Contrast-Enhanced Spectral Mammography Assessment of Patients Treated with Neoadjuvant Chemotherapy for Breast Cancer

Background: Evaluating the tumor response to neoadjuvant chemotherapy is key to planning further therapy of breast cancer. Our study aimed to evaluate the effectiveness of low-energy and subtraction contrast-enhanced spectral mammography (CESM) images in the detection of complete response (CR) for neoadjuvant chemotherapy (NAC) in breast cancer. Methods: A total of 63 female patients were qualified for our retrospective analysis. Low-energy and subtraction CESM images just before the beginning of NAC and as a follow-up examination 2 weeks before the end of chemotherapy were compared with one another and assessed for compliance with the postoperative histopathological examination (HP). The response to preoperative chemotherapy was evaluated based on the RECIST 1.1 criteria (Response Evaluation Criteria in Solid Tumors). Results: Low-energy images tend to overestimate residual lesions (6.28 mm) and subtraction images tend to underestimate them (2.75 mm). The sensitivity of low-energy images in forecasting CR amounted to 33.33%, while the specificity was 92.86%. In the case of subtraction CESM, the sensitivity amounted to 85.71% and the specificity to 71.42%. Conclusions: CESM is characterized by high sensitivity in the assessment of CR after NAC. The use of only morphological assessment is insufficient. CESM correlates well with the size of residual lesions on histopathological examination but tends to underestimate the dimensions.


Introduction
Neoadjuvant chemotherapy (NAC) is currently one of the basic methods of treatment of locally advanced breast cancer [1,2]. The main advantage of NAC use is the possibility of reducing the mass of previously inoperable tumors to allow surgical intervention. In the case of massive yet operable tumors, NAC provides a decrease of their mass to a clinical use in the United States in 2011. During CESM examination, approximately 2 min after contrast administration (chelated iodine-based X-ray contrast agent), dual-energy mammography is performed. During one examination, two images are acquired: one resembling conventional mammography (low-energy images of equal or noninferior quality to those of standard digital mammography) and another-the subtraction imagedemonstrating areas of increased contrast enhancement (like in the case of the CE-MRI that uses Gadolinium-based contrast dose) [24,25]. Another advantage of CESM is that it provides both the anatomical and morphological characteristics of the lesions in the breast and regions of increased neoangiogenesis. CESM is characterized by comparable or even better sensitivity and a higher negative predictive value (NPV) and positive predictive value (PPV). Moreover, in contrast to MRI, CESM allows the visualization of microcalcifications [26,27]. Due to the much shorter examination time, lower costs, higher comfort, and low anxiety involved, patients tend to cooperate more fully and prefer CESM to MRI [28].
Consequently, the use of CESM in the evaluation of response to NAC is justified, bearing in mind the efficiency of this method, which is comparable to CE-MRI. According to the European Society of Breast Cancer (EUSOBI) recommendation in 2017, CESM can be considered as an alternative to contrast-enhanced MRI in the case of contraindications to MRI. CESM is a younger and less-studied technique than MRI. In CESM, there is an extra radiation dose of approximately 20%, but both images-low-and high-energy-still imply an X-ray dose below the recommended dose for mammography. In patients treated with neoadjuvant chemotherapy for breast cancer when a CESM examination is planned, additional MG can be avoided, with the possibility of saving the radiation dose [29].
Therefore, our study aimed to evaluate the effectiveness of low-energy and subtraction CESM images in the detection of CR for neoadjuvant chemotherapy in breast cancer. In addition, we decided to correlate the residual changes assessed with CESM low-energy and subtraction images.

Materials and Methods
A total of 63 female patients (mean age of 53.32 ± 9.47 years; ages ranged from 33 to 75 years) were qualified for our retrospective analysis. Between 2016 and 2019 they were put forward for NAC due to breast cancer in the University Clinical Center of the Medical University of Silesia in Katowice, Poland. Inclusion criteria encompassed: a diagnosis of breast cancer (based on core needle biopsy), minimal stage of T1N+, age of more than 18 years, completion of NAC treatment, surgery after NAC, and a complete set of imaging examinations-CESM; ultrasonography (US) of the breast, lymph nodes, and abdomen; and chest X-ray. Exclusion criteria were as follows: known BRCA1 or BRCA2 mutation (to provide the highest radioprotection), pregnancy, anaphylaxis or anaphylactoid reaction to a contrast agent in medical history, e-GFR of less than 60 mL/min, chronic kidney disease in stage 3 or higher, and incomplete NAC. Patients with significant post-biopsy changes (e.g., hemorrhage) affecting image quality were also excluded. A full description of the study group is in Table 1. Due to the retrospective nature of this study, the local ethics committee of the Medical University of Silesia repealed the requirement for informed consent (decision number PCN/0022/KB/157/20). All the test procedures were carried out in compliance with the ethical principles of the 1964 Helsinki Declaration and its subsequent amendments.

