Contrast-enhanced mammography in the preoperative assessment of primary operable breast cancer: Results from a prospective, two-centre study

Abstract

Purpose

To evaluate the performance of contrast-enhanced mammography (CEM) in the preoperative assessment of patients with primary operable breast cancer and in the detection of multifocal or multicentric carcinoma (MFMCC). Secondary aims were to determine the added value of recombined images (CEM-RI) relative to low-energy images (CEM-LEI) in detecting MFMCC.

Materials and Methods

In this prospective multicenter study, 184 patients with newly diagnosed breast cancer were included. All patients underwent CEM, and findings from CEM-LEI and CEM-RI were analysed separately for the detection of MFMCC, using histopathology as the reference standard. The influence of CEM-RI findings on surgical decision-making was evaluated. All patients completed a 12-month clinical and imaging follow-up to assess residual disease and recurrence.

Results

Histopathology confirmed MFMCC in 30/184 patients (16.30%). Sensitivity for MFMCC detection was significantly higher for CEM-RI than for CEM-LEI (86.67% vs. 53.33%, p), while specificity remained high for both techniques (91.56% vs. 90.26%, p). When both image types were assessed together, overall sensitivity increased to 93.33%. CEM-LEI sensitivity decreased with increasing breast density (63.16% for American College of Radiology (ACR) A + B vs. 36.36% for ACR C + D), whereas CEM-RI maintained high sensitivity irrespective of density (78.95% for ACR A+B vs. 100% for ACR C+D). CEM-RI findings led to modification of the planned surgical procedure in 23 of 184 patients (12.43%). The reoperation rate due to R1 resection after breast-conserving therapy was 3.80%.

Conclusion

CEM-RI demonstrates significantly greater sensitivity than CEM-LEI for detecting MFMCC and contributes to clinically meaningful changes in surgical management.

Highlights

• •Accurate preoperative assessment of breast cancer extent and detection of multifocal and multicentric disease remain essential for appropriate treatment planning.
• •CEM-RI demonstrated significantly higher sensitivity than CEM-LEI for detecting multifocal multicentric disease (86.67% vs. 53.33%), while maintaining high specificity.
• •Combined assessment of low-energy and recombined images increased overall sensitivity for additional lesions detection to 93.33%.
• •CEM-RI findings led to modification of the planned surgical procedure in 12.43% of patients, with a low reoperation rate (3.80%) and no distant recurrence at 12-month follow-up.

1. Introduction

Accurate preoperative assessment of breast cancer extent and detection of multifocal and multicentric disease (MFMCC) remain essential for appropriate treatment planning, particularly in the era of increasingly adopted breast-conserving approaches and ongoing efforts toward further therapeutic deescalation, including the evaluation of percutaneous ablative techniques, e.g., cryoablation and radiofrequency ablation, in selected patient groups or the expansion of indications for accelerated partial breast irradiation (APBI) [1], [2], [3]. Precise staging and reliable detection of all malignant foci are crucial for patient safety and optimal therapeutic outcomes. Conventional imaging modalities such as full-field digital mammography (FFDM), digital breast tomosynthesis (DBT), and breast ultrasound (US) remain the fundamental methods in breast cancer diagnosis [4]. Although digital mammography is the only modality proven to reduce breast cancer mortality in screening settings, its sensitivity is limited, especially in dense breasts [5]. Consequently, functional imaging techniques have gained importance in improving diagnostic accuracy.

Contrast-enhanced mammography (CEM) adapts the principles of contrast-enhanced magnetic resonance imaging (CE-MRI) to a mammographic platform by generating recombined images from low- and high-energy acquisitions obtained after intravenous contrast administration [6]. While CE-MRI remains the reference standard for functional breast imaging, its use is constrained by high cost, limited availability, and suboptimal performance in the evaluation of suspicious calcifications [7], [8], [9]. CEM has emerged as an accessible, cost-effective, and well-tolerated alternative. Several studies have demonstrated that its sensitivity approaches that of breast CE-MRI. Moreover, CEM offers improved diagnostic performance over morphological techniques in cancer detection and has been increasingly adopted for clinical indications historically reserved for CE-MRI [6], [10]. According to the National Comprehensive Cancer Network [11], CEM may be used for preoperative assessment of breast cancer in cases where CE-MRI is not available. Evidence also suggests that CEM may provide comparable or even better benefits to CE-MRI in preoperative planning, including similar rates of surgical plan modification and superior delineation of disease extent compared with mammography, tomosynthesis, and ultrasound [10], [12].

