A Comparison of Residual Tumor Accuracy Prior to Re-Excision between Breast Magnetic Resonance Imaging and Ultrasonography
Article information
Abstract
Purpose:
We aimed to evaluate the accuracy of breast magnetic resonance imaging (MRI) versus ultrasonography (US) in detecting residual tumor prior to a second re-excision.
Methods:
Fifty-seven breast cancer patients diagnosed by vacuum-assisted breast biopsy (VABB) or excisional biopsy and scheduled for breast-conserving surgery (BCS) were included. They underwent breast MRI or US after an excisional biopsy or VABB to determine the extent of residual tumor around the excision site or multifocal cancers, followed by immediate curative surgery. We evaluated the correlation between the radiologic and pathologic findings after breast surgery.
Results:
Pathologic residual tumors were found in 42 patients (73.6%), and 13 patients (22.8%) demonstrated pathologic multifocal breast cancer. When detecting residual tumor using MRI, the sensitivity, specificity, positive predictive value, and negative predictive value were 59.5%, 46.7%, 75.8%, and 29.2%, respectively. For US, the sensitivity, specificity, positive predictive, and negative predictive values were 41.4%, 90%, 92.3%, and 34.6%, respectively. Invasive ductal breast cancers seemed to be more enhancing than ductal carcinoma in situ or extensive intraductal components containing breast cancer. MRI findings of residual tumors were not associated with residual tumor size (positive, 1.16 cm; negative, 1.0 cm; p=0.65). Thirty-nine patients (68.4%) underwent BCS, while 18 patients (31.6%) underwent mastectomy.
Conclusion:
The MRI accuracy of detecting residual lesions around previous scars was lower than that of US, except for sensitivity; however, other lesions distinct from the primary cancer and multifocal lesions were more frequently identified on MRI than US.
INTRODUCTION
Breast-conserving surgery (BCS) is the standard-of-care in the locoregional treatment of early breast cancer. Large, prospective, randomized trials have demonstrated that survival rates after BCS are equivalent to those after radical mastectomy [1,2]. Positive resection margins have been associated with a higher rate of local recurrence after BCS [3,4]. It is standard practice to obtain clear margins, even if this necessitates an additional surgical procedure, on the assumption that re-excision will convert a positive margin to a negative margin and thereby decrease the risk of local recurrence.
During the last 2 decades, tremendous advances have occurred in the performance and interpretation of breast magnetic resonance imaging (MRI). Several studies reported the higher sensitivity of contrast- enhanced magnetic resonance imaging (CE MRI) than a standard workup using conventional imaging for the detection of invasive breast cancer and disease extent visualization [5-8]. Breast MRI is helpful for women diagnosed with breast cancer who are contemplating BCS, as it provides valuable information regarding the extent of the disease. BCS is performed by many surgeons for breast cancer patients, except for those with multifocal breast cancer, central breast cancer, and extensive microcalcification. However, a previous excisional biopsy has been regarded as a relative contraindication; many surgeons have been attempting BCS because of improved preoperative imaging studies. Residual tumor after excisional biopsy has been reported in 32% to 63% of cases [9,10]. Many surgeons have used MRI for preoperative cancer staging to plan single-stage resections, especially in BCS. MRI can also help assess residual invasive breast cancer in patients with positive margins after an excisional biopsy [11]. Vacuumassisted breast biopsy (VABB) has been widely performed for the excision of benign-looking breast masses in Korea instead of excisional biopsy. In cases where breast malignancy was found, the evaluation of residual breast cancers was difficult [12].
The accurate evaluation of residual breast cancer after excisional biopsy or VABB is instrumental when planning BCS. We aimed to evaluate the accuracy of MRI for residual disease in breast cancer patients who underwent a previous excisional biopsy or excision using VABB and to compare it to the accuracy of ultrasonography (US).
METHODS
Fifty-seven patients diagnosed with breast cancer by surgical or excisional VABB at other breast clinics were referred to our institute for surgical treatment without a previous breast MRI evaluation.
The preoperative diagnostic methods were excisional biopsy in 41 cases and VABB in 16 cases. Patients were evaluated by US or CE MRI to evaluate residual lesion around the previous excision site and the presence of multifocal lesions. Approval from the Institutional Review Board was obtained for this retrospective study (2012–110).
