Background: The Detection and localization of breast lesions remain challenging in mammography and digital breast tomosynthesis (DBT) due to tissue overlap and information loss during volumetric reconstruction. Sinograms preserve the full angular projection data acquired during scanning, enabling analysis of tissue structure without reconstruction. Methods: This study proposes a direct segmentation approach for mammography and DBT sinograms using a U-Net architecture. Experiments were conducted on 1082 annotated mammography mass images from the CBIS-DDSM dataset (521 benign, 561 malignant) and 272 annotated DBT images from the Breast Cancer Screening DBT dataset (136 benign, 136 malignant). Dataset splitting was performed at the patient level to prevent data leakage, and all reported quantitative results correspond to the independent test set, with the validation set used solely for model selection and early stopping. Three input configurations were evaluated: mammography sinograms, DBT sinograms, and a combined model. Results: The mammography model achieved the highest performance (Dice: 0.94 training, 0.90 test), outperforming DBT alone (0.77 training, 0.70 test). Multimodal fusion improved DBT results (Dice: 0.84 test). Centroid analysis showed 99.11% correspondence with reference annotations (average distance: 2.83 pixels), and partial back-projection reconstructions confirmed anatomical consistency. Compared with YOLOv5x, the proposed approach provided superior lesion localization, particularly for small or multiple lesions. Conclusions: Direct sinogram segmentation is an efficient, clinically viable strategy for breast lesion detection and localization.

Diagnostic value of contrast-enhanced spectral mammography for breast lesions: qualitative and quantitative analyses

A novel and general spatiotemporal diagnostic model: Intratumoral outflow and peritumoral inflow for the differentiation and stratification of breast tumor.

Diffusion-weighted imaging (DWI) in breast MRI requires balancing image quality with acquisition time. Reducing scan time without sacrificing diagnostic performance could improve patient comfort and workflow. To compare 3-scan trace DWI (tDWI) and diagonal DWI (dDWI) for breast lesion evaluation using both phantom and clinical assessments, focusing on image quality metrics and diagnostic performance. Retrospective comparative study. A commercially available breast diffusion phantom (Caliber MRI; Boulder, CO, USA) and 92 consecutive participants were initially enrolled. After excluding 23 due to having no confirmed lesion, and 15 with non-mass lesions, 54 patients (median age 55.5 years, range 21-82) were analyzed. 3-T, tDWI (120 s) and dDWI (74 s) with identical parameters except for diffusion gradient directions, with dDWI reducing scan time by 38%. Phantom studies measured apparent diffusion coefficient (ADC) and estimated signal-to-noise ratio (eSNR). Clinical studies evaluated ADC, contrast-to-noise ratio (CNR), and contrast ratio (CR) from ROIs in lesions, normal breast tissue, and fat. Three radiologists scored lesion conspicuity and image quality. Due to non-normal data distribution, the Wilcoxon signed-rank test compared metrics between sequences. The ADC coefficient of variation (CV) was calculated. A p-value < 0.05 was considered significant. dDWI significantly improved image quality by reducing artifacts, especiallythose originating from the nipple and in the breast tissue periphery, and slightly better lesion conspicuity. No significant difference was found for eSNR in breast phantom studies (p = 0.31 for b = 0; and p = 0.84 for b = 800 s/mm2). tDWI demonstrated significantly higher CNR for breast tissue and fat. tDWI also showed lower CR values for tumor/breast tissue and lower values for tumor/fat. ADC measurements were similar between techniques (CV = 4.20%). dDWI provides a 38% shorter acquisition time than tDWI while maintaining comparable quantitative performance. dDWI demonstrates improved image quality, particularly in challenging anatomical regions, though tDWI yields higher contrast-to-noise ratio values. 3. 2.