Given the promising results reported in retrospective and limited prospective studies, the growing clinical relevance of accurate preoperative staging, and the need for robust comparative data, we conducted a prospective, two-centre study to evaluate the performance of CEM in the preoperative assessment of patients with primary operable breast cancer and in the detection of MFMCC. Secondary aims were to determine the added value of recombined images relative to low-energy images in the detection of MFMCC, and to assess the proportion of cases in which recombined imaging led to modification of the planned surgical procedure. Additionally, postoperative histopathological margins were analysed to assess the reoperation rate, and a one-year follow-up was conducted to evaluate late complications and recurrence.

2. Material and method

This prospective, two-centre study was approved by the Institutional Review Board and Ethics Committee (Decision number: BNW/NWN/0052/KB1/93/1/23/24). Written informed consent was obtained from all participants. All study procedures complied with the ethical principles of the 1964 Declaration of Helsinki and its subsequent amendments.

2.1. Patient population and study protocol

Patients were consecutively recruited from two breast units in southern Poland who presented for evaluation of lesions suspected of malignancy between January and December 2024. The cohort consisted of women with suspicious findings identified in external facilities or through screening, as well as symptomatic patients with palpable or visible masses, or other clinical features suggestive of malignancy, all of whom had not undergone prior diagnostic assessment.

Inclusion criteria comprised: newly diagnosed breast cancer eligible for surgical management, no contraindications to CEM, and written informed consent. Exclusion criteria included: planned neoadjuvant chemotherapy, metastatic or palliative-stage disease, contraindications to iodinated contrast or ionising radiation (e.g., pregnancy, contrast allergy, impaired renal function), and refusal to participate. Patients deemed unsuitable for surgery due to advanced age or significant comorbidities, as well as those declining surgical treatment despite consenting to diagnostic procedures, were also excluded (Fig. 1).

Fig. 1.

Flowchart of the study.

Patients referred from external institutions with suspicious breast lesions underwent supplemental imaging (FFDM and/or US) if required based on the available diagnostic work-up. Symptomatic patients primary reffered to our institutions underwent primary assessment with FFDM and US. All patients underwent biopsy using the modality best depicting the lesion, with a minimum of 5 cores obtained under ultrasound guidance and at least 12 cores for mammography-guided biopsy. A marker clip was placed in each sampled lesion. Both solid masses and malignant microcalcifications were considered eligible. The primary lesion identified at presentation was designated as the index lesion.

The initial treatment plan was determined at a Breast Cancer Unit (BCU) meeting based on FFDM and ultrasound findings, prior to surgical consultation and study enrolment. All study participants subsequently underwent CEM as an additional preoperative examination. When CEM recombined images revealed additional suspicious foci, a second-look ultrasound was performed; lesions detected on ultrasound were biopsied using the core needle technique. If the lesion was not sonographically visible, a stereotactic biopsy was performed, using surrounding morphological landmarks when necessary or a biopsy was performed under CEM or CE-MRI guidance. CEM was scheduled within the routine preoperative interval and did not delay treatment. Findings were reviewed at a second BCU meeting, during which modifications to the surgical plan were made as indicated. All patients underwent definitive surgery, followed by adjuvant therapy according to tumour subtype and BCU recommendations. Follow-up imaging was performed at 6 months and 12 months postoperatively.

Because some FFDM and ultrasound examinations had been performed outside our institution and at variable time intervals before presentation, only the CEM low-energy images (CEM-LEI) and recombined images (CEM-RI) were included in the analytical dataset.

