The value of contrast-enhanced energy-spectrum mammography combined with clinical indicators in detecting breast cancer in Breast Imaging Reporting and Data System (BI-RADS) 4 lesions

Introduction

Breast cancer is the most common malignant tumor in women worldwide, and seriously endangers women’s life and health (1). Early diagnosis is an important measure in the prevention and treatment of breast cancer, and it can effectively reduce the mortality rate and improve the quality of life of patients after treatment. Imaging examination is an important basis for the early diagnosis of breast cancer.

Mammography (MG), which has the advantages of a low-radiation dose and cost-effectiveness, is a commonly used imaging method, and the Breast Imaging Reporting and Data System (BI-RADS) classification standard is also widely used in clinical practice (2). However, in some dense breasts, the diagnostic efficacy of MG is reduced due to small or deep lesions that are poorly demarcated from the surrounding soft tissues (3). Contrast-enhanced spectral mammography (CESM) is a new imaging technique that can not only acquire low-energy images similar to MG, but can also provide information about the blood supply and perfusion of breast lesions, and has shown promising results in the diagnosis of breast cancer, the assessment of the extent of disease, and the evaluation of the efficacy of neoadjuvant chemotherapy (4-8). However, there is a lack of uniformity in the diagnostic criteria of CESM in clinical practice.

The quantitative CESM enhancement score based on the low-energy images has been shown to have gain value in diagnosing breast cancer (9). Lesions determined to have a BI-RADS score of 1–3 on low-energy images have a low probability of malignancy, while those with a BI-RADS score of 5 are highly likely to be malignant. However, while the probability span of lesions with a BI-RADS score of 4 is large, they are also often diagnosed incorrectly in the clinic (10). Therefore, this study sought to evaluate the value of a model that combined the BI-RADS score with the enhancement score and clinical indicators in differentiating between benign and malignant breast lesions in patients with suspected breast cancer with a BI-RADS score of 4 on low-energy images. We present this article in accordance with the TRIPOD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-24-741/rc).

Methods

Study cohort and population

The data of patients who underwent CESM examinations at the Department of Medical Imaging of the Third Affiliated Hospital of Soochow University from January 2018 to July 2023 were retrospectively collected. To be eligible for inclusion in this study, the patients had to meet the following inclusion criteria: (I) be female; (II) have CESM images showing a unilateral breast, single-focal predominant lesion, or bilateral lesions with only one operation or puncture; (III) have a BI-RADS score of 4 based on the CESM low-energy images; (IV) have no history of cancer, radiotherapy, or chemotherapy; (V) have undergone surgery or puncture within 1 month of the examination. Patients were excluded from the study if they met any of the following exclusion criteria: (I) had a previous diagnosis of breast cancer; or (II) followed-up to review efficacy. Ultimately, 170 patients were included in the study. A detailed flow chart of patient enrollment for the study is shown in Figure 1. The Ethics Committee of the Third Affiliated Hospital of Soochow University approved the research protocol [No. 2021 (Teaching) CL037], which complied with the Declaration of Helsinki (as revised in 2013). The requirement of informed consent was waived, as the study was retrospective.

Figure 1 Recruitment flow chart. CESM, contrast-enhanced spectral mammography; BI-RADS, Breast Imaging Reporting and Data System.

CESM image acquisition

The examination was performed using an MG machine (GE Senographe Essential, USA). Before scanning, the patient was given 1.5 mL/kg of a non-ionic iodine contrast agent (iodixanol injection, 50 mL, 13.5 gI) at an injection rate of 3 mL/s through the anterior cubital vein. The CESM examination was initiated 2 minutes after the injection, and the steps of the filming procedure were the cephalo-caudal position on the affected side, cephalo-caudal position on the healthy side, internal and external oblique positions on the healthy side, and internal and external oblique positions on the affected side; the total examination time was limited to 7 minutes. During collection, the compression plate was evenly pressed, pressure was controlled between 11–20 daN, and the degree of compression was limited to the maximum that the patient could tolerate. High-energy and low-energy X-rays were used for each location. The peak tube voltage of the low-energy images was 26–31 kVp, and the peak tube voltage of the high-energy images was 45–50 kVp. The low-energy images and subtraction images were reconstructed through the workstation.