Numerous neural networks based on the U-Net architecture have been developed for the segmentation of breast ultrasound images. However, the accuracy of such segmentation tasks is often compromised by the complex and variable shapes of tumors, the indistinct boundaries of lesion areas, and backgroundnoise. In this study, we develop a hybrid receptive field U-Net (HRF U-Net) to improve lesion segmentation in breast ultrasoundimages. We design an innovative hybrid receptive field module (HRFM) to replace conventional convolutional layers, integrating deformable convolutions that expand the receptive field and enhance the model's ability to capture shape and boundary features. We systematically analyze the differential feature extraction capabilities of deformable convolutions across network layers, combining them with standard convolutions, dilated convolutions, and max pooling. This configuration enables HRF U-Net to achieve a broader and more adaptive receptive field, enhancing its effectiveness in handling the complexities of breast lesion segmentation. Additionally, we introduce a large-kernel attention module (LKAM) within the skip connections, which expands the receptive field and supports adaptive feature selection, capturing long-range dependencies within the convolutional attention mechanism. This novel approach enables more precise feature extraction, effectively mitigates boundary noise during training, and substantially improves the model's segmentation performance. We used three publicly available datasets to conduct extensive experiments, including ablation studies, comparative analyses, robustness evaluations, and external validation. Datasets A and B were divided into training and validation sets for four-fold cross-validation, while Dataset C was used as the test set for external validation. We selected five widely used image segmentation metrics for validation, namely pixel accuracy, precision, recall, Jaccard index, and Dice coefficient. In addition, the statistical significance was evaluated using the paired Student's t-test with Holm-Bonferroni correction . Experimental results on three public datasets demonstrate that the proposed HRF U-Net substantially improves the efficacy of breast ultrasound image segmentation, outperforming several state-of-the-art works. Specifically, in the comparative experiments, HRF U-Net achieved scores of 96.52, 85.03, 83.19, 73.15, and 81.34 for the five metrics on Dataset A. On Dataset B, the scores reached 98.61, 90.22, 83.99, 74.50, and 83.64 . The effect size Cohen's d values for HRF U-Net compared to several typical and advanced networks all exceed 0.32, demonstrating a significant improvement in segmentation performance. In external validation, the metric scores reached 97.02, 84.35, 93.94, 77.74, and 86.86 . The HRF U-Net proposed in this paper improves the segmentation accuracy of breast ultrasoundimages.

ASFCP-Net: Axial Spatial-Frequency Domain Collaborative Perception Dual-Branch Network for breast lesion segmentation in ultrasound images

Granular cell tumor (GCT) of the breast is a rare, typically benign neoplasm of Schwann cell origin, which frequently presents a significant diagnostic challenge due to its ability to clinically and radiologically mimic invasive breast carcinoma. We present the case of a 45-year-old female who presented with a palpable, firm, non-mobile mass in the right upper inner quadrant of her breast. Diagnostic imaging was highly suspicious for malignancy, with mammography revealing a high-density, spiculated, irregular mass with architectural distortion (BI-RADS 5), and ultrasound demonstrating an irregular, hypoechoic mass with severe posterior acoustic shadowing and a "taller-than-wide" orientation. Core needle biopsy was performed due to the high suspicion of malignancy. Histopathological examination revealed nests and sheets of large, polygonal cells with abundant granular, eosinophilic cytoplasm. Immunohistochemistry (IHC) studies were critical for definitive diagnosis, showing strong positivity for S100 protein and SOX10, and granular positivity for CD68, while being negative for cytokeratin markers (AE1/AE3, CAM 5.2). This immunoprofiling confirmed the diagnosis of benign GCT. The patient was successfully treated with complete surgical excision (R0 resection) and has remained free of local recurrence at three years of follow-up. This case underscores the critical role of tissue diagnosis and characteristic IHC in differentiating this rare benign entity from malignant breast lesions, thereby preventing unnecessary aggressive treatment.