2.2. CEM protocol and image analysis

All CEM examinations were performed at two radiology centres using three dual–energy–capable digital mammography systems (Pristina, GE Healthcare, Waukesha, USA; Giotto Class 30.000, IMS Giotto, Sasso Marconi, Italy and Selenia Dimensions™, Hologic, Marlborough, USA). Contraindications were excluded prior to each examination. Following an intravenous injection of a low-osmolar, non-ionic iodinated contrast agent (1.5 mL/kg) administered at 3 mL/s with a 30-mL saline chaser, delayed dual-energy acquisitions were obtained in CEM mode. Each projection generated paired low-energy (26–30 kVp) and high-energy (45–49 kVp) images, followed by automated recombination to produce subtraction images emphasising contrast enhancement while suppressing fibroglandular background. The total examination time was approximately 10 min. After the procedure, patients were monitored for at least 30 min to exclude any contrast-related adverse reactions.

A dedicated breast radiologist at each participating centre interpreted all CEM examinations. Each reader had at least 5 years of experience in CEM interpretation and at least 10 years in mammographic imaging. Readings were not blinded, as the study was prospective and integrated into the routine clinical workflow. All CEM studies were evaluated in accordance with the standards outlined in the 2022 supplement to the Breast Imaging Reporting and Data System (BI-RADS) lexicon published by the American College of Radiology (ACR), for both low-energy and recombined images [13], [14].

Preoperative estimation of the total tumour extent was performed for each imaging modality. The reported extent represented the largest measurable dimension of the tumour. In multifocal disease, the total extent was defined as the sum of the largest diameter of each lesion plus the distances between them. These measurements, together with the tumour’s location, were used by the surgeon for operative planning.

2.3. Surgery and histopathologic reference standard

For this study, a change in surgical management for multifocal breast cancer was defined as the identification of any additional malignant focus that rendered breast-conserving surgery technically unfeasible, thereby requiring escalation to a more extensive procedure. In cases of multicentric disease, the detection of additional lesions invariably resulted in modification of the surgical plan, either to mastectomy or to a broader, multi-quadrant breast-conserving excision after placement of additional localisation needles when deemed feasible for the individual patient.

Surgical specimens were processed in one of two pathology departments affiliated with the participating centres, following standard protocols for fixation, embedding, sectioning, staining, and immunohistochemistry. At each centre, histopathological examinations were performed by three pathologists with at least 5 years of experience in breast pathology diagnostics. Multifocal (MF) breast cancer denotes the presence of two or more separate tumour foci situated inside the same quadrant or section of the breast, generally within 2–5 cm of one another. On the other hand, multicentric (MC) breast cancer comprises numerous tumours situated in distinct quadrants or regions of the breast, typically separated by more than 5 cm or located within different breast lobes. To standardise calculations, these lesions were included in a single MFMCC group. All additional neoplastic foci identified on histopathology were characterised with respect to tumour size, histologic type, and grade. Both invasive carcinomas and ductal carcinomas in situ were included in the analysis.

2.4. Outcomes

The primary outcomes include the diagnostic performance of CEM-RI compared to CEM-LEI in pre-surgical staging, and the effect of these images on surgical planning across the study population. Additional outcomes included their ability to detect previously unrecognised malignant foci. Subgroup analyses were performed by patient age, mammographic breast density, tumour subtype, and microcalcification status to evaluate whether these factors influenced the assessment. Changes in surgical management attributable to CEM findings were defined as any modification of the initial surgical plan, including conversion from partial mastectomy to mastectomy or vice versa, as well as unilateral-to-bilateral surgery prompted by the detection of contralateral malignancy.

2.5. Statistical analysis

The analysis included 184 patients who met the inclusion criteria. The histopathological diagnosis of MFMCC served as the “gold standard” and was compared with the suspicion based on CEM-LEI and CEM-RI. Normality of distribution was evaluated using the Kolmogorov–Smirnov test. Continuous variables were summarised as the arithmetic mean with standard deviation for normally distributed data, or as the median with interquartile range for non-normally distributed data. A quantitative summary was also prepared for participants’ menopausal status, histopathological cancer types, cancer staging, breast morphology, and lesion morphology observed in CEM-LEI and CEM-RI.

