Diagnostic Value of Contrast-enhanced Digital Mammography Versus Contrast-enhanced MRI for Detecting Residual Disease after Neoadjuvant Chemotherapy for Breast Cancer

Background: Preoperative evaluation of breast cancer using contrast-enhanced digital mammography (CEDM) as a possible alternative to contrast-enhanced magnetic resonance imaging (CEMRI) has gained acceptance. Our aim is to compare the diagnostic performance of CEDM and CEMRI for chemotherapeutic response in breast cancer patients who underwent neoadjuvant chemotherapy (NAC). Methods: From November 2017 to October 2018, 30 patients with invasive carcinoma who performed both CEDM and CEMRI were included. Residual malignancy sizes after NAC were compared with histopathological results. The diagnostic performances for detecting residual cancers were compared using Lin concordance and Pearson correlation coecients. Results: Thirty patients were included for the analysis. Mean tumor size after NAC was 1.22 cm (range: 0–7.0 cm) for CEDM and 1.13 cm (range: 0–5.1cm) for CEMRI compared with 1.89 cm (range: 0–12.0 cm) at nal pathology measurement. Sensitivity for Conclusions: Diagnostic value of CEDM for detecting residual tumor after NAC was comparable to that of


Background
Neoadjuvant chemotherapy (NAC) is systemic therapy applying to patients with breast cancer, especially large inoperable cancer. NAC downsizes primary tumor, which increases the chances of breast-conserving therapy. Evaluating tumor response to NAC is important for speculating long-term outcomes and future treatment plans. Contrast-enhanced magnetic resonance imaging (CEMRI), ultrasonography (US), and mammography are imaging modalities used to predict chemotherapeutic response.
Alongside development of digital mammography, Lewin et al established contrast-enhanced digital mammography (CEDM) in 2003 [1]. CEDM traces uptake of iodine-based contrast agent in breast tissue.
Past studies have showed that CEDM yields valuable information on shape and vascular enhancement, concordant with CEMRI [2]. CEDM showed better sensitivity of breast cancer detection with high speci city due to higher contrast and better lesion delineation than mammography alone, even in dense breasts [3]. CEMRI shows low speci city for breast cancer detection, high false-positive (FP) rate and leading to overdiagnosis and overtreatment of breast cancer [4].
The preoperative role of CEDM as a possible alternative to CEMRI has gained acceptance. Recently published clinical results have reported that CEDM feasibility for detecting primary cancers was similar to that of CEMRI [5]. However, few studies have compared the diagnostic performance and effects of CEDM and CEMRI for evaluating remnant lesion after NAC in breast cancer [6,7].
Our aim is to compare the diagnostic performance of CEDM and CEMRI for assessing extent of remnant lesion in breast cancer patients who received NAC and to determine the accuracy of CEDM and CEMRI for predicting pathologic complete response (pCR).

Patient selection
From November 2017 to October 2018, 340 patients were diagnosed as primary breast cancer at OO Hospital, and 53 patients underwent NAC. Patients who were 18 years or older and histologically proven breast cancer with indications for NAC were included. Patients were excluded if pregnant or had anaphylactic reaction to any contrast agent according to the guidelines of the Korean Society of Radiology [8]. This prospective study was approved by the Institutional Review Board (OOOOO 2017-09-027). The use of iodinated contrast material was approved by Korean Food and Drug Administration.
Informed consent was obtained from all patients. Patient's clinical information was collected.
Contrast-enhanced digital mammography and contrastenhanced magnetic resonance imaging techniques Mammography was carried out before and after NAC using a full-eld digital mammography unit (Lorad Selenia; Hologic, Danbury, USA) using standard craniocaudal (CC) and mediolateral oblique (MLO) views.
CEDM and CEMRI were performed before and after NAC. All CEDM examinations were performed within 7 days of CEMRI using a digital mammography device capable of dual energy CEDM acquisition (Selenia Dimensions; Hologic, Bedford, USA). Previously, techniques of CEDM and CEMRI were described in detail [9,10].
Interpretation of contrast-enhanced digital mammography and contrast-enhanced magnetic resonance imaging All US, mammography, and CEMRI reports were interpreted using the American College of Radiology (ACR) Breast Imaging Reporting and Data System (BI-RADS) by ve breast imaging radiologists with 6-24 years of experience [11]. All patients were examined rst with CEMRI, followed by CEDM. The median interval between CEDM and CEMRI was 5.1 days (range: 1-29 days). CEDM results were reviewed by a radiologist who was blinded to the CEMRI results. The radiologist was blinded to the results of other studies when reviewing CEDM and CEMRI. During CEMRI interpretation, when there is discordant result, one radiologist reviewed CEMRI and reached consensus. The largest diameter of the contrast-enhancing lesion was correlated and categorized according to the response evaluation criteria in solid tumors (RECIST) before and after NAC [12]. In cases with multiple tumors, the largest tumor was used to measure size. Previously, method of radiologic response was described in detail [13].

Histopathological analysis
We obtained carcinoma tissue before NAC using core needle biopsies. The formalin-xed para nembedded specimens were cut into 3-µm-thick sections and stained with hematoxylin and eosin for microscopic evaluation. Immunohistochemical staining was performed on 3 µm-thick TMA block sections. Estrogen receptor (ER) and progesterone receptor (PR) status was assessed using the Allred scoring method [14]. Human epidermal growth factor receptor-2 (HER-2) overexpression was evaluated using American Society of Clinical Oncology/College of American Pathologists guideline recommendations. In cases with equivocal HER-2 staining (score 2), silver in situ hybridization (Ventana Medical Systems, Tucson, AZ, USA) was performed to determine HER2 gene status. Tumor response was de ned as follows; pathologic complete response (pCR) was de ned as absence of residual invasive carcinoma in resected breast specimens (ypT0/is) [15]. Pathologic tumor response was categorized according to RECIST criteria [12]. In non-pCR cases, the largest histopathologic diameters of residual tumors were measured.