Imaging
MRI was performed using a 3.0-T magnet (General Electric Medical Systems, Milwaukee, USA). Patients were placed in the prone position with each breast in a 4-coil compression breast array. The imaging protocol and parameters were as follows: an axial T1-weighted localizer sequence through both breasts followed by a sagittal fat-saturated T2-weighted fast spin-echo acquisition through the affected breast (TR/TE, 5,000/120). Next, images were obtained using a 3-dimensional (3D) slab-interleaved spoiled gradient-echo sequence with intermittent fat-selective partial inversion for fat suppression.
After the administration of 33 mL of gadopentetate dimeglumine (Berlex Laboratories, Wayne, USA), 3D volumetric dynamic images were obtained at 70-second intervals for a total of five cycles followed by a sagittal T1 (6.3/2.9–12) fat-saturated postcontrast sequence. MRI findings were interpreted by two experienced breast radiologists who were aware of the mammographic findings, but not the pathologic findings. Both radiologists have extensive experience in interpreting MRI examinations; they specialize in MRI with focus on breast MRI. A regular thin enhancing rim (1–2 mm) around the seroma cavity was considered negative for residual disease. We regarded enhancing lesions around the previous excisional biopsy using MRI according to the Breast Imaging Reporting and Data System (BI-RADS)-MRI Lexicon Classification; positive findings were defined as a greater than category 4A lesion, and negative findings included all category 3 lesions (Figures 1, 2). We characterized the lesions around the previous excisional biopsy using a dynamic curve analysis, such as early enhancing or delayed washout curves.
Additional lesions were identified in some breasts separate from the biopsy cavity. They were characterized by the radiologist as benign, probably benign, or suspicious using an interpretation model that incorporates both the architectural and kinetic features of breast lesions. If the lesion had a spiculated appearance it was regarded as a morphologically suspicious lesion. Findings with nonmass like enhancements, including linear, ductal, and segmental distributions, and findings with a kinetic pattern showing an early rise and an early washout were regarded as kinetically suspicious lesions.
All suspicious lesions (13 patients, 15 lesions) except for lesions around the previous excision or VABB site in the breast MRI evaluation, underwent a second examination on US. When other suspicious lesions were identified in the second US, they were examined by US-guided biopsy.
Thirty-nine patients underwent preoperative US before the breast MRI evaluation. Positive ultrasonographic findings were defined according to the BI-RADS-Ultrasound Lexicon Classification system around each previous biopsy site. As above, category 4 lesions were considered positive residual lesions, and negative findings included all category 3 lesions [13].
Pathology
During the pathological examination, the excised specimens were handled according to a fixed protocol based on radiologic and pathological assessments. Specimens were marked to indicate their orientation in the breast, sectioned into 5-mm parallel slices, and radiographs were obtained. Cross-sections of the residual tumor, surrounding breast tissue, or previous excision area and nearest margins, as well as any additional grossly or radiologically suspicious areas, were sampled for microscopic examination. The patient specimens were evaluated by two experienced pathologists for the existence of residual tumor, the pathologic type and extent of any residual tumor, existence of an extensive intraductal component (EIC), multifocal lesions, and primary breast cancer hormone receptor and HER2/neu status. Positive residual tumor was pathologically defined as ductal carcinoma in situ (DCIS) or invasive carcinoma around the previous biopsy site. In patients who had invasive cancer and DCIS, measurement of the invasive component was performed.
Statistical analysis
The statistical analysis was performed using SPSS version 12.0 statistical software (SPSS Inc., Chicago, USA). Differences between categorical variables were calculated using standard chi-square methods, and the Mann-Whitney U test and Fisher exact test were used for nonnormally distributed variables. To compare the two diagnostic modalities among 39 patients, we used McNemar test. A p-value <0.05 was considered significant.
RESULTS
Clinical characteristics
Fifty-seven patients diagnosed with breast cancer by surgical biopsy (41 cases) or excisional VABB (16 cases) were enrolled in this study. The patients ranged in age from 27 to 63 years (mean, 45.3 years). Histological findings from the initial excisional biopsy or VABB revealed DCIS in 14 patients (24.5%), invasive ductal carcinoma (IDC) in 41 patients (71.9%), and other types in two patients (3.6%). Thirteen patients (22.8%) with IDC had an EIC, which was defined as intraductal carcinoma involvement of greater than 25%. The sizes of the IDCs identified from the initial excisional biopsy ranged from microinvasive to 3.8 cm. The final pathologic T stages were Tis (12.3%), T1 (70.2%), and T2 (17.5%). In three cases, the pathology slides were submitted from an outside institution for review, and the initial size was not specified. Thirteen patients (22.8%) were diagnosed with multifocal breast cancer after final surgical treatment. Information summarizing the characteristics of the study population is shown in Table 1. Information summarizing the characteristics of the study patients according to the biopsy methods is shown in Table 2.