Background/Objectives: Palliative radiotherapy (RT) to the breast or chest wall is an effective option for symptom relief in advanced or metastatic breast cancer. However, real-world data on clinical outcomes, prognostic factors, and optimal RT dose and timing in the modern systemic therapy era remain limited. This study aimed to evaluate the efficacy of palliative breast RT and explore the optimal integration of systemic therapy with RT. Methods: We retrospectively reviewed 38 patients treated with palliative RT between 2015 and 2024 at a regional tertiary center. The median gross tumor volume (GTV) dose was 50 Gy (range, 30–62.5 Gy), corresponding to a median biologically effective dose (BED) to the GTV (α/β = 4) of 78.9 Gy. Treatment response, symptom relief, toxicity, and survival outcomes were analyzed. Results: With a median follow-up of 9.5 months, median overall survival (OS) was 12.8 months. 1- and 2-year in-field local control (LC) rates were both 79.6%. In a parsimonious multivariate analysis, ≥3 prior lines of systemic therapy (HR 3.500, 95% CI 1.278–9.590, p = 0.015) was independently associated with worse OS, whereas use of SIB or GTV boost was associated with improved OS (HR 0.351, 95% CI 0.145–0.848, p = 0.020). Higher planning target volume (PTV)-BED (α/β = 4) correlated with improved LC (HR 0.909, 95% CI 0.839–0.985, p = 0.019). Symptom relief within 3 months occurred in 82%. Treatment-related toxicities were generally mild, with no grade ≥ 4 acute or grade ≥ 2 late toxicities observed. Conclusions: Palliative breast RT achieved durable local control and effective symptom palliation with acceptable toxicity. Dose-tailored RT and earlier integration within the disease course may optimize outcomes in selected patients.

Introduction Breast cancer remains a major challenge in women’s health globally. Early screening and personalized treatment can improve outcomes. This study aimed to evaluate ultrasound viscosity imaging (UVI) for distinguishing benign from malignant breast lesions and noninvasively assessing human epidermal growth factor receptor 2 (HER2) status. Materials and methods We conducted a retrospective analysis of 274 breast lesions, randomly divided into a derivation cohort and a validation cohort (VC) at a 7:3 ratio. Breast Imaging Reporting and Data System (BI-RADS) scores and UVI parameters were collected, with histopathology as the reference standard. The Boruta algorithm was used to identify the optimal viscous parameter (VP). A logistic regression model assessed the diagnostic performance of BI-RADS alone and in combination with VP. Associations between viscous parameters (VPs) and HER2 status were also examined. Results Among 40 VPs, V2.max (maximum viscosity from the Voigt model within a perilesional 2-mm rim) was identified as the optimal marker. When combined with BI-RADS, V2.max enhanced the differentiation between benign and malignant lesions (p&amp;lt;0.001), increasing the area under the curve (AUC) from 0.91 (95% CI: 0.87-0.95) to 0.96 (95% CI: 0.94-0.98). The combined model also demonstrated superior calibration, which was revalidated in the VC. Subgroup analyses confirmed its effectiveness in younger patients and those with larger lesions. Furthermore, we identified four Voigt-model-derived VPs, including V2.max, that correlated with HER2 positivity, and explored their potential histological basis. Conclusion UVI-derived VPs enhance BI-RADS diagnostic performance for breast lesions and are associated with HER2 status.

Background/Objectives: Estrogen receptor (ER) expression is a key biomarker in breast cancer (BC) and guides endocrine therapy selection. Estrogen receptor-targeted imaging with 16ɑ-[18F]-fluoro-17β-estradiol (18F-FES) PET is recommended in several clinical guidelines for noninvasive assessment of ER status. In clinical practice, 18F-FES PET may also identify ER-negative malignancies or benign breast lesions with variable uptake patterns. This study aimed to systematically characterize 18F-FES PET/MRI uptake patterns in benign breast lesions and across breast cancer subtypes defined by receptor status, histology, and molecular phenotype. Methods: This retrospective single-center study included 41 women with 50 breast lesions who underwent simultaneous 18F-FES PET/MRI prior to any treatment. Histopathology or long-term follow-up served as the standard of reference. Maximum and mean standardized uptake values (SUVmax and SUVmean) were derived using MRI-based lesion delineation. Results: Both benign and malignant breast lesions measuring < 10 mm demonstrated low 18F-FES uptake (SUVmax < 1.00). 18F-FES uptake among benign breast lesions was variable, with SUVmax ranging from 0.44 to 1.57. In contrast, ER-positive lesions ≥ 10 mm exhibited substantially higher 18F-FES uptake (median SUVmax 2.76; range 1.23-9.74) compared with ER-negative tumors of similar size (SUVmax 0.30-0.94). 18F-FES uptake was consistent across histologic BC subtypes and did not differ significantly among ER-positive molecular subtypes. No significant associations were observed with HER2 status or tumor grade. Conclusions: Awareness of the heterogeneous 18F-FES uptake patterns in benign breast lesions, as well as the limited sensitivity for detecting ER-positive tumors < 10 mm, is essential for accurate image interpretation. 18F-FES PET/MRI enables reliable assessment of ER expression in BC lesions ≥ 10 mm, with uptake patterns remaining consistent across molecular and histologic subtypes.

