Diagnostic performance of multimodal ultrasonography for molecular subtyping of breast cancer: a retrospective study

Introduction

Breast cancer is the most prevalent malignancy among women worldwide, and its incidence has consistently risen in recent years (1,2). According to the 2024 cancer statistics, approximately 310,000 new cases of breast cancer were diagnosed among women, accounting for 32% of all newly diagnosed female malignancies, surpassing lung and colorectal cancer to rank first in incidence (3). This disease therefore constitutes a substantial threat to women’s health. Breast cancer is characterized by marked heterogeneity. During tumor progression, it is influenced by multiple biological and environmental factors, leading to diverse histomorphological and pathological features and, consequently, distinct molecular biological profiles (4,5). Based on the expression status of certain hormone receptors, including estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 (HER2), breast cancer is commonly classified into several principal molecular subtypes, including luminal A, luminal B, HER2-overexpression, and triple-negative subtypes (6). Significant differences have been observed among these subtypes with respect to tumor progression, therapeutic responsiveness, and prognosis (7,8). In general, the luminal subtypes are associated with relatively indolent biological behavior, whereas the HER2-overexpression and triple-negative subtypes tend to exhibit greater aggressiveness and are associated with poorer clinical outcomes (9-11). Therefore, the early and accurate identification of molecular subtypes in clinical practice is considered essential to guiding individualized therapeutic strategies.

At present, molecular subtyping of breast cancer is mainly determined by histopathological and immunohistochemical assessment of biopsy or surgical specimens. In some patients, multigene expression profiling tests, such as Oncotype DX, MammaPrint, and Prosigna/PAM50, may also be used to provide additional molecular biological information about the tumor (12). However, these methods rely on tissue specimens and may be limited by invasiveness, sampling error, testing time, and limited accessibility. Therefore, magnetic resonance imaging (MRI), particularly dynamic contrast-enhanced MRI, diffusion-weighted imaging, and MRI-based radiomics, has also been explored for the noninvasive prediction of breast cancer molecular subtypes. Nevertheless, its broader clinical application may be limited by relatively high examination costs, strict requirements for standardized imaging acquisition protocols, and insufficient interpretability and generalizability of radiomics models. Ultrasonography, by virtue of its safety, noninvasiveness, and cost-effectiveness, has been widely adopted as a primary modality for the detection and screening of breast cancer (13,14). Commonly utilized ultrasound techniques include conventional two-dimensional grayscale imaging for assessing lesion morphology, margins, and internal echogenicity; color Doppler flow imaging for evaluating tumor vascularity; and shear wave elastography (SWE) for quantifying tissue stiffness.

Different molecular subtypes of breast cancer vary in proliferative activity, angiogenesis, stromal remodeling, and tumor microenvironment characteristics. These biological differences may be reflected in multimodal ultrasonographic features. For example, HER2-overexpression and triple-negative breast cancers usually show higher aggressiveness, faster tumor growth, and more active angiogenesis, which may correspond to richer Doppler blood flow signals and increased vascularization index (VI). In addition, enhanced stromal reaction, extracellular matrix remodeling, and tumor infiltration may increase tissue stiffness, thereby leading to higher SWE parameters such as higher maximum elastic modulus (E_max) and lesion and elasticity ratio between the lesion and surrounding normal tissue (E_ratio). In contrast, luminal subtypes generally show relatively indolent biological behavior and may present with lower vascularity and stiffness. These mechanisms provide a theoretical basis for using multimodal ultrasonography to predict molecular subtypes (15).

In routine clinical settings, multimodal ultrasound data, including two-dimensional grayscale imaging, SWE, and color Doppler flow imaging, are frequently integrated to enhance diagnostic performance. Although multimodal ultrasonography has been widely applied to differentiate benign from malignant breast lesions, few studies have integrated the three ultrasound features into a single combined model for predicting breast cancer molecular subtypes. Based on this, we hypothesized that a multimodal ultrasonographic combined model would achieve higher diagnostic accuracy for breast cancer molecular subtyping compared to single ultrasound modality. To test this hypothesis, the present study systematically evaluated the performance of multimodal ultrasound parameters in distinguishing different breast cancer molecular subtypes, aiming to provide a reference for early, noninvasive assessment of breast cancer molecular subtyping in clinical practice. We present this article in accordance with the STARD reporting checklist (available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0214/rc).