The number of lesions suspected of malignancy during the diagnostic process was correlated with the number of neoplastic lesions confirmed histopathologically using Kendall’s tau coefficient. In turn, lesion size and the diagnosis of MFMCC were correlated using the Pearson correlation coefficient. Statistical significance was set at p = 0.05. In the subsequent stage of the statistical analysis, confusion matrices were constructed to evaluate MFMCC detection in CEM compared with HP. Sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) were calculated separately for CEM-LEI and CEM-RI. The 95% confidence intervals for sensitivity and specificity were determined using the Bernoulli method.

All analyses were performed using an Excel spreadsheet and Python 3.13.5 (SciPy, NumPy, Pandas, Matplotlib, and Seaborn libraries).

3. Results

3.1. Patients and lesions characteristics

A total of 184 patients were enrolled in the study. The median age at enrolment was 65 years (range, 36–84 years). The study population comprised 182 (98.92%) women and 2 (1.08%) men. Among female patients, 137 (74.45%) were postmenopausal, 40 (21.62%) reported a family history of breast cancer, and 6 (3.24%) were identified as carriers of breast cancer–associated gene mutations.

Regarding histopathological characteristics, the most frequent histological type was invasive ductal carcinoma (IDC), with or without associated ductal carcinoma in situ (DCIS), observed in 124 (93.80%) lesions. The second most common type was invasive lobular carcinoma (ILC), identified in 19 (10.27%) cases.

On CEM-LEI, the most common breast composition was fatty–glandular tissue (ACR category B), noted in 74 (40.00%) patients, followed by predominantly fatty breasts (ACR category A) in 46 (25.95%) patients. On CEM-RI, the most prevalent background parenchymal enhancement (BPE) pattern was minimal, observed in 119 (64.32%) cases, followed by mild enhancement in 34 (18.38%) cases. The most frequent morphological lesion type on CEM-LEI was a mass, identified in 105 (56.76%) cases. On CEM-RI, the most common enhancement pattern was mass enhancement, observed in 154 (83.24%) cases. Detailed patient characteristics, including histological tumour subtypes and imaging features of the lesions, are presented in Table 1.

Table 1.

Patients’ characteristics.

Characteristics	Value (n = 184)
Sex
Female Male	182 (98.91%) 2 (1.08)
Age median (y)	65 (age range 36 – 84)
Menopausal status
Premenopausal Postmenopausal	47 (25.82% of women) 135 (74.17% of women)
BC family history	39 (21.43%)
Mutations	6 (3.26%)
Histology
IDC ILC DCIS Other	124 (67.03%) 19 (10.27%) 22 (11.89%) 19 (10.27%)
Grade
1 2 3	74 (40.22%) 87 (47.28%) 23 (12.50%)
T category
Tis T1 T2 T3	21 (11.41%) 97 (52.72%) 62 (33.69%) 4 (2.17%)
N category N0 N1 N2 N3	156 (84.78%) 22 (11.96%) 5 (2.72%) 1 (0.54%)
ACR density
A B C D	48 (26.09%) 73 (39.67%) 45 (24.46%) 18 (9.78%)
BPE
Minimal Mild Moderate Marked	118 (64.13%) 34 (18.48%) 20 (10.86%) 12 (6.52%)
Index lesion morphology – CEM-LEI
Mass Asymmetry Architectural distortion Microcalcifications Mass + microcalcifications Asymmetry + microcalcifications	105 (57.07%) 32 (17.39%) 23 (12.50%) 11 (5.98%) 10 (5.43%) 3 (1.63%)
Index lesion morphology – CEM-RI
Mass enhancement Non-mass enhancement Enhancing asymmetry	153 (83.15%) 26 (14.13%) 5 (2.72%)
Index lesion measure (mm)
CEM-LEI CEM-RI HP	19.36 ± 11.54 20.78 ± 14.04 19.42 ± 12.78

Abbreviations: BC – breast cancer; ACR – American College of Radiology; BPE – background parenchymal enhancement; CEM-LEI – contrast enhanced mammography low energy images; CEM-RI – contrast enhanced mammography recombinant images.