Neoadjuvant Systemic Therapy
After the end of breast cancer diagnosis, each patient's final decision concerning the proper treatment was made by the multidisciplinary breast cancer team (BCU) unit, taking into account patients' preferences. Following the decision by the BCU team, 98% of patients received chemotherapy based on anthracyclines and taxanes (for a period of 12 weeks), including 30% in the "dose-dense" regimen. Only one patient did not receive anthracyclines due to a limited dose of anthracyclines in earlier treatment of Hodgkin's disease. Due to the lack of public funding for trastuzumab in this center in the years 2016-2019 (no drug prescription government program), all the cancer patients with positive HER2 were treated outside our center and were not included in this analysis.

CESM Examination
All CESM examinations were performed in our hospital. They were carried out with the use of a digital mammography device dedicated to performing dual-energy CESM acquisitions (SenoBright, GE Healthcare, 3000 N. Grandview Blvd., Waukesha, WI, USA). An intravenous injection of 1.5 mL/kg of body mass of non-ionic contrast agent was performed using a power injector at a rate of 3 mL/s with a bolus chaser of 30 mL of saline. In CESM mode, the device automatically performed a pair of exposures (low-and high-energy) in each view. Specific image processing of low-energy and highenergy images was done. This processing aimed to obtain subtraction images to highlight contrast enhancement and suppress structured noise due to fibroglandular breast tissue. The morphological information obtained from low-energy images of CESM is similar to the morphological information given by standard mammography and the functional information obtained from subtraction images of CESM visualizes the vascularization of breast lesions [24,25,30]. For each view, the CESM technique made it possible to obtain two images: a low-energy acquisition at 26-30 kVp and a high-energy acquisition at 45-49 kVp, with these values depending on breast density and thickness. Motion blur could be sometimes observed on subtraction images due to movements between the acquisition of low-and high-energy images. All of the images obtained were in the DICOM format [31]. The total examination time was usually 10 min. After examination, the patients were observed for approximately 30 min for any adverse reactions that may have occurred after administration of the contrast agent.

Imaging Interpretation
All CESM examinations were carried out two times: just before the beginning of NAC and as a follow-up examination 2 weeks before the end of chemotherapy to evaluate its effect (and to inform decisions about possible changes in therapeutic strategies). For our retrospective analysis, we performed renewed CESM image evaluations with every patient. Such evaluations were made separately by two different radiologists with 20 years of experience and a minimum of 5 years' experience in CESM image interpretation. Radiologists were blinded from each other and from the results of histopathological examination. CESM images were assessed according to the Breast Imaging-Reporting and Data System (BI-RADS) scale. Three measurements were taken in the CC and MLO projections, while the statistical analysis encompassed one-i.e., the largest dimension of the tumor. The tumor dimensions were compared while analyzing: • Low-energy images from two consecutive contrast-enhanced spectral mammograms (taken before the start and at completion of neoadjuvant chemotherapy); • Subtraction images from two consecutive contrast-enhanced spectral mammograms (taken before the start and at completion of neoadjuvant chemotherapy).
The response to preoperative chemotherapy was evaluated based on the Response Evaluation Criteria in Solid Tumors (RECIST) 1.1 criteria. The response was classified as follows: complete response (CR, disappearance of all lesions); partial response (PR, ≥30% dimensional reduction), stable disease (SD, <30% dimensional reduction/<20% dimensional increase), and progressive disease (PD, ≥20% dimensional increase) [17,32]. For this article, PR, SD, and PD have been classified as "non-CR".
After surgery, a comparison of the NAC response, evaluated with the use of CESM (both low-energy and spectral images), to the histopathological study results was undertaken. Our analysis of histopathology (HP) was used as a "golden standard" against which the effectiveness of both CESM image types in NAC response detection could be analyzed.