Methods

Study population

Patients with breast cancer who were admitted to our hospital between January 2022 and January 2025 were retrospectively enrolled, and their clinical records were collected and analyzed. The inclusion criteria were as follows: (I) fulfillment of the diagnostic criteria for breast cancer as per the Guidelines for the Diagnosis and Treatment of Breast Cancer (2022 Edition) (16), with molecular subtype confirmed by histopathological assessment; (II) age >18 years; (III) first-time diagnosis; (IV) unilateral disease involvement; (V) no prior administration of relevant therapeutic interventions before admission; (VI) voluntary completion of conventional ultrasonography, color Doppler ultrasonography, and SWE prior to surgery; and (VII) availability of complete clinical data. Meanwhile, the exclusion criteria were as follows: (I) coexistence of other malignant tumors; (II) failure to obtain postoperative molecular subtype results from lesion tissue; (III) presence of severe early-stage dysfunction of major organs, including the heart, lungs, liver, and kidneys; (IV) concomitant inflammatory breast disease; (V) conditions interfering with imaging examinations; and (VI) pregnancy or lactation. According to the empirical principle for sample size estimation in multivariate analysis, the sample size should generally be 5–10 times the number of variables intended for inclusion. In this study, 10 variables were included. After considering an approximately 10% rate of missing or invalid data, the required sample size was estimated to be 55–110 cases. A total of 155 participants were ultimately enrolled in this study, which met the analytical requirements. According to the molecular classification determined by pathological examination, the patients were categorized into HER2-overexpression subtype (n=32), triple-negative subtype (n=31), luminal A subtype (n=26), and luminal B subtype (n=66) groups.

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Nanjing Lishui People’s Hospital (No. 2026KY0409-01) and informed consent was taken from all the patients.

Histopathological and immunohistochemical assessment

The histopathological diagnosis and molecular classification of all patients were determined based on postoperative pathological examination and immunohistochemical findings. Estrogen receptor (ER) and progesterone receptor (PR) status were assessed according to the percentage of positively stained tumor cell nuclei. ER or PR positivity was defined as positive nuclear staining in ≥1% of tumor cells, and the percentages of ER- and PR-positive cells were recorded. HER2 status was determined based on immunohistochemistry (IHC) results and fluorescence in situ hybridization (FISH) testing when necessary. IHC 3+ was defined as HER2-positive, whereas IHC 0 or 1+ was defined as HER2-negative. IHC 2+ was considered equivocal or uncertain HER2 expression and required further FISH testing. Cases showing HER2 gene amplification on FISH were classified as HER2-positive, whereas those without HER2 gene amplification were classified as HER2-negative.

Based on ER, PR, HER2, and Ki-67 expression status, together with the final molecular classification in the pathological reports, breast cancer was categorized into the HER2-overexpression subtype, triple-negative subtype, luminal A subtype, and luminal B subtype. The HER2-overexpression subtype was defined as ER-negative, PR-negative, and HER2-positive. The triple-negative subtype was defined as ER-negative, PR-negative, and HER2-negative. The luminal A subtype was defined as ER- and/or PR-positive, with high PR expression (≥20%), HER2-negative status, and low Ki-67 expression (<14%). The HER2-negative luminal B subtype was defined as ER- and/or PR-positive, HER2-negative, and with either low PR expression (≤20%) or high Ki-67 expression (≥14%). The HER2-positive luminal B subtype was defined as ER- and/or PR-positive, HER2-positive, and with any level of Ki-67 expression. In this study, HER2-negative luminal B and HER2-positive luminal B tumors were collectively classified as the luminal B subtype.

Conventional ultrasonography

Prior to examination, patients were instructed to assume the supine position on the examination bed with full exposure of the breasts. A linear-array probe with a frequency range of 7.5–12 MHz was used. Systematic scanning was performed in both the transverse and longitudinal planes to facilitate examination of all four quadrants of the breast (lower outer, upper outer, upper inner, and lower inner), as well as the adjacent tissues and regional lymph nodes. Relevant ultrasonographic parameters were documented, including tumor morphology, margin characteristics, calcification status, and posterior acoustic features.

Color Doppler ultrasonography

The imaging mode was subsequently switched to color Doppler. Lesion vascularity was evaluated according to the Adler semiquantitative grading criteria (17) (Table 1). To ensure consistency, all grading assessments and measurements were performed by the same ultrasonographer with more than 5 years of relevant clinical experience. Microvascular flow imaging was subsequently applied to further visualize intralesional blood flow. The section exhibiting the most abundant vascular signals was selected, and the lesion boundary was manually delineated. The VI was then automatically calculated by the ultrasound system. Measurements were repeated three times, and the mean value was recorded as the final result.