3.2. CEM diagnostic performance in MFMCC assessment

All index lesions were visualised on CEM-RI, and 178 (96.7%) on CEM-LEI. The mean index lesion size on HP was 19.42 ± 12.78 mm; on CEM-LEI, it was 19.36 ± 11.54 mm, and on CEM-RI, it was 20.78 ± 14.04 mm. The lesion size assessed on CEM-RI showed a slightly stronger correlation with the histopathological measurement than CEM-LEI (Pearson’s r = 0.66, p < 0.001 vs. r = 0.61, p < 0.001, respectively).

In the study population, histopathological examination confirmed the presence of at least one MFMCC in 30 cases. CEM-LEI detected 16 MFMCC cases (53.33%), whereas CEM-RI detected 26 cases (86.67%). The agreement with histopathological findings, expressed by the Phi coefficient, was higher for CEM-RI than for CEM-LEI (Φ = 0.71, p < 0.001 vs. Φ = 0.43, p < 0.001, respectively). Detailed data, presented as confusion matrices, regarding MFMCC detectability for both image types are shown in Fig. 2. Among the four lesions not visualised on CEM-RI, two lesions were located outside the mammographic field of view on both CEM-RI and CEM-LEI, while two lesions presented as microcalcifications and were visualised on CEM-LEI. Three of the undetected foci were invasive carcinoma NST, and one was DCIS.

Fig. 3.

A. Contrast-enhanced mammography low-energy images (CEM-LEI): the white arrow indicates an irregular mass in the outer quadrants of the right breast. B. Contrast-enhanced mammography recombinant images (CEM-RI): the white arrow shows heterogeneous contrast enhancement at the site of the lesion seen on CEM-LEI. The yellow arrows indicate two additional foci of contrast enhancement. All lesions were confirmed as invasive ductal carcinoma on histopathological examination. C. Invasive ductal carcinoma (low grade), cribriform type, present in isolated ducts scattered within the fatty tissue.

Fig. 2.

Confusion matrices illustrating the detectability of multifocal or multicentric carcinoma (MFMCC) for both image types. (A) Diagnostic performance of low-energy contrast-enhanced mammography images (CEM-LEI). (B) Diagnostic performance of recombined contrast-enhanced mammography images (CEM-RI).

The sensitivity of CEM-LEI for the detection of MFMCC was 53.33%, specificity was 90.26%, PPV was 51.61%, and NPV was 90.85%. For CEM-RI, sensitivity was 86.67%, specificity was 91.56%, PPV was 66.67%, and NPV was 97.24%. Detailed data regarding the diagnostic performance of both image types for entire cohort, as well as stratified by breast density categories and BPE are presented in Table 2.

Table 2.

Diagnostic performance of both image types.

Modality		CEM-LEI		CEM-RI
	All (n = 184)	ACR A + B (n = 121)	ACR C + D (n = 63)	All (n = 184)	ACR A + B (n = 121)	ACR C + D (n = 63)	BPE minimal + mild (n = 152)	BPE moderate + marked (n = 32)
Accuracy	84.24 ± 2.69%	85.12 ± 2.62%	82.54 ± 2.80%	90.76 ± 2.13%	89.26 ± 2.28%	93.65 ± 1.80%	90.13 ± 2.20%	93.75 ± 1.78%
Sensitivity	53.33 ± 3.68%	63.16 ± 3.56%	36.36 ± 3.55%	86.67 ± 2.51%	78.95 ± 3.01%	100.00%	81.82 ± 2.84%	100%
Specificity	90.26 ± 2.19%	89.22 ± 2.29%	92.31 ± 1.96%	91.56 ± 2.05%	91.18 ± 2.09%	92.31 ± 1.96%	91.54 ± 2.05%	91.67 ± 2.04%
NPV	90.85 ± 2.13%	92.86 ± 1.90%	87.27 ± 2.78%	97.24 ± 1.21%	95.88 ± 1.47%	100%	96.75 ± 1.31%	100%
PPV	51.61 ± 3.68%	52.17 ± 3.68%	50.00 ± 3.69%	66.67 ± 3.48%	62.50 ± 3.57%	73.33 ± 3.26%	62.07 ± 3.58%	80 ± 2.95%

Abbreviations: CEM-LEI – contrast enhanced mammography low energy images; CEM-RI – contrast enhanced mammography recombinant images; ACR – American College of Radiology; BPE – background parenchymal enhancement; NPV – negative predictive value; PPV – positive predictive value.