Diagnostic accuracy of MRI
From mammography, a glandular breast composition of approximately 51% to 75% was observed in 19 cases (33.3%) and greater than 75% glandular breast composition was observed in 38 cases (66.7%). Forty-two of 57 women demonstrated residual intraductal or invasive cancer at histologic examination after a re-excision lumpectomy or mastectomy, with a prevalence of 73.6%. Information regarding the diagnostic accuracy of MRI in evaluations of residual disease is summarized in Table 3. The sensitivity and specificity of MRI were 59.5% (25/42) and 46.7% (7/15), respectively. Positive predictive values and negative predictive values were 75.8% (25/33) and 29.2% (7/24), respectively. The pathologic characteristics of the 42 residual tumors with MRI findings are shown in Table 4. Residual invasive breast cancers (without EIC) were very well detected on MRI, but invasive carcinoma with EIC and DCIS were not well-identified (IDC, 80%; IDC+EIC, 16.6%; DCIS, 57.1%; p=0.03). Thirteen additional lesions detected on breast MRI were classified by the interpreting radiologists as suspicious, and underwent tissue sampling. Three lesions underwent mammographically guided needle localization and excisional biopsy, nine lesions underwent US-guided core biopsy, and one lesion was sampled by excisional biopsy through US-guided lesion marking. The pathology findings of the additional lesions observed in the ipsilateral breast on MRI were as follows: of the 13 suspicious lesions, five were malignant, with one case of DCIS and four cases of IDC. The overall prevalence of synchronous malignancy in the ipsilateral breast in this study was five of 57 (8.7%). The remaining eight lesions were benign on histologic examination.
Diagnostic accuracy of US
Thirty-nine patients underwent US before re-excision. Twenty-nine patients demonstrated residual intraductal or invasive cancer from the histologic examination after a re-excision lumpectomy or mastectomy, with a prevalence of 74.3%. Information regarding the disease diagnoses is summarized in Table 5. The US findings of residual tumor had a 41.4% (12/29) sensitivity, 90% (9/10) specificity, 92.3% (12/13) positive predictive value, and 34.6% (9/26) negative predictive value. Although the ultrasonographic evaluation of residual tumor had a lower sensitivity than that of MRI, US had a higher specificity, positive and negative predictive values, and accuracy than MRI evaluation. The accuracy of US for residual disease around the excisional site was higher than MRI (US: odds ratio [OR], 3.21, p<0.01; MRI: OR, 0.33, p=0.04). According to the pathology of the residual tumor in 39 patients, extensive intraductal tumor was more readily identified on MRI, but pure IDCs were not well-identified by either modality (Table 6).
DISCUSSION
The accurate interpretation of imaging of the postsurgical breast depends on the availability of high-quality images and pertinent medical and surgical breast histories. Many findings can be mistaken for cancer; skin thickening, architectural distortion, and other indicators of malignancy can be seen in both malignant and benign conditions. Prior to breast surgery, trauma, BCS, and lumpectomy can result in scarring and distortion on imaging studies. These findings can be mistakenly interpreted as suspicious for malignancy. Thus, understanding expected postsurgical imaging findings is important to ensure accurate interpretations and treatment recommendations.
Breast MRI is usually performed after breast cancer diagnosis by excisional biopsy or VABB [14]. Acute breast changes seen on breast MRI after surgical procedures include variable signal intensity hemorrhage on nonenhanced T1-weighted images, early or intense enhancement, or both. Hematomas and seromas manifest as a cavity with variable signal intensity (low to high) depending on whether they are acute or chronic [15]. Generally, hematomas and seromas have a thinner, smoother capsule, whereas residual or recurrent tumor is more nodular and irregular. Granulation tissue usually enhances moderately, and enhancement is delayed; however, fast enhancement may lead to a false-positive diagnosis [16]. In our studies, the false-positive rate of MRI was 53.3%, while that of US was 10%.