Image analysis

The image analysis was performed on a breast-specific post-processing workstation. The images of each patient were analyzed by two attending physicians who each had more than three years of experience in breast imaging diagnosis. The scoring criteria of the 5th edition of the BI-RADS of the American College of Radiology (revised in 2013) was used, and the scores were based on the morphology of the lesions, edges, density, calcification, and other signs in the low-energy images. In cases of disagreement between the results, a higher-level physician participated in the voting decision. The subtraction images enhancement score was based on the visual analysis of the two physicians, with a total of three points for enhancement. If the physicians provided inconsistent ratings, as mentioned above, a third expert was consulted. A score of 0 indicated negative or no significant enhancement of the lesion (i.e., the enhancement of the lesion and the surrounding tissue was similar); a score of 1 indicated moderate enhancement (i.e., the lesion was moderately more enhanced than the surrounding tissue); and a score of 2 indicated significant enhancement (i.e., the lesion was significantly more enhanced than the surrounding tissue). Typical images with different degrees of enhancement are shown in Figure 2.

Figure 2 Typical images showing the graded enhancement of lesions. (A) 0, negative or no significant enhancement; (B) 1, moderate enhancement; (C) 2, significant enhancement. The white arrows indicate the location of the lesion.

Statistical analysis

All the statistical analyses were performed using R software (version 4.1.0). The normally distributed continuous variables are expressed as the mean ± standard deviation, and the non-normally distributed continuous variables are expressed as the median (interquartile range). Differences between continuous variables among different groups were compared using the unpaired t-test or the Mann-Whitney U-test, and differences between categorical variables were compared using the chi-square test. All the models were constructed using logistic regression. The area under the curve (AUC) of the receiver operator characteristic curve was calculated to assess the discrimination of the model. The Youden index was used to determine the optimal cut-off value. The likelihood ratio test was used for comparisons between models. A decision curve analysis (DCA) was used to calculate the net benefit (NB) of the model, using the following formula: NB = number of true positives − (number of false positives × weighting factor)/number of subjects, weighting factor = threshold probability/(1-threshold probability) (11). At a specific threshold probability, the higher the NB value, the better the model. The internal validation of the model was carried out using bootstrap re-sampling (500 times), and the calibration curve was plotted. The intercept and slope of the calibration curve were used to evaluate the calibration of the model. An intercept of 0 and a slope of 1 indicated that the probability predicted by the model was exactly the same as the actual probability. A P value <0.05 was considered statistically significant.

Results

Pathological findings

A total of 170 patients with 170 lesions were included in this study, of whom 69 (40.6%) had benign lesions and 101 (59.4%) had malignant lesions. Of the 69 benign lesions, adenomas accounted for 26 (37.7%), mammary gland adenopathies for 19 (27.5%), cysts for 6 (8.7%), intraductal papillomas for 6 (8.7%), grade I–II lobular tumors for 7 (10.1%), and inflammatory lesions for 5 (7.2%). Of the 101 malignant lesions, invasive ductal carcinomas accounted for 70 (69.3%), intraductal carcinomas for 14 (13.9%), mucinous carcinomas for 5 (5.0%), papillary carcinomas for 4 (4.0%), lobular carcinomas for 3 (3.0%), ductal carcinomas in situ for 3 (3.0%), chondrosarcomas for 1 (1.0%), and malignant tumors of other mesenchymal tissue origin for 1 (1.0%).

General information

The general data, imaging features, and enhancement characteristics of the malignant and benign groups are compared in Table 1. The age of the patients in the malignant group was higher than that of the patients in the benign group, and there was also a statistically significant difference between the two groups in terms of the distribution of menopause and a family history of breast cancer (P=0.001). There was no significant difference between the malignant and benign groups in terms of the distribution of lesion location (P=0.111). In terms of the distribution of the BI-RADS score and degree of enhancement, the difference in distribution between the two groups was statistically significant (P<0.001).