Non-palpable breast cancer lesions can be treated with breast-conserving surgery, for which precise localization of lesions is key. Many localization techniques are available, but all have limitations. The European Institute of Oncology designed the RADIOSO Milan Trial to evaluate the clinical impact of SCOUT® radar localization (Merit MedicalSystems Inc.), a wire-free, non-radioactive, non-magnetic technology for precise localization of non-palpable breast lesions. RADIOSO was a single-center, prospective, observational study that enrolled patients with histologically confirmed non-palpable breast lesions eligible for breast-conserving surgery. Respective surveys were used to evaluate patient, radiologist, and surgeon satisfaction with the clinical impact of SCOUT®. A total of 298 patients were enrolled after giving informed consent; 254 were included in the final study analysis. The reflector was placed and removed successfully in all cases. In 16% of cases, the reflector was placed during the biopsy, improving patient workflow. In total, 98% of radiologists, 89.8% of surgeons, and 91.8% of patients declared themselves satisfied with the use of SCOUT®. No significant SCOUT®-related artifacts were observed, regardless of the radiological modality chosen. SCOUT® is a promising preoperative localization device for non-palpable breast and axillary lesions. It facilitates a single-step placement approach by enabling the concurrent deployment of the radar reflector during the initial biopsy, optimizing clinical workflow. NCT05963464.

This pilot study aimed to evaluate whether segmented coronal breast maximum intensity projection (MIP) images generated from contrast-enhanced CT (CE-CT) could improve the detectability of breast lesions incidentally visualized in clinical settings where the breast is not the primary imaging target. This retrospective study included 71 patients with breast cancer who underwent CE-CT for preoperative staging before surgery. Three radiologists independently assessed the presence or absence of breast lesions using a 3-tiered scoring system on both segmented coronal breast MIP and 5-mm axial images. Reading times were recorded. Readers then compared paired breast MIP and axial images and indicated which reconstruction type was more effective for breast lesion detection. The Wilcoxon signed-rank test with Bonferroni correction was used to compare reading times, and the McNemar-Bowker test was used to compare 3-tiered lesion detectability scores. Segmented coronal breast MIP images were read faster than axial images by 2 of the 3 readers (median [interquartile range]: 4 [4-7] vs 9 [8-11] seconds and 3 [2-6] vs 11 [9-14.5] seconds; P < .001). No significant difference was found for the third reader (P = .446). For lesion detectability, breast MIP images were significantly superior to axial images for reader 1 (P = .029), reader 2 (P = .043), and reader 3 (P = .004). Breast MIP images were preferred in 76.1% (54/71), 95.8% (68/71), and 83.1% (59/71) of cases, respectively. Segmented coronal breast MIP images from CE-CT may improve lesion detectability and reduce interpretation time compared with conventional axial images.

Breast Masses in Adolescent Females: A Scoping Review.