An important and controversial issue surrounding breast MRI is the dilemma of managing additional unexpected incidental enhancing lesions found on MRI that were undetected on physical examination, mammography, or US. Although overlaps in the appearances of benign and malignant lesions limits the utility of MRI evaluation for residual disease, MRI can depict additional suspicious lesions likely to represent multicentric or multifocal disease; these findings often alter surgical management. Our study is a rare clinical report to evaluate residual tumors around previous biopsy sites using preoperative breast MRI.
Lim et al. [17] suggest that preoperative MRI can potentially lower the rate of incompletely excised malignancies by identifying additional occult cancer prior to surgery, and does not lead to an increase in the mastectomy rate; however, because some benign lesions are indistinguishable from suspicious or malignant lesions, excessive surgical procedures are unnecessarily performed in a significant portion of patients. Meta-analyses found an 11% rate of aggressive surgery for true-positive MRI findings in the ipsilateral breast and 4.1% in the contralateral breast [18,19]. Pengel et al. [20] showed that preoperative MRI did not significantly affect the overall rate of incomplete tumor excision, but yielded a significantly lower rate of incompletely excised IDC. The reduction in incomplete excisions after MRI was smaller than the rate of prior treatment change resulting from MRI findings.
Lee et al. [21] reported on the impact of MRI on surgical treatment planning after excisional biopsy. Eighty-two patients underwent contrast-enhanced MRI before further surgery and subsequently underwent either a re-excision lumpectomy or mastectomy with histopathologic correlations. Residual carcinoma, either invasive or DCIS, was present in 59.8% of the breasts. The sensitivity and specificity of MRI for detecting residual disease were 61.2% and 69.7%, respectively, and the positive predictive value of identifying an additional suspicious lesion was 33.3%. MRI can show additional suspicious lesions likely to represent multicentric or multifocal disease. These findings changed the original treatment plan for approximately 30% of breasts.
In assessing the diagnostic accuracy of MRI in showing the presence and extent of residual disease, we found that MRI had a sensitivity, specificity, positive predictive value, and accuracy of 59.6%, 46.6%, 75.8%, and 56.1%, respectively. Our results were similar to those of previously reported studies [16,21].
Many investigators have asserted that overestimation of the extent of tumor on breast MRI results could lead to unnecessary, extensive surgery. In a retrospective study presented at the 2008 annual meeting of the American Society for Clinical Oncology, the percentage of women who underwent mastectomy was increased, and the use of preoperative MRI was also increased [22]. Houssami et al. [18] performed a systemic review and meta-analysis of the accuracy of MRI from 19 studies; MRI staging resulted in more extensive breast surgery in a significant proportion (6.6%) of women after additional lesions were identified.
This can be especially problematic when multiple enhancing lesions distant from the primary tumor site are identified. The biopsy of additional enhancing lesions that prove to be benign may compromise the patient’s cosmetic results after BCS.
A previous surgical biopsy is likely to change the breast parenchyma and result in postoperative changes, such as rim enhancement on MRI. Frei et al. [23] found that the time interval between excision and MRI influenced the specificity of MRI; a specificity of 75% was achieved when MRI was performed between 28 and 35 postoperative days.
It is likely that the interpretation of breast MRI was different resulting from the radiologists’ differing experiences regarding MRI postoperative findings and initial pathologic findings. They were more likely to easily accept real-time US than breast MRI findings, and were likely to categorize probable benign lesions around excisional sites as suspicious breast lesions. Additionally, there was a greater proportion (47%) of cases containing DCIS and extensive intraductal carcinoma components; thus, the diagnostic accuracy regarding residual tumor on breast MRI might be lower than that on US.
Limitations of the current study include that it was a retrospective analysis with a small cohort, and we were able to evaluate concurrent breast US and breast MRI in only 39 patients to compare the two modalities.
The accuracy of MRI for residual lesion around the previous scar was lower than that of US, except for sensitivity. However, other distinct lesions from the primary cancer or multifocal lesions were more frequently identified on MRI than US. Breast MRI could detect residual lesions from invasive carcinoma more easily than from DCIS or in IDC with EIC lesions. However, suspicious lesions detected on breast MRI must be re-evaluated in a second examination using US with core needle biopsy or MRI-guided biopsy because of the lower specificity and positive predictive value of MRI. Possible changes in therapeutic planning resulting from preoperative MRI findings should be decided by a multidisciplinary team.
Notes
CONFLICT OF INTEREST
The authors declare that they have no competing interests.