Development of prediction models

The following three models were developed for the study: Model 1: BI-RADS score; Model 2: BI-RADS score + enhancement score; and Model 3: BI-RADS score + enhancement score + age + menopause + a family history of breast cancer. As Figure 3A shows, the AUCs of Models 1, 2, and 3 were 0.830 [95% confidence interval (CI): 0.769–0.890], 0.858 (95% CI: 0.800–0.915), and 0.900 (95% CI: 0.851–0.949), respectively. The best cut-off value for Model 1 was 0.586, corresponding to a sensitivity of 86.1% and a specificity of 72.5%, that for Model 2 was 0.624, corresponding to a sensitivity of 84.2% and a specificity of 78.3%, and that for Model 3 was 0.766, corresponding to a sensitivity of 74.3% and a specificity of 94.2%.

Figure 3 Comparison of the ROC and decision curves of the three models. (A) Comparison between the ROC curves. (B) Decision curve analysis of three models. Model 1, BI-RADS score; Model 2, BI-RADS score + enhancement score; Model 3, BI-RADS score + enhancement score + age + menopause + a family history of breast cancer. AUC, area under the curve; ROC, receiver operator characteristic curve; BI-RADS, Breast Imaging Reporting and Data System.

Model comparison and internal validation

Comparisons of the AUCs between Models 1, 2, and 3 are shown in Table 2. Compared with Models 1 and 2, the likelihood ratio test chi-square value of Model 3 was 30.980 (P<0.001) and 20.656 (P<0.001), respectively, suggesting that Model 3 was superior to Models 1 and 2. Figure 3B shows a comparison of the DCA curves of the three models; a higher NB was associated with better models at specific threshold probabilities. The internal validation of Model 3 was carried out using bootstrap re-sampling (500 times) with an AUC of 0.880. The calibration curve (intercept: 0.034; slope: 0.807) based on the bootstrap method was used to evaluate the calibration of the model (Figure 4), and the predicted probability of Model 3 was in good agreement with the actual prevalence in the cohort.

Figure 4 Calibration curve for the model that combined the BI-RADS score with the enhancement score and clinical indicators (i.e., Model 3) in the prediction of breast cancer. Model 3, BI-RADS score + enhancement score + age + menopause + a family history of breast cancer. BI-RADS, Breast Imaging Reporting and Data System.

Discussion

In this study, among patients with suspected breast cancer with a BI-RADS score of 4, the AUC of the model that combined the BI-RADS score, enhancement score, and clinical indicator was 0.900, and the corresponding optimal cut-off value was 0.766, with a sensitivity of 74.3%, and a specificity of 94.2%. Moreover, based on the AUC and DCA results, Model 3 was superior to Models 1 and 2. These results indicate that the use of comprehensive imaging features, enhancement degree, and clinical indicators can improve the diagnostic efficiency of benign and malignant breast lesions in patients with a BI-RADS score of 4.

CESM is an emerging MG technology. After an intravenous injection of contrast agent, dual-energy X-ray is used for photography. The peak tube voltage of a low-energy image is 26–31 kVp, which is similar to that of an ordinary MG image (12). The image obtained by logarithmic subtraction can reduce the overlapping effect of breast glands, highlight the uptake area of the iodine contrast agent, and reflect the blood supply of lesions, and thus has great advantages in clinical application (13,14). However, there is a lack of uniformity in the diagnostic criteria of CESM.

In an early attempt to quantitatively evaluate the feasibility of CESM enhancement, Deng et al. (15) showed that the enhancement degree of CESM in breast cancer was significantly higher than that in benign lesions. Jochelson et al. (16) reported that CESM detected 50 of 52 malignant lesions in the breast (sensitivity 96.2%), while conventional MG detected only 42 (sensitivity 80.8%), thus CESM was considered to be more effective in the detection of malignant lesions. The AUC of the combined diagnostic model of the present study was 0.900, which is not particularly high; however, this could be related to the exclusion of patients with more benign lesions from the present study. In addition, the mechanism by which malignant tumor lesions often show significant enhancement on CESM images may be related to factors such as neovascularization, increased blood supply, and increased permeability (17,18).

