Comparative analysis of digital mammography and contrast-enhanced mammography in diagnosing suspicious breast calcifications: implications for surgical decision-making

Abstract

Background

Breast calcifications can be an indicator of early-stage breast cancer. While Digital Mammography (DM) remains the gold standard for detecting breast microcalcifications, its ability to differentiate between benign and malignant types based solely on morphology and distribution is limited. Contrast-Enhanced Mammography (CEM) incorporates pathological contrast-enhancement, leading to a more confident diagnosis, and offering a faster and more cost-effective tool than breast magnetic resonance imaging. CEM allows for a single scan assessing both microcalcification morphology using low energy images and lesion vascularity using high energy subtracted images.

Purpose

This study aimed to compare the diagnostic accuracy of CEM and DM in evaluating suspicious breast calcifications, with a focus on characterizing breast masses (benign vs. malignant), delineating tumor extent, and assessing the clinical impact of these imaging modalities on surgical decision-making.

Results

Out of the 54 identified breast lesions, 19/54, (35%) were benign and 35/54, (65%) were malignant. Calculated sensitivity, specificity, positive and negative predictive values and total accuracy of DM were 91.4%, 70.8%, 62.7%, 93.9%, and 78% respectively as compared to 85.7%, 90.7%, 83.3%, 92.2%, and 89% for CEM.

Regarding size estimation, Results showed that size of tumor by CEM was more correlated to that by pathology, correlation coefficient (r = 0.86), than DM and pathology,(r = 0.588). Mean difference in diameter (compared to pathology) for CEM was 1.28 mm, and 2.8 mm for DM, with no effects on surgical decision-making.

Cronbach’s Alpha as a measure of reliability was 0.740 between size measurement by DM and pathology, indicating acceptable reliability. However, it was 0.92 between CEM and pathology, indicating excellent reliability.

Conclusions

CEM exhibits higher specificity than DM in assessing suspicious breast microcalcifications. CEM provided tumor size measurements that correlated more with pathological findings compared to DM. While these measurement discrepancies were minor and did not affect surgical planning, CEM proved impactful in when multicentricity was identified by CEM but overlooked in initial DM evaluations. This was evident in patients with dense breasts or instances of satisfaction of search error, where subtle lesions indicative of multicentricity were initially missed, highlighting CEM’s potential to enhance surgical decision-making in such scenarios.

Background

Breast calcification continues to be a huge challenge in terms of diagnosis and management [1, 2–3]. Nearly 55% of non-palpable breast malignancies manifest as isolated calcifications, while over 80% of in situ breast neoplasms are characterized by microcalcifications as their sole radiographic indicator [4].

Breast calcifications, are categorized as benign or suspicious based on their morphological features and spatial distribution patterns on Digital Mammography (DM). These calcifications are systematically classified using the Breast Imaging Reporting and Data System (BI-RADS), a standardized framework established by the American College of Radiology (ACR) to ensure consistent assessment and reporting of imaging findings [4, 5].

This classification system plays a pivotal role in aiding differential diagnosis and risk stratification to optimize patient management and surgical decision-making [6].

Breast tumors exhibit pronounced neoangiogenesis. This vascular permeability facilitates the extravasation of contrast agents into the tumoral interstitium, resulting in elevated signal intensity and enhancing radiological delineation of malignant lesions [7].

Contrast-Enhanced Mammography (CEM) is a promising modality, combining capabilities to concurrently assess morphological features and contrast enhancement patterns. Studies have demonstrated that CEM exhibits diagnostic accuracy comparable to Magnetic Resonance Imaging (MRI) in breast cancer detection, with some evidence suggesting its potential for superior specificity [8].

CEM offers insights into enhancement patterns linked to breast microcalcifications. However, CEM is not yet a definitive diagnostic tool. Therefore, the development of complementary strategies is essential to address its current constraints and optimize its integration into diagnostic workflows [9].

Evaluating disease extent in patients with ductal carcinoma in situ (DCIS) or non-palpable invasive breast cancer remains a clinically complex endeavor. CEM synergizes the diagnostic strength of DM particularly its sensitivity in detecting early malignancies manifesting as microcalcifications with advanced functional insights into lesion morphology, spatial distribution, and vascular characteristics. These integrated capabilities enhance preoperative assessment, offering critical data to guide surgical planning and optimize therapeutic interventions [10].

This study aimed to compare the diagnostic accuracy of CEM and DM in evaluating suspicious breast calcifications, with a focus on characterizing breast masses (benign vs. malignant), delineating tumor extent, and assessing the clinical impact of these imaging modalities on surgical decision-making.

Methods

This study was a cross-sectional prospective study. It was conducted from March 2023 to March 2024. The relevant ethics committee approved the study.

Subjects

Fifty adult female patients were included, with 54 identified breast lesions.

Inclusion criteria:

• Patients with suspicious breast calcifications, BIRADS 4B or 4C, without an associated breast mass, identified during DM National Annual Screening Program.

