**3. 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.


Abbreviations: PL-E CESM—low-energy CESM images prior to NAC; NL-E CESM—low-energy CESM images after NAC; PS CESM—subtraction CESM images prior to NAC; NS CESM—subtraction CESM images after NAC; NHP—histopathological examination after NAC.

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).

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.

**Figure 1.** Correlations between: (**A**) the maximum tumor size before NAC in low-energy and subtraction CESM images: R = 0.89, *p* < 0.01; (**B**) the maximum tumor size after NAC in low-energy and subtraction CESM images: R = 0.57, *p* < 0.01.

**Figure 2.** Correlations between: (**A**) the maximum tumor sizes after NAC in low-energy CESM images and in histopathology results: R = 0.26, *p* < 0.04; (**B**) the maximum tumor sizes after NAC in low-energy CESM images and in histopathology results: R = 0.67, *p* < 0.01.

**Table 3.** Individual therapeutic responses to NAC using low-energy and subtraction CESM images.


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.


**Table 4.** Differences in NAC response depending on the type of breast cancer.

Abbreviations: CR—complete response; IDC—invasive ductal carcinoma; ILC—invasive lobular carcinoma; HP—histopathology examination.

> 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.


Abbreviations: CR—complete response; PR—partial response; SD—stable disease; PD—progressive disease; PPV—positive predictive value; NPV—negative predictive value.

> Figure 3 presents the differences in ROC curves for low-energy and subtraction CESM images in detecting CR.

> Figures 4 and 5 present an assessment of the therapeutic responses in two pairs of CESM images.

**Figure 3.** ROC curves based on the tested diagnostic methods (Youden Index:0.44, proposed cut-off point: 1.00): (**A**) for low-energy CESM images, the value of the AUC field was 0.718 at a standard error of 0.091 and *p* < 0.0172; (**B**) for subtraction CESM images, the value of the AUC field was 0.755 at a standard error of 0.064 and *p* < 0.0001.

**Figure 4.** Assessment of therapeutic response in low-energy MLO (**A**,**C**) and subtraction 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.

**Figure 5.** Assessment of therapeutic response in low-energy CESM CC images (**A**,**C**) and subtraction CESM CC images (**B**,**D**) before NAC (IDC LumB, G2 T3N1)showing irregular infiltration on the border of the outer quadrants of the left breast with high density (**A**), revealing pathological contrast enhancement on subtraction CESM images (**B**). 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.
