**4. 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].

In our study, the sensitivity of low-energy images in forecasting CR reached 33.33%, the specificity was 92.86%, the PPV was 70%, and the NPV was73.58%. After the conversion of subtraction images, the sensitivity of CESM in CR detection in a group of patients after NAC was 85.71%, its specificity was 71.42%, the PPV was 60%, and the NPV was 90.90%. Similar results for subtraction CESM images have been acquired by other authors such as Patel et al. [35] (64 patients, sensitivity: 95%, specificity: 66.7%, PPV: 55.9%, NPV: 96.7%), Iotti et al. [36] (46 patients, sensitivity: 100%, specificity: 84%, PPV: 57%, NPV: 100%), and Barra et al. [37] (33 patients, sensitivity: 76%, specificity: 62.5%, PPV: 86%, NPV: 45.4%). All mentioned studies, including ours, showed that, in the assessment of the CR following NAC, subtraction CESM images reached significantly higher sensitivity. Unfortunately, their specificity was lower. These results demonstrate that imaging techniques, even after intravenous administration of a contrast agent, may not differentiate between residual infiltration lesions and co-existing inflammatory/reactive lesions. A similar problem occurs with MRI, which tends to underestimate residual changes [23].

In their meta-analysis, Tang et al. [38] demonstrated that the pooled sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio (DOR) of the pathological response of breast cancer to NAC assessed by CESM were:

0.83 (95% CI, 0.66–0.93), 0.82 (95% CI, 0.68–0.91), 4.66 (95% CI, 2.59–8.41), 0.20 (95% CI, 0.10–0.43), and 22.91 (95% CI, 8.66–60.62), respectively. The same parameters for MRI, which is considered to be the best method for assessing response to NAC, were as follows: sensitivity: 0.77 (95% CI, 0.67–0.84), specificity: 0.82 (95% CI, 0.73–0.89), PLR: 4.35 (95% CI, 3.00–6.33), NLR: 0.28 (95%CI, 0.20–0.39), and DOR: 15.48 (95% CI, 9.97–24.03). Based on these findings, it can be concluded that CESM has the same specificity and higher sensitivity than MRI and is more accurate in the pathological evaluation of NAC response in breast cancer treatment.

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) [43–45]. 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 [46–48].

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 [49–51]. 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 governmen<sup>t</sup> program). There is a need for multi-institutional studies with larger groups of patients free of the limitations described above in the future.