• Patients with a breast mass associated with suspicious calcifications (BIRADS 4 or 5) identified by DM (Fig. 1).

[See PDF for image]

Fig. 1

Flow chart demonstrating inclusion and exclusion criteria in our study

Exclusion criteria:

• Patients with contraindications to contrast media injection such as; renal dysfunction, pregnancy, and allergies to contrast medium (Fig. 1).

Control:

Selected patients who had unilateral breast lesions had their contralateral free breast side used as the control for our study (46 patients had unilateral breast lesions, so we had 46 controls).

Sampling method:

Fifty eligible patients were selected using a convenience sampling method. They all provided written informed consent before undergoing examinations requested in our study.

CEM procedure

A certified technician performed CEM on a Senographe Pristina mammography system (GE Healthcare) following these steps:

• An intravenous (IV) line was placed in the patient’s arm.

• A non-ionic iodine contrast agent (Ultravist or Omnipaque 300) was injected at a dose of 1.5 ml/kg body weight.

• High-energy (HE) and low-energy (LE) mammography images were acquired before and after contrast administration.

• Subtraction images, in the standard positions craniocaudal (CC) and mediolateral oblique (MLO) views were obtained with compression so that the breast parenchyma becomes less visible, to isolate the iodine signal, highlighting areas of blood vessel growth (angiogenesis).

• The entire procedure took approximately 7–10 min.

Image analysis and interpretation

Two radiologists independently assessed the DM and CEM images, the first with 10 years of experience and the second with 15 years of experience (when both radiologists disagreed, we took the opinion of the senior radiologist), for:

• Lesion size, location, shape, margins, and density and calcification characteristics on DM, which include distribution and morphology were analyzed and categorized as suspicious or non-suspicious according to ACR BIRADS Lexicon 2013 [4].

  Inter-reader agreement between both radiologists was calculated in the form of ‘Cohen’s kappa’: 0.94, ‘prevalent adjusted bias-adjusted kappa’ (PABAK): 0.94 and ‘observed agreement’: 97%. This indicates ‘Almost Perfect Agreement’: Cohen’s Kappa (κ ≈ 0.94) and PABAK both indicate ‘Excellent Reliability’ (per Landis and Koch: κ > 0.8 = Almost Perfect).

• Enhancement patterns were evaluated for CEM images and were analyzed and categorized according to ACR BIRADS for Contrast Enhanced Mammography (CEM) 2022 [11].

  Inter-reader agreement between both radiologists was calculated in the form of ‘Cohen’s Kappa’: 0.957, ‘PABAK’: 0.963. This indicates ‘Almost Perfect Agreement’ (per Landis and Koch criteria: κ > 0.8 = Almost Perfect) and ‘High Consistency’ (Both readers agreed on 53/54 lesions, 98.15%, reflecting ‘Strong Reliability’).

• The contralateral sides of the breast were used as control if diagnosed as free of suspicious lesions.

Histopathological correlation

Histopathological results served as the gold standard for this study. Patients with BIRADS IV or V lesions, by DM and/or CEM, had their results compared to histopathological reports.

Lesions were biopsied under ultrasound (US) guidance if the lesion was visualized by US (15 lesions). If the lesion was only seen by DM, stereotactic biopsy was done using true cut needle biopsy (16 lesions), via 14 -18-gauge needles, or referred for surgical excision (23 lesions).

For patients with suspicious breast calcifications, which turned out to be benign lesions by biopsy (19 lesions), or patients with malignant lesions who were referred to the oncology department for neoadjuvant chemotherapy (12 lesions); the histopathological results were based on core needle biopsies.

For patients with malignant lesions that were elected for surgery without prior neoadjuvant chemotherapy (23 lesions); the final surgical specimens were used.

Statistical analysis

Sensitivity, specificity, positive and negative predictive values (PPV, NPV), and total accuracy were estimated by taking the probability of malignancy; micro-calcific clusters as detected by DM were considered as positive test results. Also, mass or non-mass enhancement (NME) detected by CEM were considered as positive test results.

The diagnostic indices ± 95% confidence interval (CI) were calculated for DM and CEM alone and when combining them together. P-value was considered significant if ≤ 0.05.

Wilcox-signed rank test was used to compare largest diameter by both DM and CEM in comparison to histopathology. Non parametric correlation analysis using spearman Rho test was used to test association of tumor size by different measures (DM, CEM and histopathology).

All statistical calculations were done using computer programs Microsoft Excel 2007 (Microsoft Corporation, NY, USA) and SPSS version 27.0, Statistical Package for the Social Science (Chicago, ILI).

Results

This study included 50 patients. None of them had adverse reactions to injected contrast media. The ages ranged from 30 to 76 years with a mean age of 53.

All patients had positive lesions on one side of the breast by DM. Four patients out of 50 had bilateral breast lesions. So we considered to have 54 malignant lesions and 46 controls.