Statistical analysis
In the absence of an a priori hypothesis, calculation of sample size could not be performed. Categorical data are summarized as frequencies and percentages. Tumor sizes measured by CEDM and CEMRI were compared to those measured on pathology (reference standard) within a range of -1 cm-1 cm. Agreements in histological size between CEDM and CEMRI measurements were evaluated using Bland-Altman plots. Lin's concordance and Pearson correlation coe cient were estimated to assess agreement between CEDM, CEMRI and histopathologic results. Sensitivity (true positive (TP) rate) was de ned as the probability by which a no-pCR can be detected by CEDM or CEMRI (TP/(TP + false negative (FN))); speci city (true negative (TN) rate) was de ned as probability by which a pCR can be detected by CEDM or CEMRI (TN/(false positive (FP) + TN)); positive predictive value (PPV) was de ned as probability by which no-radiologic CR(rCR) predicts no-pCR (TN/(FN + TN); and negative predictive value (NPV) was de ned as probability by which rCR predicts pCR (TP/(TP + FP)). The results were calculated with corresponding 95% con dence intervals (CIs).The McNemar's test was used to determine differences between CEDM and CEMRI results. A p-value < 0.05 (2-tailed) was considered statistically signi cant. All statistical analyses were performed with PASW Statistics for Windows, Version 18.0 (SPSS Inc., Chicago, IL, USA).

Discussion
NAC is used to test response or resistance of chemotherapeutic agent and reduce micrometastatic diseases [16,17]. Evaluating tumor response after NAC is important for determining future treatment plans and predicting long-term outcomes. Currently, CEMRI is the imaging modality for response monitoring [18]. CEMRI yields superior outcomes compared to mammography and US for identifying residual disease following NAC [19,20]. However, CEMRI has several limitations, such as long exam time, high cost, low accessibility, and longer interpretation time. In comparison, CEDM is known for shorter procedure time, cheaper price, easy interpretability, shorter interpretation time, and no claustrophobia risk [10,21,22]. Also, female patients do need to schedule their imagining appointment according to menstrual cycle. Some clinicians have suggested to conduct CEDM and CEMRI according to menstrual phase; however, some argue no clear pattern in background parenchymal enhancement variation across the menstrual cycle for either CEDM or CEMRI [23]. CEDM also revealed equivalent result for detecting primary and multifocal cancers and estimating extent of cancers compared to CEMRI [5].
In present study, CEDM was evaluated for detecting and measuring remnant tumors after NAC. Although CEDM underestimated remnant tumor size in some patients, the difference between CEDM and pathologic results was less than 1 cm in 63.3% of cases. Barra et al. [24] reported similar results, with a difference between CEDM and pathologic size within 1 cm in almost 70% of cases. Limits of agreement with the residual tumor were lower for CEMRI than for CEDM, which was consistent with that found by Patel and colleagues [6].
Regarding diagnostic performance, CEDM demonstrated better speci city, PPV, and NPV compared to CEMRI, though CEMRI had better sensitivity than CEDM. Similarly, other studies reported comparable speci city and PPV for CEDM [6,25]. All women with radiologic CR on CEMRI had negative CEDM results.
Our results also showed good correlation and concordance of CEDM with histopathologic result after NAC (Figs. 2 and 3, Table 3).
CEDM and CEMRI both underestimated the extent of remnant cancers (mean tumor sizes from CEDM, CEMRI, and pathology were 1.22 cm, 1.13 cm, and 1.89 cm, respectively). Iotti et al. [26] also reported underestimation of tumor size by CEDM and CEMRI after NAC. Wu et al. [27] explained this discrepancy as an antiangiogenic effect of chemotherapeutic agents used in CEDM and CEMRI, such as taxanes and anthracyclines. This anti-vascular effect, which lacks a surrounding in ammatory response, lowers enhancement of tumor tissue and, thus, can result in tumor underestimation [20].
Limitations of the present study are associated several factors. First, we enrolled a small number of patients. This was possibly because patients lacked an understanding of CEDM. Since CEDM was recently introduced in South Korea, patients were not familiar with it, and most refused to undergo additional mammography because of the pain induced by breast compression during the procedure. Second, the ability to detect residual ductal carcinoma in situ (DCIS) or suspicious microcalci cations remains unknown for both CEDM and CEMRI. CEDM and CEMRI identify occult breast cancer using mechanism of tumor angiogenesis [28]; however, the underlying mechanism of tumor angiogenesis depends on subtype of breast cancer. Neo-angiogenesis is more prominent in IDC and is not always exist in DCIS and lobular carcinomas [29]. In our study, none of pCR patients showed residual DCIS component on residual tumor. The feasibility of CEDM for DCIS should be further evaluated. Third, we could not differentiate the diagnostic accuracy of CEDM according to presence or absence of microcalci cations.
Finally, we could not assess the accuracy of CEDM according to molecular subtype, and further investigations are needed.
Our study also has several strengths. First, this is the rst CEDM study conducted among breast cancer patients who received NAC in South Korea.. Second, we found that CEDM is secure and practical. There was no technical failures during procedure and one patient had side effects during CEDM examinations. The patient vomiting one minutes following injection of contrast media. The symptoms disappeared soon. Finally, because this was a prospective study, there was no inherent bias.

Conclusions
In conclusion, CEDM showed diagnostic results comparable to CEMRI in demonstrating residual cancers and estimating residual tumor extent after NAC. CEDM can be as effective as CEMRI for evaluating NAC patients. Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interest
The authors declare that they have no competing interests.