3.3. Changes in surgery

Among the included patients, 129 (69.75%) underwent breast-conserving surgery, 32 (17.30%) underwent total mastectomy, and 23 (12.43%) underwent subcutaneous mastectomy with either immediate or delayed implant-based reconstruction.

In 23 cases (12.43%), the extent of surgery was modified during the BCU meeting based on CEM-RI findings. In 15 cases (65.22%), a more extensive surgical procedure was indicated. In 6 cases (26.09%), CEM-RI enabled visualization of the index lesion that had not been detected by other imaging modalities. In 2 cases (8.70%), the extent of surgery was reduced based on CEM-RI findings. Detailed data regarding changes in surgical management are presented in Table 3.

Table 3.

Changes in surgical approach.

Changes in surgery	n = 23 (%)
BCT to subcutaneous mastectomy	8 (34.78%)
BCT to total mastectomy	7 (30.43%)
No surgery (lesion visualisation) to BCT	4 (17.39%)
No surgery (lesion visualisation) to total mastectomy	2 (8.70%)
Subcutaneous mastectomy to BCT	2 (8.70%)

Abbreviations: BCT – breast conserving surgery.

3.4. 12-months follow-up

All 184 patients completed the 12-month follow-up. In 7 patients (3.80%), histopathological examination revealed microscopically positive margins (R1), and reoperation was performed within a maximum of one month after the initial surgical procedure. All lesions were invasive carcinoma of NST, and all patients underwent BCT. In 3 patients (1.63%) with R0 resection, isolated DCIS cells < 1 mm were identified; after completion of radiotherapy, a surveillance strategy was adopted. At the 12-month follow-up, no evidence of disease recurrence was observed in these cases.

Local recurrence at the surgical bed was diagnosed in 2 patients (1.09%) within 12 months after surgery - one patient had previously undergone breast-conserving therapy (BCT) and the other mastectomy. Additionally, nodal recurrence was detected in 2 patients (1.09%) at 12 months.

Regarding postoperative complications within the first three months after surgery, 6 patients (3.26%) developed a large, persistent hematoma at the surgical site requiring evacuation. Delayed wound healing was observed in 4 patients (2.17%). One patient (0.54%) developed a surgical site infection with abscess formation requiring surgical revision and antibiotic therapy. In one patient (0.54%), bronchospasm occurred during anaesthesia, necessitating discontinuation of anaesthesia and postponement of the procedure until the following day. One patient (0.54%) developed shock on the second postoperative day following administration of metamizole. No distant complications were observed during the 12-month follow-up period.

4. Discussion

In this prospective multicentre study, we compared the diagnostic performance of CEM-LEI, corresponding to standard mammographic images that assess morphological characteristics of breast lesions, with that of CEM-RI, which provides a functional evaluation of intralesional contrast enhancement, for the detection of MFMCC. Our findings demonstrate that CEM-RI achieved significantly higher sensitivity than CEM-LEI (86.67% vs 53.33%), highlighting the added value of contrast-enhanced functional imaging in assessing lesion multiplicity. These results are consistent with our previous single-centre retrospective analysis of 999 patients, in which CEM-RI also showed superior sensitivity compared with CEM-LEI (87.63% vs. 24.74%) [15]. The primary aim of the present study was to evaluate the added value of CEM-RI over standard mammography in preoperative staging. However, it should be emphasised that CEM interpretation inherently includes assessment of both CEM-LEI and CEM-RI images, which constitute integral components of a single diagnostic examination. Notably, CEM-RI failed to detect two MFMCC lesions presenting as microcalcifications, which were visible on CEM-LEI. When both image types are considered jointly (in accordance with the BI-RADS lexicon), the overall sensitivity of the CEM examination increases to 93.33% [16]. The lack of enhancement of lesions presenting as microcalcifications on CEM-RI is consistent with the literature, which indicates that malignant microcalcifications may or may not demonstrate contrast enhancement, and that the sensitivity of CEM-RI for malignant microcalcifications is approximately 71.4% [17], [18]. The two missed MFMCC lesions were located outside the field of view of CEM and were identified in the subclavicular region. These lesions were detected by conventional ultrasound, which is routinely performed in all patients for complementary breast assessment and evaluation of axillary, subclavicular, and supraclavicular lymph nodes, emphasizing the importance of a multidisciplinary and multimodal evaluation of patients with breast cancer [19].