Furthermore, three cases exhibited multiple lesions in the ipsilateral breast with contrast enhancement on CEM, all of which were later confirmed as malignant by pathological examination. Upon retrospective review of the initial DM scans, suspicious microcalcifications that had been overlooked in the initial mammographic assessment were identified. These omissions were attributed to several factors, including satisfaction of search error, where subtle lesions indicative of multicentricity were missed, potentially impacting surgical decision-making; the presence of dense breast tissue; the small size of the lesions; or their initial misinterpretation as benign-appearing calcifications on DM. These lesions were considered false negative (FN) for DM.

The 50 cases were classified into 4 groups according to the breast density. Out of the 50 cases, 5 (10%) cases were classified as ACR A, 11 (22%) cases were classified as ACR B, 29 (58%) cases were classified as ACR C and 5 (10%) cases were classified as ACR D.

According to the lesion location, we classified the detected 54 lesions into 5 groups and further according to their final diagnosis either by pathological assessment of biopsy and surgical samples (Table 1). Upper outer quadrant lesions (UOQ) were the most prevalent in the “malignant lesions” group (14 out of 35 malignant lesions).

Table 1. Distribution of benign and malignant groups within the studied population according to the site of their suspected lesions

Site of lesion	Total no	Benign	Malignant
UOQ	21	7	14
UIQ	2	0	2
LOQ	2	0	2
LIQ	6	3	3
CENTRAL	14	8	6
MULTICENTERIC	9	1	8

According to the final histopathological diagnosis, out of 54 lesions, 35 lesions turned out to be malignant. Invasive duct carcinoma (IDC) was the most common final histopathological diagnosis in the “malignant lesions” group (65.5%). DCIS alone was found only in 5.7%, while both DCIS and IDC were found combined in about 11.4%. Mucinous adenocarcinoma was the least common, with only one case detected in our study (Table 2).

Table 2. Final histopathological diagnosis upon the “malignant lesions” group

Final histopathological diagnosis	Number of lesions (total = 35)
DCIS	4
IDC	20
DCIS + IDC	6
ILC	4
Mucinous adenocarcinoma	1

Out of the 54 detected lesions, microcalcifications were isolated presenting findings in only six cases, other findings are listed in (Table 3). These microcalcifications were divided according to their pattern and distribution according to ACR BI-RADS lexicon 2013, details provided in (Table 4).

Table 3. Associated DM findings among studied lesions

DM associated findings	No of lesions (total 54)	%
Isolated suspicious calcifications	6	11.1%
+ Masses	23	42.6%
+ Focal asymmetry	14	24.5%
+ Architectural distortion	10	25.9%
+ Internal mammary lymph nodes	2	3.7%
+ Dilated ducts	2	3.7%
+ Parenchymal edema	1	1.85%
+ Skin thickening	1	1.85%

Table 4. DM findings among studied lesions regarding micro-calcifications form and distribution

Form of calcification	Number of lesions (total 54)
Coarse heterogeneous	11
Amorphous	9
Fine pleomorphic	29
Fine linear branching	5

Distribution of microcalcifications	Number of lesions (total = 54)
Diffuse	2
Regional	5
Grouped	34
Ductal linear	5
Segmental	8

Upon correlating the DM findings to the final diagnoses, 32 lesions were true positives (TP), 19 lesions were false positives (FP), 3 lesions were FN and 46 lesions were true negatives (TN) (Table 5).

Table 5. TP, TN,FN and FP results of DM and CEM

DM	Final diagnosis malignant	Final diagnosis benign	Total
Malignant	32 TP (a)	19 FP(b)	51/51
Benign	3 FN (c)	46 TN (d)	49/49
Total	35 (a + c)	65 (b + d)	100 (a + b + c + d)
CEM	Final Diagnosis Malignant	Final Diagnosis Benign	Total
Malignant	30 TP (a)	6 FP( b)	36/100
Benign	5 FN (c)	59 TN (d)	64/100
Total	35 (a + c)	65 (b + d)	100 (a + b + c + d)

Regarding CEM findings, 36 (66.6%) lesions showed contrast uptake and 18 (33.3%) lesions showed no contrast uptake.

Enhancing lesions were classified into; enhancing mass lesions (13 lesions, 36.1%), NME lesions (17 lesions, 47.2%) and lesions that showed both mass and NME (5 lesions, 16.6%) (Fig. 2).

[See PDF for image]

Fig. 2

A 42-year-old female presented with large right breast lump. A CC and B MLO views of DM, right heterogeneously dense breast showing large mass with partially non circumscribed ill-defined margins located centrally and extending to UOQ “arrow” with associated extensive areas of pleomorphic micro-calcifications with regional distribution seen in both UOQ and lower inner quadrant (LIQ) “arrows”. C and D representing CC and MLO views of CEM showing confluent enhancing masses with the central one showing negative contrast on posterior aspect as it is representing a complex cystic lesion “arrow” with associated UOQ mass enhancement and LIQ regional NME with microcalfications seen exhibiting misregistration artifacts “arrows”. This patient turned to be a multi-centric IDC

Out of the 23 lesions that showed NME pattern; focal NME pattern was most common (8 lesions), followed by regional NME (6 lesions) then comes segmental NME (5 lesions) (Fig. 3), followed by linear, diffuse, combined linear with diffuse and combined focal with diffuse NME each represented by only one lesion.