Histopathological Examination
The histopathological examination was conducted in the Histopathology Laboratory of our center by two pathologists with broad experience in breast cancer diagnostics. The greatest dimension of the tumor, necessary for determining the T descriptor in the pTNM classification apart from the macroscopic measurement, was verified histopathologically. This verification was undertaken using a microscope and the Olympus Cell Sens Dimension ® software (Japan, 2013). Tumors up to 2 cm were excised whole, serially, on a cross-sectional basis with a margin of 0.2 to 0.4 cm and embedded in a paraffin block after each cross-section. Tumors measuring over 2 cm that could not fit within a single paraffin block were divided into two or more parts by making parallel cuts of the lesion. Next, they were marked in pairs with ink of the same color and the individual layers were given numbers to allow for the restoration of the largest section of the tumor. The T value of the tumor was the total of the parallel measurements of the particular parts of the lesion.

Statistical Analysis
Tumors that underwent histopathology examination were used as the "gold standard" and compared to the tumor sizes derived from CESM images. The normality of the distribution was assessed using the Shapiro-Wilk test and the continuous variables were summarized using the arithmetic mean with standard deviation for data following a normal distribution or a median with quartiles 1 and 3 for data demonstrating anonnormal distribution. Comparison of four respective aspects referring to the maximal tumor dimension (defined as the maximum of three dimensions measured in the CESM and histopathology) included the Pearson's correlation coefficient (R-value) in order to measure the strength of the relationship between low-energy and subtraction images of CESM measurements. Paired t-tests were used to assess mean differences between each analyzed study participant. The correlations of data are illustrated by plotting the actual measurements while all paired measurements for each patient are summarized using paired linear plots. A p-value of <0.05 was considered statistically significant. The diagnostic performance indexes for low-energy and subtraction images for complete response and non-complete response were tested using the Clopper-Pearson test with 95% confidence intervals. Statistical analyses were carried out using Statistica 10 (StatSoft Inc., Tulsa, OK, USA) and MedCalc Statistical Software 16.4.3 (MedCalc Software by, Ostend, Belgium) software.

Results
Before neoadjuvant chemotherapy, the average size of the tumors varied from 34.4 mm for low-energy CESM to 34.3 mm for CESM subtraction images. After neoadjuvant chemotherapy, their average sizeswere17.6 mm for low-energy images and 8.5 mm for CESM subtraction images. The average size of the lesions in the histopathological examination was 11.1 mm ( Table 2). The average reduction of the tumors reached 52.22% of the initial tumor mass based on low-energy images, and this even reached 78.76% in the case of CESM subtraction images. Table 2. Dimensions of the lesions before and after neoadjuvant chemotherapy for low-energy and subtraction CESM images and histopathological examination. When comparing the maximum tumor dimensions before neoadjuvant chemotherapy for low-energy and subtraction CESM images, a high degree of correlation in the Spearman's analysis (R = 0.89, p < 0.01) was noticed. When the comparison between the maximum tumor dimensions after neoadjuvant therapy for low-energy and subtraction CESM images is considered, the correlation between the results can be described as moderate (R = 0.57, p < 0.01) (Figure 1).