Liew et al. (19) found that CESM can reduce the biopsy rate of suspicious lesions in the breast, especially for BI-RADS 4A lesions. Among 105 lesions, 22 lesions showed no enhancement, and it was subsequently confirmed that all 22 lesions were benign. Our research, which used a model method, also confirmed that enhancement grading based on BI-RADS scores can improve the diagnostic accuracy of models. Therefore, we believe that comprehensive model-based analyses could prevent some unnecessary biopsies or suggest that some suspicious lesions require active, early follow-up or biopsy, and thus change the clinical treatment strategy to some extent (20); however, more studies need to be conducted to confirm this conclusion.

The pathogenesis of breast cancer is related to many factors. Global data show that the incidence of breast cancer in women aged under 25 years is low, and the incidence of breast cancer in people aged over 25 years tends to increase with age (21,22). In this study, the age of patients in the benign group was 44.93±12.65 years, and that of patients in the malignant group was 53.63±13.08 years, and the age of the patients in the malignant group was significantly higher than that of the patients in the benign group (P<0.001). The proportion of menopausal patients and those with a family history of breast cancer was also significantly higher in the malignant group than the benign group (P<0.001), and the inclusion of age, menopause, and a family history of breast cancer in the prediction model resulted in a diagnostic gain. Due to the limited sample size, the study did not group patients by age or breast type. However, further research is needed to confirm the conclusion in different age and glandular type.

In addition to MG and CESM, breast ultrasound and dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) are commonly used in the differential diagnosis of benign and malignant breast lesions. Breast ultrasound can not only observe the shape, edge features, echo, size and distribution of the calcification of breast lesions, but can also detect the blood flow of lesions (23). New ultrasound techniques, such as elastic imaging and contrast-enhanced ultrasound (24,25), can also improve the accuracy of breast cancer diagnoses; however, the results are closely related to the detection location and the experience of the examining physicians. DCE-MRI can not only show the enhanced characteristics of lesions, but can also generate blood flow-time change curves in selected areas of interest, with high diagnostic accuracy, and is considered the gold standard for breast cancer diagnosis (26,27). However, due to its high cost and long examination time, it is contraindicated for patients with pacemakers and claustrophobia, which limits its clinical application to a certain extent. CESM provides more lesion enhancement information than conventional MG, but it still has some issues (e.g., it requires a small amount of radiation and has decreased diagnostic efficiency in dense breasts). In short, CESM, ultrasound, DCE-MRI, and other imaging technologies complement each other, jointly contributing to the early diagnosis and treatment of breast cancer.

This study had a number of limitations. First, other risk factors for breast cancer, such as race, fertility and lactation status, and hormone use, were not taken into account. Thus, we intend to include more risk factors in subsequent studies. Second, to avoid the duplication of the clinical data of the samples, the study only included patients with a unilateral mammary gland, single lesion as the main lesion, or a bilateral lesion with only one operation or puncture, which might have caused selection bias. In addition, the pathological results of puncture cannot completely replace postoperative pathology results, but they occupy a small part. Finally, the sample size of the study was small, there was no subgroup analysis of different ages and glandular types, and external validation was lacking. In the future, the sample size should be expanded to coordinate the multi-center verification of the study results.

Conclusions

In suspected breast cancer patients with a BI-RADS score of 4, our model that combined the BI-RADS score with the enhancement score and clinical indicators had improves diagnostic efficiency. Our combined model could have great clinical value in the diagnosis of breast cancer.

Acknowledgments

Funding: This research was supported by the China National Natural Science Foundation Youth Program [No. 82001858, Principal Investigator (PI): M.Z.].