SWE

The breast examination mode was activated, and the elastography function was subsequently initiated. The transducer was gently placed on the surface of the lesion, and image acquisition was performed after stabilization of the elastographic display. The maximum elastic modulus (E_max), minimum elastic modulus (E_min), mean elastic modulus (E_mean), and the ratio of lesion stiffness to that of surrounding normal tissue (E_ratio) were measured and recorded for further analysis.

Quality control

All ultrasonographic examinations were performed by the same sonographer with more than 5 years of experience in breast ultrasonography. All stored ultrasound images were independently reviewed and assessed by two sonographers with more than 5 years of experience in breast ultrasonography, and both sonographers were blinded to the histopathological results and molecular subtypes of the patients during image interpretation. For qualitative variables, including tumor morphology, margin characteristics, calcification, posterior acoustic features, and Adler blood flow grade, the Kappa test was used to assess interobserver agreement. For quantitative variables, including VI, E_max, E_min, E_mean, and E_ratio, the intraclass correlation coefficient (ICC) was used to evaluate interobserver agreement. If discrepancies occurred between the two sonographers, a consensus was reached through discussion; if disagreement persisted, a senior sonographer with more than 10 years of experience in breast ultrasonography made the final decision.

Outcome measures

Conventional ultrasonographic parameters, color Doppler flow parameters, and SWE parameters were compared between the breast cancer cases with different molecular subtypes.

Statistical analysis

All statistical analyses were conducted with SPSS version 26.0 (IBM Corp., Armonk, NY, USA). Continuous variables conforming to a normal distribution were expressed as the mean ± standard deviation (SD). Comparisons of continuous variables among multiple groups were performed using one-way analysis of variance. When the overall difference was statistically significant, pairwise comparisons were subsequently conducted using Bonferroni correction. Categorical variables were presented as counts and percentages, and intergroup differences were evaluated using the Chi-squared test. When the overall difference was statistically significant, pairwise comparisons for categorical variables were further performed with Bonferroni correction. Multivariate logistic regression analysis was performed to identify the independent predictors for breast cancer molecular subtype. The combined multimodal ultrasonographic model was constructed based on the independent predictors identified by multivariate logistic regression analysis. In this study, a one-vs-rest strategy was used to evaluate the diagnostic performance of the model; that is, each molecular subtype was separately defined as the positive outcome, whereas the remaining subtypes were defined as the negative outcome. The predicted probabilities generated from the combined model were used to construct receiver operating characteristic (ROC) curves for each subtype, in order to evaluate the diagnostic value of multimodal ultrasonography for breast cancer molecular subtyping. The area under the curve (AUC), sensitivity, specificity, and their 95% confidence intervals (CIs) were calculated to assess the diagnostic performance of the model. The DeLong test was used to compare the AUCs between the combined multimodal ultrasonographic model and each individual ultrasound parameter. An AUC ≥0.80 was considered to indicate favorable diagnostic performance, whereas an AUC ≥0.90 was considered to indicate excellent diagnostic performance. A two-sided P value <0.05 was considered to indicate statistical significance.

Results

Comparison of baseline characteristics among different molecular subtypes

No statistically significant differences were observed among the four molecular subtype groups with respect to baseline characteristics, including age and body mass index (BMI) (P>0.05) (Table 2).

Comparison of conventional ultrasonographic parameters among different molecular subtypes

Significant differences were identified in conventional ultrasonographic parameters among the four molecular subtypes of breast cancer (P<0.05).

Regarding tumor morphology, lesions in the HER2-overexpression subtype were predominantly regular in shape, whereas nearly equal proportions of regular and irregular morphology were observed in the triple-negative subtype. In contrast, tumors in the luminal A and luminal B subtypes were more frequently characterized by irregular morphology.

With respect to margin characteristics, microlobulated margins were most commonly observed in the triple-negative subtype, while spiculated margins were most commonly observed in the luminal A subtype, and indistinct margins in the luminal B subtype.

In terms of calcification, the lowest incidence was noted in the triple-negative subtype, whereas the highest incidence was observed in the luminal A subtype.

