dense breast tissue (American College of Radiology breast density classification (ACR) D) and mixed density breast tissue. Radiologists were most likely to correctly classify a margin in fatty tissue (ACR A). Surgeons were most likely to correctly classify a margin in scattered fibroglandular tissue (ACR B). However, the sample size was too small to demonstrate statistically significant differences.

Figure 2: Number of margins correctly and incorrectly classified using the 3D images by the surgeon and the radiologist (n=216).

The proportion of patients who could have avoided re-excision was calculated as the number of patients in whom all margins that were positive on histopathology were correctly identified as positive on 3D specimen tomosynthesis by (a) the surgeon’s retrospective interpretation and (b) the radiologist’s retrospective interpretation. Overall, 11 patients (20%) could have avoided re-excision if all 3D images using tomosynthesis slicing function were adequately assessed by the surgeon in real-time. Twenty-seven (50%) patients could have avoided returning to theatre if the images had been reported intra-operatively by a breast radiologist.

Table 1: Patient demographics and operative details.

Table 2:  Exhibiting patient’s tumour type and size.

Table 3: Sensitivity and specificity of radiologist and surgeon assessment of margins. For dual assessment, a margin is considered positive when either radiologist or surgeon assessed it as positive.

Discussion

This study shows that intra-operative use of the 3D tomosynthesis slicing function can reduce the positive margin rate by 20% during margin assessments. In terms of patient outcomes this study shows that 50% of patients could have avoided re-excision if a breast radiologist assessed 3D images during surgery for negative margins.  We found radiologists accurately assessed 54.3% of positive margins versus 30.9% by the surgeon using 3D specimen tomosynthesis. The 3D tomosynthesis specimen X-ray enables accurate interpretation of margins, as seen in Figure 3. UK guidelines mandate intra-operative specimen radiography for all impalpable lesions and that radiographs should be reported to or by the operating surgeon within 20 minutes. However, practices may vary across European countries where radiologist reporting may also be required [11].

Figure 3: (A) 2D specimen mammogram displaying excised tissue with the irregular lesion. The margins appear adequate, with excision margins displayed and a satisfactory amount of healthy tissue between the mass and the specimen’s edge. (B) A single 1mm tomosynthesis slice 5mm from the anterior surface  of the same specimen providing a view that is unobstructed by tissue above or below. It shows visible spicules extending laterally to the margins are subtle, yet a significant finding.