[See PDF for image]

Fig. 3

A 48-year-old female with known left breast cancer. A CC and B MLO views of DM of left breast showing segmental pleomorphic micro-calcifications in UOQ. C and D representing CC and MLO views of CEM showing segmental clumped enhancement. This patient upon biopsy had combined DCIS and IDC

Upon correlation with pathology results or follow up, 72.2% of lesions that showed no contrast uptake were benign. While, 37.8% of lesions turned out malignant even though they had no contrast uptake on CEM (Fig. 4).

[See PDF for image]

Fig. 4

A 45-year-old female had left-sided US guided biopsy from suspicious breast lesion that turned to be IDC. A MLO and B CC views on DM of left dense breast showing extensive area of pleomorphic micro-calcifications having linear pattern. C Zoomed view at area of calcifications with a clip seen at the posterior extent. D and E representing CC and MLO views of CEM showing no enhancement corresponding to noted calcifications. These calcifications on biopsy showed DCIS

On the contrary, 92.3% of enhancing mass lesions, and 100% of lesions that showed both mass and NME were proven malignant by histopathology.

Regarding the diagnostic indices of DM and CEM performance in assessment of micro-calcifications, they are summarized in (Table 6).

Table 6. Results of DM versus CEM in the evaluation of patients with suspicious breast calcifications

Accuracy measures	Digital mammography		CEM
	(%)	95% confidence interval (CI)	(%)	95% CI
Sensitivity	91.4%	(76.9–98.2)	85.7%	(69.7–95.2)
Specificity	70.8%	(59.7–81.8)	90.7%	(83.7–97.8)
PPV	62.7%	(49.5–76.0)	83.3%	(71.2–95.5)
NPV	93.9%	(83.1–98.7)	92.2%	(82.7–97.4)
Total accuracy	78%	(69.9–86.1)	89.0%	(82.9–95.1)

Combining DM and CEM, our results showed that sensitivity was raised to 100%, specificity was 90.7%, PPV was 83.3% and NPV raised to 100%.

Size correlation by DM versus CEM

Accurate lesion sizing through excision biopsy was obtained for 23 out of the 36 malignant breast lesions detected in our study population.

One of the study limitations was the wide time interval between radiological investigations in the form of DM and CEM and the surgical excision biopsy for the other 13 malignant lesions detected, as those cases were referred to neoadjuvant chemotherapy after histopathological diagnosis. Therefore, those 13 cases were excluded from the size correlation study.

Results showed that the size of the lesion measured by CEM was more correlated to that by pathology, with a correlation coefficient (r) = 0.86. DM results in correlation to pathology were (r) = 0.588, (Figs. 5 and 6).

[See PDF for image]

Fig. 5

Correlation between lesion largest diameter by DM and by pathology

[See PDF for image]

Fig. 6

Correlation between lesion largest diameter by CEM and by pathology

Also, the mean difference between the lesion’s largest diameter assessment by CEM and pathology was 1.28 mm, but DM in correlation to pathology showed a larger mean difference, 2.8 mm (Tables 7 and 8).

Table 7. Comparison of largest diameter (cm) of breast lesions by DM and after surgical dissection

	N	Mean	Standard deviation	Median	Minimum	Maximum	P value
Largest diameter by DM	23	5.517	3.1628	5.000	1.3	13.5	0.487
Largest diameter by pathology	23	5.239	3.1066	5.000	2.0	14.0

P value is significant ≤ 0.05, test statistics used for paired data comparison is Wilcoxon signed rank test

Table 8. Comparison of largest diameter (cm) of breast lesions by CEM and after surgical dissection

	N	Mean	Standard deviation	Median	Minimum	Maximum	P value
Largest diameter by CEM	18	4.900	2.6090	4.500	1.0	9.0	0.777
Largest diameter by pathology	18	5.028	2.6415	4.500	2.0	10.0

P value is significant ≤ 0.05, test statistics used for paired data comparison is Wilcoxon signed rank test

Cronbach’s Alpha as a measure of reliability or repeatability was found to be 0.740 between lesion size measurement by DM and pathology, indicating acceptable reliability. On the other hand, Cronbach’s Alpha was 0.92 between CEM and pathology, indicating excellent reliability.

CEM examination impact on surgical decision for cases with suspicious breast microcalcifications

Regarding the size of the largest diameter of breast lesion by CEM, even though it was more correlated to that by pathology than that between DM and pathology. Yet these minute differences in measurements (less than 4 mms) did not affect the final surgical decision-making.

However, the results of CEM showed that the FP rate (9.3%) is significantly lower than that in DM (29.2%). This means when a surgical decision is to be prompted, it is better to be with CEM guidance.