Regarding posterior acoustic features, posterior enhancement was most frequently identified in the triple-negative subtype and posterior attenuation in the luminal A subtype; meanwhile, no significant posterior change was most frequently observed in the luminal B subtype, followed by the HER2-overexpression subtype (Table 3).

Comparison of color Doppler ultrasonographic parameters among the different molecular subtypes

Statistically significant differences among the four molecular subtypes were also identified from the color Doppler ultrasonographic parameters (P<0.05). Regarding Adler blood flow grading, grades 0–I were most frequently observed in patients with the luminal A subtype and grades II–III in the HER2-overexpression subtype. In terms of the VI, significantly higher values were observed in the triple-negative subtype compared with the other three molecular subtypes (Table 4).

Comparison of SWE parameters among the different molecular subtypes

Significant differences were observed in the E_max and E_ratio among the four molecular subtypes of breast cancer (P<0.05). Compared with the luminal A and luminal B subtypes, the HER2-overexpression and triple-negative subtypes exhibited markedly elevated E_max and E_ratio values. Meanwhile, E_min and E_mean did not differ significantly among the subtypes (P>0.05) (Table 5).

Association between multimodal ultrasonographic features and the molecular subtypes of breast cancer

Breast cancer molecular subtype was defined as the dependent variable. Variables that demonstrated statistical significance in univariate analysis were entered into a multivariate logistic regression model as independent variables. The results indicated that calcification status, VI, Adler blood flow grading, and E_ratio were independently associated with breast cancer molecular subtype (P<0.05) (Table 6).

Diagnostic performance of multimodal ultrasonographic features in the molecular subtyping of breast cancer

ROC curve analysis demonstrated that the combination of multimodal ultrasonographic parameters (composite indicator) yielded significantly higher AUC in differentiating among the molecular subtypes than did any single parameter alone (all P values <0.05), along with favorable sensitivity and specificity. The corresponding ROC curves and detailed diagnostic indices are presented in Figure 1 and Table 7.

Figure 1 ROC curves of multimodal ultrasonographic features for the identification of breast cancer molecular subtype. (A) ROC curve of multimodal ultrasonographic features in identifying the HER2-overexpression subtype. (B) ROC curve of multimodal ultrasonographic features in identifying the triple-negative subtype. (C) ROC curve of multimodal ultrasonographic features in identifying the luminal A subtype. (D) ROC curve of multimodal ultrasonographic features in identifying the luminal B subtype. E_ratio, the elasticity ratio between the lesion and surrounding normal tissue; HER2, human epidermal growth factor receptor 2; ROC, receiver operating characteristic; VI, vascularization index.

DeLong test

DeLong’s test was used to compare the AUC between the combined multimodal ultrasonographic model and each individual ultrasound parameter. The results showed that, for the identification of the HER2-overexpression, triple-negative, luminal A, and luminal B subtypes, the AUC of the combined model was significantly higher than those of individual parameters, including calcification, Adler blood flow grade, VI, and E_ratio. Moreover, the 95% CIs for all AUC differences did not cross 0, and the differences were statistically significant (all P values <0.05) (Table 8).

Discussion

In this study, a multivariate logistic regression model consisting of multimodal ultrasonographic parameters demonstrated that calcification status, VI, Adler blood flow grading, and E_ratio were independently associated with breast cancer molecular subtypes. These findings indicate that ultrasound-derived imaging features provide substantial value in the molecular classification of breast cancer. Calcification represents a critical imaging characteristic of breast cancer, particularly in early detection and subtype differentiation. In this study, with the luminal B subtype serving as the reference group, calcifications were more frequently observed in the HER2-overexpression and luminal A subtypes, whereas they were less commonly identified in the triple-negative subtype. These observations are consistent with previously published reports (18-20). From a pathological perspective, the mechanisms underlying calcification formation vary across molecular subtypes. Tumors with HER2 overexpression subtype are characterized by high proliferative activity, rendering them more susceptible to intratumoral ischemia and necrosis. Cellular debris derived from necrotic tissue may provide a substrate for calcium salt deposition, thereby facilitating the development of microcalcifications (21). Luminal A breast cancers are often accompanied by intraductal components or low-grade ductal carcinoma in situ, and microcalcification constitutes a typical radiologic manifestation of intraductal lesions, which may account for the relatively higher incidence of calcification in this subtype (22,23). In contrast, triple-negative breast cancer typically presents as a rapidly growing solid mass with a predominance of invasive components and relatively limited intraductal involvement. Imaging manifestations tend to be mass formation rather than calcific foci, indicating a comparatively lower likelihood of calcification (24,25).