Our findings are consistent with those reported by other authors, particularly the study by Bicchierai et al., in which an analysis of 991 patients demonstrated a CEM sensitivity of 97.7% for the detection of additional foci, with only five lesions missed [20]. In other retrospective studies, CEM sensitivity for detecting MFMCC has varied widely, ranging from 16.7% to 85.29% [21], [22], [23], [24], [25]. Such variability is primarily due to differences in the number of patients included and the relatively small cohort sizes. The ongoing PROCEM trial [NCT04437602], which is investigating the added value of CEM in preoperative breast cancer staging, holds promise for clarifying the current evidence, and its results are expected soon [26].

In our study, both CEM-LEI and CEM-RI demonstrated high specificity for detecting MFMCC, with a slight advantage for CEM-RI (91.56% vs 90.26%, respectively). These results are consistent with those reported by other authors, in which CEM also achieved high specificity for detecting additional tumour foci, ranging from 91.2% to 94.9% [20], [22]. However, these findings contrast with the overall specificity of CEM reported in broader patient populations, where meta-analyses have shown substantially lower specificity, not exceeding 80% [10], [27], [28]. This discrepancy may be attributed to differences in the study populations; in our study, a selected cohort of patients with a primary diagnosis of breast cancer was evaluated, and additional foci showing contrast enhancement are highly suspicious for MFMCC.

Additionally, we analysed the impact of breast density and BPE on the diagnostic performance of CEM in detecting MFMCC. For CEM-LEI, sensitivity in breasts with density categories ACR A + B was 63.16%, and it decreased substantially in breasts classified as ACR C + D, where sensitivity was 36.36%. In contrast, for CEM-RI, sensitivity in ACR A + B breasts was 78.95%, whereas in ACR C + D breasts it reached 100%, as all four undetected cases were within the ACR A + B group. These findings confirm the well-established consensus that standard mammographic images, corresponding to CEM-LEI, demonstrate a significant decline in diagnostic sensitivity with increasing fibroglandular tissue density, which also translates into lower detectability of MFMCC in this subgroup of patients. Conversely, CEM-RI demonstrated high sensitivity for detecting malignant lesions and additional tumour foci, regardless of mammographic breast density [6], [29], [30]. For CEM-RI, BPE may represent a potential confounding factor. However, in our study cohort, sensitivity for minimal and mild BPE was 81.82%, compared with 100% for moderate and marked BPE. This difference reflects the unequal distribution of cases between the groups (152 vs 32), with all undetected cases occurring in the minimal + mild BPE group. Overall, this analysis indicates that CEM-RI maintains high diagnostic performance irrespective of the degree of background parenchymal enhancement.

Based on CEM-RI findings, the BCU modified the planned surgical approach in 23 cases (12.43%). In six cases (26.09%), CEM-RI enabled visualisation of the index lesion within fibroglandular tissue that other imaging modalities had not identified. In 11 cases (48.82%), a more extensive surgical procedure was indicated due to the presence of MFMCC, and in 4 cases (17.39%), the decision was driven by a larger lesion size than that observed with other diagnostic methods. In two patients (8.70%), the extent of surgery was reduced. At the 12-month follow-up, no residual disease or recurrence was observed in any of these patients. Postoperative complications occurred in three patients (13.04%), consisting of hematoma or delayed wound healing. In our previous retrospective analysis, we observed a slightly higher decision change rate of 19.90%, which is primarily attributable to differences in cohort size [31]. In other published studies, the decision change rate has ranged from 20.00% to 25.40% [20], [32], [33], [34].