In addition, 3 cases showed multicentricity detected by CEM but were missed in initial DM reports. This significantly affected the surgical decision-making (upgrading from breast-conserving surgery to total mastectomy) (Fig. 7).

[See PDF for image]

Fig. 7

A 34-year-old female presented with left breast lump. A CC and B MLO views of DM of left breast showing asymmetry seen at inner half of breast on CC view and seen central along the nipple line of MLO view with associated 3 grouped pleomorphic micro-calcifications seen around the lesion “arrows”. C Zoomed view of asymmetry and micro-calcifications. D And E representing CC and MLO views of CEM showing mass enhancing lesion corresponding to area of asymmetry central and extending to upper quadrant and on CC view located on inner part of breast with associated extensive clumped NME with segmental distribution at LIQ. This case had a multi-centric IDC

Discussion

The presence of breast calcifications can be a significant indicator of early-stage breast cancer, including DCIS and even non-palpable invasive forms [12]. While DM remains the gold standard for detecting breast microcalcifications, its ability to differentiate between benign and malignant types based solely on morphology and distribution can be limited. CEM emerges as a promising approach by incorporating pathological contrast enhancement, potentially leading to a more confident diagnosis compared to DM, and offering a faster and potentially more cost-effective alternative to breast MRI [13]. Additionally, CEM allows for a single scan to assess both microcalcification morphology using LE images and lesion vascularity using subtracted HE image [14, 15].

This study aimed to compare the diagnostic accuracy of CEM and DM in evaluating suspicious breast calcifications, with a focus on characterizing breast masses (benign vs. malignant), delineating tumor extent, and assessing the clinical impact of these imaging modalities on surgical decision-making.

In our study, the ages ranged from 30—76 years with a mean age 53, while in a study done by Houben et al. [12] to assess the diagnostic accuracy of CEM in cases with suspicious calcifications, the ages ranged from 49 to 75 years with mean age 60.5 years. On another study by Nicosia et al. [15], also about the role of CEM in microcalcifications assessment, the ages ranged from 45 to 59 years with mean age 51 years.

Out of the 50 cases, 34 (68%) females had dense breast “including ACR C and D” in our study population. This is comparable to Nicosia et al. [15] who found that 78% of their cases had breast density classified as ACR C and D.

As regards lesion location, we found that 38.8% of lesions were in upper outer quadrant (UOQ), which was the most frequent among all other sites. While Nicosia et al. [15] found that about 59.9% of lesions were detected at the UOQ.

We had 65% of lesions diagnosed as “malignant lesions” by final histopathology. While Houben et al. [12], had only 44% of studied lesions with suspicious microcalcifications diagnosed as malignant. Yet Nicosia et al. [15] found that 66% of studied lesions were malignant.

In this study, 65.5% of malignant lesions in our study were diagnosed with IDC. DCIS alone was found only in 5.7%, while DCIS and IDC were both found combined in about 11.4%. Houben et al. [12] had DCIS in (50.4%) of malignant lesions group. They had 27 IDC cases, 2 cases of ILC, only 1 case of invasive mucinous carcinoma, and another case of intraductal papillary carcinoma. However, we had no cases of intraductal papillary carcinoma diagnosed in our study.

Although invasive lobular carcinoma (ILC) typically presents on DM as masses, asymmetries, architectural distortion, or rarely as microcalcifications [16]. We observed four cases of ILC with suspicious breast calcifications.

We found that 42.6% of all lesions presented with associated mass lesions, while only 2.7% of lesions in the study by Nicosia et al. [15] presented with associated masses.

Unlike many similar studies that primarily focused on categorizing microcalcifications as either suspicious or non-suspicious, our research emphasized the specific morphology and distribution patterns. In our study, the predominant morphology of microcalcifications in the lesions was fine pleomorphic, accounting for 53.7% of cases. The most frequent distribution pattern observed was grouped microcalcifications, present in 62.9% of cases.

The diagnostic performance parameters for assessing breast calcifications using DM in our study were as follows: sensitivity of 91.43%, specificity of 70.77%, PPV of 62.75%, NPV of 93.88%, and overall accuracy of 78%. Houben et al. [12] reported similar results for LE images, with a sensitivity of 90.8%, PPV of 54.1%, and NPV of 84.2%. However, their study’s specificity (39%) was significantly lower than that observed in our study.

Houben et al. [12] reported that 76.8% of lesions showed enhancement on CEM. In our study, 66.6% of lesions exhibited enhancement, with the NME pattern being the most common. Louka et al. [17] in a study on diagnosing breast microcalcifications on CEM and histopathological correlation, found that 96.2% of malignant lesions (25 out of 26 cases) showed enhancement on CEM in patients with BI-RADS 4, 5, and 6 mammographic suspicious breast calcifications on DM [17].

CEM demonstrated significantly higher specificity than DM, with 90.7% (CI 83.7–97.8) compared to 70.8% (CI 59.7–81.8), p-value < 0.05, in our study. Houben et al. [12] found no significant difference between the specificity of LE and HE images of CEM (39% vs. 36.6%) [12].