Adler blood flow grading and VI semiquantitatively and quantitatively reflect tumor vascularity, respectively, and together provide insight into tumor perfusion and microvascular density. In our study, a highly consistent distribution pattern was observed between these two parameters across molecular subtypes. With the luminal B subtype serving as the reference category, the luminal A subtype was more likely to present with Adler grades 0–I and lower VI values, whereas the HER2-overexpression and triple-negative subtypes were more likely to present Adler grades II–III and higher VI values. This distribution suggests marked differences in hemodynamic characteristics and vascularization levels among the molecular subtypes of breast cancer (26,27). In terms of mechanism, luminal A breast cancer is typically hormone receptor-positive and is generally associated with less aggressive biological behavior. Lower proliferative activity and slower tumor growth may lead to limited activation of pro-angiogenic signaling pathways, resulting in insufficient neovascularization and a relatively low microvessel density within the tumor microenvironment (28). Consequently, lower VI values are observed. Meanwhile, reduced perfusion and diminished Doppler-detectable flow signals contribute to the predominance of low-flow patterns (Adler grades 0–I) in this subtype. Conversely, HER2-overexpression and triple-negative breast cancers are commonly characterized by higher proliferation indices and greater invasiveness. In these subtypes, enhanced expression of pro-angiogenic factors within the tumor microenvironment may simultaneously promote increased vascular formation and perfusion. In HER2-overexpression tumors, persistent activation of the HER2 signaling pathway has been shown to upregulate vascular endothelial growth factor (VEGF) and other angiogenic mediators, thereby stimulating neovascularization (29). Triple-negative breast cancer is frequently associated with a pronounced hypoxic microenvironment, in which angiogenesis may be further enhanced through hypoxia-inducible factor (HIF)-related pathways (30). Under the influence of these molecular mechanisms, intratumoral microvessels become more abundant and irregularly distributed, leading to elevated VI values. Concurrently, the presence of a denser and more active vascular network increases lesion perfusion and enhances the visibility of blood flow signals, thereby contributing to higher Adler blood flow grades (II–III) (31,32).

E_ratio constitutes a key parameter in SWE, reflecting the relative difference in stiffness between the tumor and the surrounding normal tissue; a higher E_ratio value indicates that the lesion is relatively harder compared to adjacent parenchyma. In our study, elevated E_ratio values were found to be associated with the triple-negative and HER2-overexpression subtypes, whereas lower E_ratio values were associated with the luminal A subtype. This phenomenon may be attributable to the propensity of triple-negative and HER2-overexpression tumors to undergo infiltrative growth and to markedly activate peritumoral stromal reactions. Sustained activation of tumor-associated fibroblasts may promote collagen deposition and extracellular matrix remodeling, thereby resulting in a denser tumor architecture and significantly increased tissue stiffness. Consequently, the stiffness contrast between tumor tissue and adjacent normal breast tissue is amplified (33,34).

The ROC curve analysis demonstrated that the combined application of multimodal ultrasonographic features—including calcification status, VI, Adler blood flow grading, and E_ratio—yielded a significantly greater AUC than did any single parameter alone, with both sensitivity and specificity reaching favorable levels. These findings suggest that integrative analysis of multimodal ultrasound parameters can substantially enhance the accuracy of molecular subtyping in breast cancer. The results further underscore the potential utility of multimodal ultrasonography in the noninvasive assessment of breast cancer molecular subtypes. Moreover, they support its application in clinical practice, particularly in the context of early screening and the formulation of individualized therapeutic strategies.

However, several limitations should be acknowledged. First, this was a single-center retrospective design, which may have been subject to inherent selection bias. In addition, the sample size was relatively small, and the distribution of cases among the molecular subtypes was not completely balanced, which may have affected the statistical power and robustness of the findings to some extent. Second, ultrasonographic examination is inherently operator-dependent. Although all procedures were performed by experienced sonographers, a certain degree of subjective measurement bias might have been introduced, particularly during the manual delineation of the region of interest in VI assessment and SWE. Third, systematic internal validation using bootstrap resampling or a split-sample approach was not performed in this study, and this study did not include an independent external validation cohort; therefore, the stability and generalizability of the model performance remain to be further confirmed across different centers, regions, or patient populations. Future multicenter, large-scale, prospective studies incorporating a broader range of imaging parameters and pathological indicators are warranted, and both internal validation and independent external validation cohorts should be included to further assess the robustness and generalizability of the findings.