Our study demonstrated a relatively low reoperation rate following R1 resection after BCT, observed in only 7 patients (3.80%). In the literature, reoperation rates after BCT are considerably higher, ranging from 10% to 30% in the general population. Reported rates vary by histological subtype, ranging from approximately 18% for invasive carcinoma NST to up to 35% for invasive lobular carcinoma [35], [36], [37], [38]. We attribute the low reoperation rate in our cohort primarily to the use of CEM in combination with adjunctive ultrasound for evaluating regional lymph nodes and potential lesions outside the mammographic field of view, as well as to the multidisciplinary team's experience in interpreting both conventional mammographic and CEM images. In recent years, CEM has proven to be a valid, feasible alternative to CE-MRI for presurgical staging of BC, with diagnostic performance comparable to CE-MRI and a lower false-positive rate in retrospective studies [10], [39], [40], [41].

This study has several limitations. Although conducted prospectively and across multiple centres, the study cohort was relatively small, comprising only 184 patients. The limited sample size, particularly in subgroup analyses stratified by breast density and background parenchymal enhancement (BPE), may have affected the stability and precision of the diagnostic performance estimates. In addition, the unequal distribution of cases between subgroups may have influenced sensitivity results, particularly for CEM-RI in the ACR C + D and moderate + marked BPE categories, where no false-negative cases were observed. Moreover, both participating institutions were tertiary referral centres. While this setting ensures a high level of diagnostic expertise and standardised multidisciplinary management, it may reduce the generalizability of the findings to institutions with different levels of experience or technical resources. Finally, the follow-up period was limited to 12 months. Although no residual disease or distant complications were observed during this interval, longer follow-up is necessary to adequately evaluate long-term oncologic outcomes and the potential impact of CEM-guided surgical decision-making on recurrence rates.

5. Conclusions

CEM-RI demonstrated significantly higher sensitivity than CEM-LEI for the detection of MFMCC, while maintaining high specificity. The combined assessment of low-energy and recombined images further increased the overall diagnostic sensitivity of CEM, underscoring the complementary value of both components. CEM-RI contributed to clinically relevant modifications of surgical management in a substantial proportion of patients, without increasing short-term recurrence rates. Thanks to the use of CEM in the preoperative staging of breast cancer in the studied population, a very low reoperation rate was achieved, suggesting that this method may help reduce the rate of positive surgical margins and decrease the number of women requiring additional procedures. These findings support the role of CEM, particularly the recombined images, as a valuable tool in the preoperative staging of breast cancer.

Bioetics statement

This prospective, two-centre study was approved by the Institutional Review Board and Ethics Committee of Medical University of Silesia (Decision number: BNW/NWN/0052/KB1/93/1/23/24), obtained in January 2024.

Written informed consent was obtained from all participants. All study procedures complied with the ethical principles of the 1964 Declaration of Helsinki and its subsequent amendments.

CRediT authorship contribution statement

Michał Nycz: Writing – review & editing, Writing – original draft, Formal analysis, Data curation, Conceptualization. Iwona Gisterek-Grocholska: Writing – review & editing, Writing – original draft, Visualization, Formal analysis, Data curation. Joanna Szczudło-Chraścina: Writing – original draft, Visualization, Resources, Methodology, Investigation, Data curation. Kamil Krupa: Writing – review & editing, Writing – original draft, Resources, Formal analysis. Anna Grażyńska: Writing – review & editing, Writing – original draft, Validation, Supervision, Resources, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Urszula Szulc: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Katarzyna Steinhof-Radwańska: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Michał Hrapkowicz: Writing – original draft, Resources, Methodology, Formal analysis, Data curation, Conceptualization. Agnieszka Niewiadomska: Writing – review & editing, Writing – original draft, Formal analysis, Data curation. Karolina Snopek-Miśta: Writing – original draft, Methodology, Formal analysis, Data curation, Conceptualization. Agnieszka Bobola: Writing – original draft, Visualization, Validation, Methodology, Formal analysis, Data curation. Anna Barczyk-Gutkowska: Writing – review & editing, Writing – original draft, Supervision. Andrzej Lorek: Writing – review & editing, Writing – original draft, Validation, Supervision, Project administration. Aleksandra Kuźbińska: Writing – original draft, Visualization, Methodology, Data curation, Conceptualization.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.