Maximal
A certain correlation, defined as moderate (R = 0.44, p < 0.01), can also be observed upon the comparison of the maximum tumor reductions for low-energy and subtraction CESM images. In terms of comparing the measurements of the maximum size for low-energy and subtraction CESM images following NAC and the maximum size in the histopathological examination, there was a low level of correlation for low-energy CESM images (R = 0.26, p < 0.04) and a high level of correlation for subtraction CESM images (R = 0.67, p < 0.01) (Figure 2).
Both pairs of images tended to imprecisely estimate the sizes of residual lesions. In the case of low-energy images, the size of these lesions was overestimated (the average overestimation value was 6.28 mm), whereas in the case of subtraction CESM images, residual lesions were underestimated (the average underestimation value was 2.75 mm).
According to the RECIST 1.1 guidelines, the low-energy images with morphological assessment only revealed 15.87% CR (n = 10) and 84.13% non-CR (n = 53). In the case of subtraction CESM images, these parameters were 47.62% (n = 30) and 52.38% (n = 33), respectively. Histopathological examination demonstrated CR in 33.33% (n = 21) of cases and non-CRs in 66.67% (n = 42). A detailed description of the particular responses to NAC can be found in Table 3.   Abbreviations: CR-complete response; PR-partial response; SD-stable disease; PD-progressive disease.
In the histopathological examination for invasive ductal carcinoma (IDC), a CR of 39.22% (20 out of 51 of tumors) could be achieved, whereas for invasive lobular carcinoma (ILC) it was16.67% (1 out of 6 tumors). In the case of mixed IDC/ILC, CR was not achieved in any of the tumors. In Table 4, differences in the NAC response depending on the type of breast cancer analyzed by CESM and histopathological examination are shown. Comparing the two types of CESM images to the histopathological examination, Table 5 presents the sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) of both images in the prediction of the CR. The sensitivity of lowenergy images in forecasting CR amounted to 33.33%, while its specificity was 92.86%. In the case of the subtraction CESM images, the sensitivity amounted to 85.71% and the specificity to 71.42%. Table 5. Diagnostic performance indexes for the assessment of complete (CR) and non-complete response (PR, SD, PD) according to RECIST 1.1criteria using low-energy and subtraction CESM images compared to histopathology results.     MLO (B,D). Before NAC, a tumor can be seen in the upper outer quadrant of the right breast with high density and polycyclic outlines, accompanied by enlarged lymph nodes in the axillary fossa (A), revealing pathological contrast enhancement in subtraction CESM images (B). Following NAC, in the tumor field, there is a visible focal asymmetry, with a density lower than the residual glandular tissue (C), without pathological contrast enhancement (D). Based on the low-energy images, the therapeutic response was classified as partial response (PR). Based on the subtraction images, the therapeutic response was classified as complete response (CR), which was acknowledged in the HP examination. Additionally, satellite foci are visible in subtraction CESM images, which were confirmed in core-needle biopsy (smaller arrow) (B). Following NAC, there was a visible focal asymmetry, with a density slightly lower than the infiltration before NAC (C), shown again without pathological contrast enhancement (D). Based on the low-energy CESM images, the therapeutic response was classified as stable disease (SD). Based on the subtraction CESM images, the therapeutic response was classified as complete response (CR), which was acknowledged in the HP examination.