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-24-741/rc

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-24-741/coif). M.Z. received funding from the China National Natural Science Foundation Youth Program (No. 82001858). The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The Ethics Committee of the Third Affiliated Hospital of Soochow University approved the study protocol [No. 2021 (Teaching) CL037], and the requirement of informed consent was waived, as the study was retrospective.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Zubair M, Wang S, Ali N. Advanced Approaches to Breast Cancer Classification and Diagnosis. Front Pharmacol 2020;11:632079. [Crossref] [PubMed]
2. Mori M, Akashi-Tanaka S, Suzuki S, Daniels MI, Watanabe C, Hirose M, Nakamura S. Diagnostic accuracy of contrast-enhanced spectral mammography in comparison to conventional full-field digital mammography in a population of women with dense breasts. Breast Cancer 2017;24:104-10. [Crossref] [PubMed]
3. Lee JM, Arao RF, Sprague BL, Kerlikowske K, Lehman CD, Smith RA, Henderson LM, Rauscher GH, Miglioretti DL. Performance of Screening Ultrasonography as an Adjunct to Screening Mammography in Women Across the Spectrum of Breast Cancer Risk. JAMA Intern Med 2019;179:658-67. [Crossref] [PubMed]
4. Richter V, Hatterman V, Preibsch H, Bahrs SD, Hahn M, Nikolaou K, Wiesinger B. Contrast-enhanced spectral mammography in patients with MRI contraindications. Acta Radiol 2018;59:798-805. [Crossref] [PubMed]
5. Xiang W, Rao H, Zhou L. A meta-analysis of contrast-enhanced spectral mammography versus MRI in the diagnosis of breast cancer. Thorac Cancer 2020;11:1423-32. [Crossref] [PubMed]
6. Xu W, Zheng B, Chen W, Wen C, Zeng H, He Z, Qin G, Li Y. Can the delayed phase of quantitative contrast-enhanced mammography improve the diagnostic performance on breast masses? Quant Imaging Med Surg 2021;11:3684-97. [Crossref] [PubMed]
7. Daniaux M, Gruber L, De Zordo T, Geiger-Gritsch S, Amort B, Santner W, Egle D, Baltzer PAT. Preoperative staging by multimodal imaging in newly diagnosed breast cancer: Diagnostic performance of contrast-enhanced spectral mammography compared to conventional mammography, ultrasound, and MRI. Eur J Radiol 2023;163:110838. [Crossref] [PubMed]
8. Wang S, Wang Z, Li R, You C, Mao N, Jiang T, Wang Z, Xie H, Gu Y. Association between quantitative and qualitative image features of contrast-enhanced mammography and molecular subtypes of breast cancer. Quant Imaging Med Surg 2022;12:1270-80. [Crossref] [PubMed]
9. James JJ, Tennant SL. Contrast-enhanced spectral mammography (CESM). Clin Radiol 2018;73:715-23. [Crossref] [PubMed]
10. Meng M, Li H, Zhang M, He G, Wang L, Shen D. Reducing the number of unnecessary biopsies for mammographic BI-RADS 4 lesions through a deep transfer learning method. BMC Med Imaging 2023;23:82. [Crossref] [PubMed]
11. Megna R, Assante R, Zampella E, Gaudieri V, Nappi C, Cuocolo R, Mannarino T, Genova A, Green R, Cantoni V, Acampa W, Petretta M, Cuocolo A. Pretest models for predicting abnormal stress single-photon emission computed tomography myocardial perfusion imaging. J Nucl Cardiol 2021;28:1891-902. [Crossref] [PubMed]
12. Dromain C, Thibault F, Muller S, Rimareix F, Delaloge S, Tardivon A, Balleyguier C. Dual-energy contrast-enhanced digital mammography: initial clinical results. Eur Radiol 2011;21:565-74. [Crossref] [PubMed]