We also noted that although CEM’s PPV appeared higher than DM’s (83.3% vs. 62.7%), the overlapping confidence intervals indicated no significant difference. CEM had a significantly lower FP rate (9.3%) compared to DM (29.2%), suggesting CEM is preferable for surgical-decision making. Houben et al. [12], found no significant difference in FP rates between CEM (36.37%) and DM (34%). Zhao et al. [9] explained in his very recent study about breast suspicious microcalcifications on CEM, that benign lesions with blood vessels, infections or inflammation, benign skin lesions, and imaging artefacts, all may cause FP results.

Zhao et al. [9] emphasized that the lack of enhancement on HE images cannot reliably rule out malignancy. Accurate diagnosis requires evaluating both the morphological characteristics and enhancement patterns of suspicious microcalcifications. Although HE imaging alone cannot definitively distinguish between benign and malignant calcifications, malignant lesions are more commonly associated with enhancement than benign ones. Consequently, suspicious calcifications should be carefully assessed using LE images or DM. This aligns with our findings, which identified five FN cases, and therefore, we also recommend further histopathological correlation when DM results are clear cut and unequivocal, even if CEM results are negative.

Sensitivity and NPV were comparable between DM and CEM in our study, aligning with Houben et al. [12]. Total accuracy was higher with CEM, though not significantly, likely due to the small sample size. Louka et al. [17], reported higher sensitivity (96%) for CEM in detecting malignant lesions with suspicious microcalcifications.

Our results are consistent with Cheung et al. [18], who found CEM provided additional enhancement information for accurate cancer diagnosis in lesions with microcalcifications, reporting a sensitivity of 89%, specificity of 87%, PPV of 77%, and NPV of 95% [18]. Cheung et al., [19] also reported a sensitivity of 93–100% and specificity of 63–88%, showing significant improvement over DM.

We also agree with the study by Sherif et al., [10] about the characterization of suspicious breast microcalcification by CEM, who found the specificity of CEM in detecting malignant lesions was 83.3%, with overall diagnostic accuracy of 96.77%. This showed that CEM succeeded in the characterization of breast microcalcifications into benign and malignant, compared to DM alone were all microcalcifications were considered suspicious.

Shetat et al. [13], found CEM valuable for assessing suspicious microcalcifications, with non-mass enhancement indicating high-grade DCIS or invasive components, and enhancement paucity favoring benign or low-grade DCIS.

Ploumen et al. [20] found that CEM enhancement and calcification features help differentiate between invasive breast cancer, DCIS, and benign lesions, with enhancement absence in calcifications mainly associated with low-grade DCIS, consistent with our results.

Also, Cao et al. [1] concluded that the ability of CEM to predict the malignancy status of suspicious breast calcifications were significantly improved by adding enhancing features from HE images, especially by incorporating peri-calcification features.

We evaluated combined results of DM and CEM in the evaluation of suspicious breast calcifications, they showed a sensitivity of 100% with a CI (90–100), specificity of 90.7% with a CI (59.7–77.6), PPV of 83.3% with a CI (52.1–77.6) and NPV 100% with CI (92.3–100). To our knowledge, most previous papers of the same interest did not study diagnostic indices of both LE and HE imaging mammographic techniques combined.

On another aspect, the main objective of pre-operative breast imaging is to assess the extent of the disease after an initial diagnosis. Accurately measuring the tumor size and determining its progression through T staging are essential for planning surgery, and also, whether or not neoadjuvant chemotherapy is involved for breast cancer patients [21].

Specimen radiography is useful for diagnosing suspicious microcalcifications but cannot predict the likelihood of underestimation of IDC. This determination requires microscopic histopathological analysis. To prevent underestimation, it is essential to accurately target the biopsy and obtain invasive tissue for microscopic evaluation. Not all microcalcifications in specimens contain invasive elements. CEM can aid in identifying biopsy sites by highlighting enhancement features such as masses and solid enhancements, thereby improving the diagnosis of invasive disease [18, 19].

One of the main issues we aimed to address in our study was accurate sizing of lesions by CEM. Our findings indicated that the tumor size measured by CEM showed a stronger correlation with pathology, (r) = 0.858 compared to DM, (r) = 0.588. The reliability of size measurements between DM and pathology was acceptable (Cronbach’s Alpha = 0.740), while the reliability between CEM and pathology was excellent (Cronbach’s Alpha = 0.92). These results align with Cheung et al.’s [19] study, which demonstrated that CEM features can accurately predict IDC underestimation.

Our results align with Cheung et al. [22], who found that CEM provides a more accurate assessment of disease extent compared to DM, with mean differences of 0.5 mm for CEM and 4.2 mm for DM.

The mean difference in tumor diameter between CEM and histopathology was 1.28 mm, compared to 2.8 mm between DM and histopathology. However, these differences did not significantly impact surgical decision-making, as surgeons typically ensure an oncologically safe margin of over 4 mm around the calcifications [23].