Conclusions

Calcification status, VI, Adler blood flow grading, and E_ratio were identified as independent predictors of molecular subtype in breast cancer. The application of multimodal ultrasonography shows potential diagnostic value in predicting breast cancer molecular subtypes and may provide valuable noninvasive auxiliary information for individualized therapeutic decision-making.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0214/rc

Data Sharing Statement: Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0214/dss

Peer Review File: Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0214/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0214/coif). The 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 and its subsequent amendments. The study was approved by the Ethics Committee of Nanjing Lishui People’s Hospital (No. 2026KY0409-01) and informed consent was taken from all the patients.

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. Giaquinto AN, Sung H, Newman LA, et al. Breast cancer statistics 2024. CA Cancer J Clin 2024;74:477-95. [Crossref] [PubMed]
2. Arnold M, Morgan E, Rumgay H, et al. Current and future burden of breast cancer: Global statistics for 2020 and 2040. Breast 2022;66:15-23. [Crossref] [PubMed]
3. Siegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA Cancer J Clin 2024;74:12-49. [Crossref] [PubMed]
4. Zhang X. Molecular Classification of Breast Cancer: Relevance and Challenges. Arch Pathol Lab Med 2023;147:46-51. [Crossref] [PubMed]
5. Turner KM, Yeo SK, Holm TM, et al. Heterogeneity within molecular subtypes of breast cancer. Am J Physiol Cell Physiol 2021;321:C343-54. [Crossref] [PubMed]
6. Xu X, Lu L, Zhu L, et al. Predicting the molecular subtypes of breast cancer using nomograms based on three-dimensional ultrasonography characteristics. Front Oncol 2022;12:838787. [Crossref] [PubMed]
7. Naeimzadeh Y, Heidari Z, Razban V, et al. Deregulated MicroRNAs involved in P53 Signaling Pathway in Breast Cancer with Focus on Triple-negative Breast Cancer. Curr Mol Pharmacol 2024;17:e18761429263841. [Crossref] [PubMed]
8. Hu E, Wang Y, Chen G, et al. Trifluridine/Tipiracil Exhibits Potent Antitumor Activity in Triple-Negative Breast Cancer by Targeting SKP2 and Inducing Cell Cycle Arrest and Apoptosis. Current Pharmaceutical Analysis 2025;22:35-43.
9. Johnson KS, Conant EF, Soo MS. Molecular Subtypes of Breast Cancer: A Review for Breast Radiologists. J Breast Imaging 2021;3:12-24. [Crossref] [PubMed]
10. Ye X, Yang L, He Q, et al. Reconceptualizing the clinicopathological features, locoregional therapy and prognostic factors of occult breast cancer in the era of molecular subtyping. Women Health 2023;63:105-14. [Crossref] [PubMed]
11. Vi C, Mandarano G, Shigdar S. Diagnostics and Therapeutics in Targeting HER2 Breast Cancer: A Novel Approach. Int J Mol Sci 2021;22:6163. [Crossref] [PubMed]
12. Tripathi D, Davies NM, Rajinikanth PS, et al. Advancements in Targeted Therapies and Pharmacogenomics for Personalized Breast Cancer Treatment: The Role of Gene SNPs in Treatment Resistance. Curr Gene Ther 2025; Epub ahead of print. [Crossref]
13. Dou T, Chen Y, Liu L, et al. Radiogenomic analysis of clinical and ultrasonic characteristics in correlation to immune-related genes in breast cancer. Sci Rep 2025;15:15918. [Crossref] [PubMed]
14. Dan Q, Zheng T, Liu L, et al. Ultrasound for Breast Cancer Screening in Resource-Limited Settings: Current Practice and Future Directions. Cancers (Basel) 2023;15:2112. [Crossref] [PubMed]
15. Zhu JY, He HL, Jiang XC, et al. Multimodal ultrasound features of breast cancers: correlation with molecular subtypes. BMC Med Imaging 2023;23:57. [Crossref] [PubMed]