Discussion
Early assessment of NAC response to breast cancer and correct differentiation between patients with the complete pathological response (pCR) and without NAC response is the key point in NAC therapy. Proper evaluation is crucial to the future clinical perspectives and therapy for each patient. Obtaining complete pathological response results in better event-free survival (EFS) and overall survival (OS) rates [33,34].
The largest pretreatment tumor dimensions in our analysis of low-energy and subtraction CESM images were similar and there was a statistical difference between these modalities (R = 0.89, p = 0.01). However, these differences became significant following neoadjuvant chemotherapy (R = 0.55, p = 0.01). This was because post-NAC tumors reduce their density and then become difficult to distinguish from glandular tissue based on morphological images alone. On the other hand, the functional information provided by CESM in subtraction images showed that the residual infiltration was visible, and the type of breast tissue did not affect its visualization. Moreover, in our study, the comparison of the measurement of the maximal sizes of residual changes after NAC, evaluated using CESM subtraction images and histopathology, showed a high correlation (R = 0.67, p = 0.01). Iotti et al. [36], while comparing the sizes of tumors after NAC with a histopathological examination, proved that CESM showed greater coherence with histopathology than MRI (Lin's coefficient were 0.81 and 0.59, respectively). In contrast, Patel et al. [35] achieved the opposite results, where MRI showed higher compatibility with histopathology than CESM (Lin's concordance coefficient was 0.75 for CESM and 0.76 for MRI; Pearson's correlation was 0.77 for CESM and 0.80 for MRI). The lack of consistency between researchers indicates the need for further research-multicenter, on a larger group of patients, using equipment available on the market from various companies.
In our study, Low-energy images tended to overestimate the dimensions of residual lesions following NAC, while subtraction CESM images tended to underestimate them. Similar results were acquired by Patel et al. [35] and Iotti et al. [36], where CESM and histopathology results underestimated the size of post-NAC tumors by 5mm and 4.1mm, respectively. It should also be emphasized that the underestimation of the dimensions of residual lesions in our study had no impact on the scope of surgical treatment. Since CESM is a method involving vascularization of the tumor foci, the effect of excessive reduction in vascularization around the tumor during NAC may account for the weaker enhancement of the residual tumor mass on follow-up CESM, underestimating the actual dimension of residual lesions. A similar problem arises with MRI, which tends to underestimate residual lesions in follow-up examinations [35,36]. Over-or underestimation of residual disease can also be a result of the fact that, due to neoadjuvant chemotherapy, induction of cellular changes leading to the elimination of cancerous cells occurs before the decrease in tumor size [39,40]. Moreover, after the eradication of cancerous cells in the region of the tumor, the residual mass visible on radiological images may still be present. It consists of fibrotic tissue left after the therapy. To overcome this problem, Xing et al. [41] and Moustafa et al. [42] suggested relying not only on RECIST 1.1 criteria in the evaluation of NAC response but also creating a special mathematical model. It consists of a combination of the measurement of the largest diameter of the target lesion together with subjective identification of the difference in the intensity of the contrast uptake before and after NAC. After that, a combination of the summation of the number of pixels and their intensity within the area of interest before and after NAC is included in this model. It should be noted that, after using this method, CESM remained a method of high sensitivity and specificity in the evaluation of NAC response. Such results prove that CESM is one of the best methods for diagnostic imaging available for the analysis of the response to neoadjuvant chemotherapy.
The histological and molecular heterogeneity of tumors cannot be evaluated with a simple examination with a contrast medium, and biopsies are limited, especially in large tumors. The need to better define the heterogeneity of tumors will involve the aid of new methodologies currently under study, such as radiomics. Radiomics in MRI can be more effective in the diagnosis of breast cancer and in the histological and morphological assessment of tumors. For CESM, a radiomics model achieved a significantly better discriminative ability compared to the standard clinical model (AUC, 0.81 vs. 0.55, p < 0.01) [43][44][45]. In recent years, it has also been suggested to use background parenchymal enhancement (BPE) in the assessment of responses to NAC, which describes the enhancement of the normal breast tissue related to physiological vascularization. La Forgia et al., indicated that BPE is an important aspect that conditions the diagnosis and that it is a potential predictive factor in the response to neoadjuvant cancer therapies in graphic contrast examinations, as already confirmed in MRI and CESM examinations [46][47][48].
CESM, while being a relatively novel diagnostic imaging technique, has become a useful tool in diagnosing and evaluating breast cancer stages. Subtraction images improve the diagnostic specificity of low-energy images, providing more precise measurement of tumor size as well as the possibility of identifying multifocal diseases, especially in women with dense breast tissue. Thus, the effectiveness of CESM in breast cancer diagnosis is comparable to MRI and is a promising tool to serve as a basic imaging technique in patients with symptomatic breast cancer or for the detection of multifocal and multicentric breast cancers [49][50][51]. As we have shown in our study, CESM is also a useful and effective method in assessing the pathological response to NAC. Post-NAC treatment monitoring is extremely important for planning surgical treatment, reducing the number of mutilating mastectomies, and replacing them with breast-conserving surgeries. However, the most accurate measurements possible should be made to avoid underestimating the size of the tumor and increasing the extent of the operation. The advantage of CESM over other methods so far available is its precise definition of the tumor before NAC thanks to the directly integrated visualization of suspicious calcifications in the low-energy images and enhancement in the recombined images, which is not possible with MRI [52].
Our study had several limitations. Firstly, this single-institution research was carried out on a relatively small number of female patients due to the limited number of qualifications for NAC. Moreover, all CESM examinations were conducted on a single vendor system. Finally, we did not include patients with HER2-positive cancer in our study due to the lack of public funding for trastuzumab in this center in the years from2016-2019 (no drug prescription government program). There is a need for multi-institutional studies with larger groups of patients free of the limitations described above in the future.

Conclusions
Due to the possibility of assessing vascularity, CESM is characterized by high sensitivity in the assessment of CR after NAC. The use of morphological assessment alone is insufficient. CESM correlates well with the sizes of residual lesions from histopathological examination but tends to underestimate the dimensions.

Conflicts of Interest:
The authors declare no conflict of interest.