Similarly, Houben et al. [12] reported that CEM reduces measurement error in disease extent assessment, though it may slightly overestimate the extent. These minor discrepancies did not significantly impact surgical decision-making.

Houben et al. [12] took into account the impact of their findings on surgical decision-making, as this is the most relevant outcome when assessing the extent of the disease. Their study found no significant statistical differences in surgical treatment plans based on LE images or the entire CEM exam. We concur with their conclusions regarding the impact of size estimation on surgical outcomes. However, in our study, CEM detected multicentricity in three cases initially missed by DM, leading to a significant change in surgical decisions, upgrading from breast-conserving surgery to primary mastectomy. This aspect was not considered by Houben et al. [12].

Our study had several limitations;

The sample size was limited, and the patients were not consecutive. The patient population was a selected group recalled from a national screening program, where screening radiologists decide on recalls. Different radiologists might have selected different patients, but this is the standard practice in our program.

CEM was not mandatory before biopsy in current clinical practice; as a result, some patients were hesitant to undergo contrast medium injection. Additionally, the wide time interval between CEM examinations and surgical excision biopsy was a limitation, as 13 cases were selected to receive neoadjuvant chemotherapy after histopathological diagnosis and therefore were excluded from the size correlation study.

Conclusions

Our study concluded that CEM exhibits higher specificity than DM in assessing suspicious breast microcalcifications. Although other accuracy metrics did not significantly differ between the two modalities, CEM provided tumor size measurements that correlated more closely with pathological findings compared to DM. While these measurement discrepancies were minor and did not affect surgical planning, CEM proved impactful in cases where multicentricity was identified by CEM but overlooked in initial DM evaluations. This was particularly evident in patients with dense breasts or instances of satisfaction of search error, where subtle lesions indicative of multicentricity were initially missed, highlighting CEM’s potential to enhance surgical decision-making in such scenarios.

Acknowledgements

Not applicable.

Author contributions

1st author MEAI “writing manuscript, cases review” 2nd author M.ME “collecting cases” 3rd author “helping in monitoring cases and planning for surgical decision” 4th and 5th authors “results review”.

Availability of data and material

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Abbreviations

ACR

American College of Radiology

BIRADS

Breast Imaging Reporting and Data System

CC

Cranio-caudal

CEM

Contrast Enhanced Mammography

CI

Confidence interval

DCIS

Ductal carcinoma insitu

DM

Digital Mammography

FN

False negative

FP

False positive

IDC

Invasive ductal carcinoma

ILC

Invasive lobular carcinoma

IV

Intravenous

HE

High energy

LE

Low energy

LIQ

Lower inner quadrant

LOQ

Lower outer quadrant

MLO

Mediolateral-oblique

MRI

Magnetic Resonance Imaging

NME

Non-mass enhancement

NPV

Negative predictive value

PABAK

Prevalent adjusted bias-adjusted kappa

PPV

Positive predictive value

(r)

Correlation coefficient

SPSS

Statistical Package for the Social Science

TN

True negative

TP

True positive

UIQ

Upper inner quadrant

UOQ

Upper outer quadrant

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Cao, K; Gao, F; Long, R; Zhang, FD; Huang, CC; Cao, M; Yu, YZ; Sun, YS. Peri-lesion regions in differentiating suspicious breast calcification-only lesions specifically on contrast enhanced mammography. J Xray Sci Technol; 2024; 32, pp. 1-4.

2. Covington, MF. Contrast-enhanced mammography implementation, performance, and use for supplemental breast cancer screening. Radiol Clin North Am; 2021; 59, 1 pp. 113-128.

3. Kornecki, A. Current status of contrast enhanced mammography: a comprehensive review. Can Assoc Radiol J; 2022; 73, 1 pp. 141-156.

4. Sickles EA (2013) ACR BI-RADS® Atlas, Breast imaging reporting and data system. American College of Radiology, p 39

5. Kim, EY; Lee, KH; Yun, JS; Park, YL; Park, CH; Jang, SY; Ryu, JM; Lee, SK; Chae, BJ; Lee, JE; Kim, SW. Impact of residual microcalcifications on prognosis after neoadjuvant chemotherapy in breast cancer patients. BMC Womens Health; 2024; 24, 1 187.

6. Spak, DA; Plaxco, JS; Santiago, L; Dryden, MJ; Dogan, BE. BI-RADS® fifth edition: a summary of changes. Diagn Interv Imaging; 2017; 98, 3 pp. 179-190.1:STN:280:DC%2BC1c3gtFektA%3D%3D

7. Ghaderi, KF; Phillips, J; Perry, H; Lotfi, P; Mehta, TS. Contrast-enhanced mammography: current applications and future directions. Radiographics; 2019; 39, 7 pp. 1907-1920.