16. Health Commission Of The People’s Republic Of China N. National guidelines for diagnosis and treatment of breast cancer 2022 in China (English version). Chin J Cancer Res 2022;34:151-75.
17. He MN, Lv K, Jiang YX, et al. Application of superb microvascular imaging in focal liver lesions. World J Gastroenterol 2017;23:7765-75. [Crossref] [PubMed]
18. Algazzar MA, Elsayed E, Alhanafy AM, et al. Breast cancer imaging features as a predictor of the hormonal receptor status, HER2neu expression and molecular subtype. Egypt J Radiol Nuclear Med 2020;51:93.
19. Gumowska M, Mączewska J, Prostko P, et al. Is There a Correlation between Multiparametric Assessment in Ultrasound and Intrinsic Subtype of Breast Cancer? J Clin Med 2021;10:5394. [Crossref] [PubMed]
20. Lin CJ, Xiao WX, Fu T, et al. Calcifications in triple-negative breast cancer: Molecular features and treatment strategies. NPJ Breast Cancer 2023;9:26. [Crossref] [PubMed]
21. Wang H, Chen W, Jiang S, et al. Intra- and peritumoral radiomics features based on multicenter automatic breast volume scanner for noninvasive and preoperative prediction of HER2 status in breast cancer: a model ensemble research. Sci Rep 2024;14:5020. [Crossref] [PubMed]
22. Venkatesh SL, Oseni TO, Bahl M. Symptomatic ductal carcinoma in situ (DCIS): Upstaging risk and predictors. Clin Imaging 2021;73:101-7. [Crossref] [PubMed]
23. Yamaguchi R, Watanabe H, Mihara Y, et al. Histopathology of non-mass-like breast lesions on ultrasound. J Med Ultrason 2001;2023:375-80. [Crossref] [PubMed]
24. Lohitvisate W, Pummee N, Kwankua A. Mammographic and ultrasonographic features of triple-negative breast cancer compared with non-triple-negative breast cancer. J Ultrasound 2023;26:193-200. [Crossref] [PubMed]
25. Wang K, Zou Z, Shen H, et al. Calcification, Posterior Acoustic, and Blood Flow: Ultrasonic Characteristics of Triple-Negative Breast Cancer. J Healthc Eng 2022;2022:9336185. [Crossref] [PubMed]
26. Li X, Zhang J, Zhang G, et al. Contrast-Enhanced Ultrasound and Conventional Ultrasound Characteristics of Breast Cancer With Different Molecular Subtypes. Clin Breast Cancer 2024;24:204-14. [Crossref] [PubMed]
27. Li J, Huo G, Lei X, et al. Correlation analysis of multiparametric magnetic resonance imaging features and molecular subtypes of breast cancer. J Clin Imaging Sci 2025;15:37. [Crossref] [PubMed]
28. Kraby MR, Krüger K, Opdahl S, et al. Microvascular proliferation in luminal A and basal-like breast cancer subtypes. J Clin Pathol 2015;68:891-7. [Crossref] [PubMed]
29. Jafari S, Ostadrahimi A, Farjami A, et al. Synergistic anticancer effects of Chrysin and trastuzumab in HER2-Positive breast cancer cells. Sci Rep 2025;15:41215. [Crossref] [PubMed]
30. Feng D, Gao J, Liu R, et al. CARM1 drives triple-negative breast cancer progression by coordinating with HIF1A. Protein Cell 2024;15:744-65. [Crossref] [PubMed]
31. Sachani P, Dhande R, Parihar P, et al. Enhancing the Understanding of Breast Vascularity Through Insights From Dynamic Contrast-Enhanced Magnetic Resonance Imaging: A Comprehensive Review. Cureus 2024;16:e70226. [Crossref] [PubMed]
32. Wang K, Yang X, Yang S, et al. A diagnostic test of two-dimensional ultrasonic feature extraction based on artificial intelligence combined with blood flow Adler classification and contrast-enhanced ultrasound for predicting HER-2-positive breast cancer. Transl Cancer Res 2025;14:640-50. [Crossref] [PubMed]
33. Acerbi I, Cassereau L, Dean I, et al. Human breast cancer invasion and aggression correlates with ECM stiffening and immune cell infiltration. Integr Biol (Camb) 2015;7:1120-34. [Crossref] [PubMed]
34. Akinpelu A, Akinsipe T, Avila LA, et al. The impact of tumor microenvironment: unraveling the role of physical cues in breast cancer progression. Cancer Metastasis Rev 2024;43:823-44. [Crossref] [PubMed]

(English Language Editor: J. Gray)