13. Covington MF, Salmon S, Weaver BD, Fajardo LL. State-of-the-art for contrast-enhanced mammography. Br J Radiol 2024;97:695-704. [Crossref] [PubMed]
14. Monticciolo DL, Newell MS, Moy L, Lee CS, Destounis SV. Breast Cancer Screening for Women at Higher-Than-Average Risk: Updated Recommendations From the ACR. J Am Coll Radiol 2023;20:902-14. [Crossref] [PubMed]
15. Deng CY, Juan YH, Cheung YC, Lin YC, Lo YF, Lin G, Chen SC, Ng SH. Quantitative analysis of enhanced malignant and benign lesions on contrast-enhanced spectral mammography. Br J Radiol 2018;91:20170605. [Crossref] [PubMed]
16. Jochelson MS, Dershaw DD, Sung JS, Heerdt AS, Thornton C, Moskowitz CS, Ferrara J, Morris EA. Bilateral contrast-enhanced dual-energy digital mammography: feasibility and comparison with conventional digital mammography and MR imaging in women with known breast carcinoma. Radiology 2013;266:743-51. [Crossref] [PubMed]
17. Viallard C, Larrivée B. Tumor angiogenesis and vascular normalization: alternative therapeutic targets. Angiogenesis 2017;20:409-26. [Crossref] [PubMed]
18. Maiborodin I, Mansurova A, Chernyavskiy A, Romanov A, Voitcitctkii V, Kedrova A, Tarkhov A, Chernyshova A, Krasil'nikov S. Cancer Angiogenesis and Opportunity of Influence on Tumor by Changing Vascularization. J Pers Med 2022;12:327. [Crossref] [PubMed]
19. Liew ALF, Lim HMY, Fok ECM, Loke SC, Tan EY, Chong BK, Lee YS, Chan PMY, Chotai N. Can Contrast-Enhanced Spectral Mammography (CESM) Reduce Benign Breast Biopsy? Breast J 2022;2022:7087408. [Crossref] [PubMed]
20. Tang YC, Cheung YC. Contrast-enhanced mammography-guided biopsy: technique and initial outcomes. Quant Imaging Med Surg 2023;13:5349-54. [Crossref] [PubMed]
21. Renard F, Van Eycken L, Arbyn M. High burden of breast cancer in Belgium: recent trends in incidence (1999-2006) and historical trends in mortality (1954-2006). Arch Public Health 2011;69:2. [Crossref] [PubMed]
22. Luo C, Li N, Lu B, Cai J, Lu M, Zhang Y, Chen H, Dai M. Global and regional trends in incidence and mortality of female breast cancer and associated factors at national level in 2000 to 2019. Chin Med J (Engl) 2022;135:42-51. [Crossref] [PubMed]
23. Liu H, Hou CJ, Tang JL, Sun LT, Lu KF, Liu Y, Du P. Deep learning and ultrasound feature fusion model predicts the malignancy of complex cystic and solid breast nodules with color Doppler images. Sci Rep 2023;13:10500. [Crossref] [PubMed]
24. Wang B, Yang J, Tang YL, Chen YY, Luo J, Cui XW, Dietrich CF, Yi AJ. The value of microvascular Doppler ultrasound technique, qualitative or quantitative shear-wave elastography of breast lesions for predicting axillary nodal burden in patients with breast cancer. Quant Imaging Med Surg 2024;14:408-20. [Crossref] [PubMed]
25. Fu N, Li J, Wang B, Jiang Y, Li S, Niu R, Wang Z. Diagnostic performance of contrast-enhanced ultrasound in breast lesions: what diagnostic models could be used for lesions of different sizes? Gland Surg 2023;12:1654-67. [Crossref] [PubMed]
26. Bartella L, Liberman L, Morris EA, Dershaw DD. Nonpalpable mammographically occult invasive breast cancers detected by MRI. AJR Am J Roentgenol 2006;186:865-70. [Crossref] [PubMed]
27. Li Y, Yang Z, Lv W, Qin Y, Tang C, Yan X, Yin T, Ai T, Xia L. Role of combined clinical-radiomics model based on contrast-enhanced MRI in predicting the malignancy of breast non-mass enhancements without an additional diffusion-weighted imaging sequence. Quant Imaging Med Surg 2023;13:5974-85. [Crossref] [PubMed]