8. Hua, B; Chen, J; Li, QR; Ge, JD; Yuan, T; Quan, GM. Breast non-mass enhancement lesions on contrast-enhanced mammography: modified breast image reporting and data system classification. Clin Radiol.; 2025; 83, 106807.1:STN:280:DC%2BB1Mbls1KgsA%3D%3D

9. Zhao, X. Breast suspicious microcalcifications on contrast-enhanced mammograms: practice and reflection. Int J Gen Med; 2025; [DOI: https://dx.doi.org/10.2147/IJGM.S494188]

10. Sherif, O; Fakhry, S; Fatma, M; Hagar, SE. Role of contrast-enhanced spectral mammography in characterization of suspicious breast microcalcification. Med J Cairo Univ; 2023; 91, 3 pp. 259-265.

11. Lee CH, Phillips J, Sung JS, Lewin JM, Newell MS (2022) Contrast-enhanced mammography (CEM): a supplement to ACR BI-RADS® mammography 2013. American College of Radiology. Breast Imaging Reporting and Data System (BI-RADS®). https://www.acr.org/Clinical-Resources/Reporting-and-Data-Systems/Bi-Rads

12. Houben, IP; Vanwetswinkel, S; Kalia, V; Thywissen, T; Nelemans, PJ; Heuts, EM; Smidt, ML; Meyer-Baese, A; Wildberger, JE; Lobbes, MB. Contrast-enhanced spectral mammography in the evaluation of breast suspicious calcifications: diagnostic accuracy and impact on surgical management. Acta Radiol; 2019; 60, 9 pp. 1110-1117.

13. Shetat, OM; Moustafa, AF; Zaitoon, S; Fahim, MI; Mohamed, G; Gomaa, MM. Added value of contrast-enhanced spectral mammogram in assessment of suspicious microcalcification and grading of DCIS. Egypt J Radiol Nucl Med; 2021; 52, pp. 1-7.

14. Coffey, K; Jochelson, MS. Contrast-enhanced mammography in breast cancer screening. Eur J Radiol; 2022; 156, 110513.

15. Nicosia, L; Bozzini, AC; Signorelli, G; Palma, S; Pesapane, F; Frassoni, S; Bagnardi, V; Pizzamiglio, M; Farina, M; Trentin, C; Penco, S. Contrast-enhanced spectral mammography in the evaluation of breast microcalcifications: controversies and diagnostic management. Healthcare (Basel); 2023; 11, 4 511.

16. Manning, P; Fazeli, S; Lim, V; Ladd, WA; Eghtedari, M; Chong, A; Rakow-Penner, R; Ojeda-Fournier, H. Invasive lobular carcinoma: a multimodality imaging primer. Radiographics; 2022; 42, 4 pp. E115-E116.

17. Louka, AL; Nassef, HH. Diagnosis of breast microcalcifications with contrast enhanced digital mammography and histopathological correlation. Med J Cairo Univ; 2020; 88, 9 pp. 1561-1569.

18. Cheung, YC; Juan, YH; Lin, YC; Lo, YF; Tsai, HP; Ueng, SH; Chen, SC. Dual-energy contrast-enhanced spectral mammography: enhancement analysis on BI-RADS 4 non-mass microcalcifications in screened women. PLoS ONE; 2016; 11, 9 e0162740.

19. Cheung, YC; Chen, K; Yu, CC; Ueng, SH; Li, CW; Chen, SC. Contrast-enhanced mammographic features of in situ and invasive ductal carcinoma manifesting microcalcifications only: help to predict underestimation?. Cancers (Basel); 2021; 13, 17 4371.1:CAS:528:DC%2BB3MXisFCqs7jL

20. Ploumen, RA; de Mooij, CM; Gommers, S; Keymeulen, KB; Smidt, ML; van Nijnatten, TJ. Imaging findings for response evaluation of ductal carcinoma in situ in breast cancer patients treated with neoadjuvant systemic therapy: a systematic review and meta-analysis. Eur Radiol; 2023; 33, 8 pp. 5423-5435.

21. Youn, I; Choi, S; Choi, YJ; Moon, JH; Park, HJ; Ham, SY; Park, CH; Kim, EY; Kook, SH. Contrast enhanced digital mammography versus magnetic resonance imaging for accurate measurement of the size of breast cancer. Br J Radiol; 2019; 92, 1098 20180929.

22. Cheung, YC; Tsai, HP; Lo, YF; Ueng, SH; Huang, PC; Chen, SC. Clinical utility of dual-energy contrast-enhanced spectral mammography for breast microcalcifications without associated mass: a preliminary analysis. Eur Radiol; 2016; 26, pp. 1082-1089.

23. Lobbes, MB; Vriens, IJ; van Bommel, AC; Nieuwenhuijzen, GA; Smidt, ML; Boersma, LJ; van Dalen, T; Smorenburg, C; Struikmans, H; Siesling, S; Voogd, AC. Breast MRI increases the number of mastectomies for ductal cancers, but decreases them for lobular cancers. Breast Cancer Res Treat; 2017; 162, pp. 353-364.