Abstract Background: Invasive lobular carcinoma (ILC) accounts for approximately 10-15% of invasive breast cancers and is characterized by a unique histologic growth pattern that often results in subtle imaging findings. Notably, up to 30% of cases may be mammographically occult. ILC is also characterized by an atypical metastatic pattern, with a tendency to spread to sites such as the gastrointestinal tract, peritoneum, and orbit. Presentation as isolated periorbital swelling is exceedingly rare and presents a significant diagnostic challenge, particularly in the context of a negative screening mammogram. Case Presentation: A 56-year-old woman initially presented to her primary care physician with left periorbital swelling, redness, and discomfort. Symptoms were recurrent for over a year, and during this time she was evaluated by optometrists, ophthalmologists, and plastic surgeons without a definitive diagnosis. Empiric treatment for presumed allergic conjunctivitis and sinusitis provided little improvement before symptoms would recur. Screening mammogram during this time was negative for malignancy. She eventually underwent a brain MRI for persistent symptoms which demonstrated nonspecific enhancement of the left orbit. Given ongoing symptoms, she underwent surgical resection of the infraorbital mass, and pathology revealed metastatic ILC. The patient had no breast-related symptoms, and repeat diagnostic mammography and ultrasound failed to identify a primary tumor. Subsequent breast MRI detected a subtle right breast lesion, which was confirmed as ILC on biopsy. Discussion: This case highlights the diagnostic limitations of mammography in detecting ILC and underscores the importance of maintaining clinical suspicion when evaluating persistent orbital symptoms. Orbital metastases are a rare and often an overlooked manifestation of ILC, especially when it is the initial presentation. This case also highlights the value of MRI in the diagnostic workup of ILC, even with negative mammography. Conclusion: Periorbital swelling may represent the initial and sole manifestation of ILC. Given the potential for ILC to spread to atypical sites, providers should be aware of this rare presentation and include breast cancer in the differential diagnosis for unexplained orbital symptoms. Further imaging with MRI should be considered, even in the setting of a negative mammogram, if there is high clinical suspicion for ILC to help avoid delays in diagnosis. Citation Format: S. A. Shaker, D. Ramirez. Atypical Presentation of Invasive Lobular Carcinoma Presenting as Periorbital Swelling Despite Negative Mammogram [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS5-06-12.

Abstract Background: In the last decades, the proportion of breast cancer patients receiving breast-conserving surgery has increased, reaching 70-80% in developed countries. In case of non-palpable lesions, surgical excision requires some form of breast localization. While wire-guided localization has long been considered gold standard, it carries several limitations, including logistical difficulties, the potential for displacement and patient discomfort, and re-excision rates reaching 21%. Other techniques (radioactive seed or radio-occult lesion localization, intraoperative ultrasound, magnetic, radiofrequency and radar localization) have been developed with the aim of overcoming these disadvantages. However, comparative data on the rates of successful lesion removal, negative margins and re-operations are limited. In the majority of studies, the patient’s perspective with regard to discomfort and pain level has not been evaluated. The aim of MELODY (MEthods for LOcalization of Different types of breast lesions) is to evaluate different imaging-guided localization methods with regard to oncological safety, patient-reported outcomes, and surgeon and radiologist satisfaction. Methods: The EUBREAST and the iBRA-NET have initiated the MELODY study to assess breast localization techniques and devices from several perspectives (NCT05559411, http://eubreast.org/melody). MELODY is a prospective intergroup cohort study which enrolls female and male pts. requiring breast-conserving surgery and imaging-guided localization for invasive breast cancer or DCIS. Multiple or bilateral lesions and neoadjuvant chemotherapy are allowed. Primary outcomes are: 1) Intended target lesion and/or marker removal, independent of margin status on final histopathology, and 2) Negative resection margin rates at first surgery. Secondary outcomes are, among others: rates of second surgery and secondary mastectomy, Resection Ratio (defined as actual resection volume divided by the calculated optimum specimen volume), duration of surgery, marker dislocation rates, rates of marker placement or localization failure, comparison of patient-reported outcomes, rates of “lost markers” and diagnostician/radiologist’s and surgeon’s satisfaction as well as the health economic evaluation of the different techniques. Target accrual: 7,416 patients. Enrollment started in January 2023. The study will be conducted in 30 countries and is supported by the Oncoplastic Breast Consortium (OPBC), AWOgyn, AGO-B and SENATURK. Financial support was provided by Endomag, Merit Medical, Sirius Medical and Hologic. Citation Format: J. Ryu, T. Kühn, N. Ditsch, J. Harvey, S. Hartmann, N. Cabioglu, N. Canturk, A. Karakatsanis, J. de Boniface, T. Filtenborg Tvedskov, L. Pankratjevaite, M. Gasparri, S. Alran, E. Schlichting, A. Lowery, A. Esgueva, R. Di Micco, D. Murawa, M. Kontos, M. Muneer Khan, Y. Masannat, V. Fabiano, K. Abdelwahab, L. Rebaza, F. Peintinger, M. Correia, E. Giblin, G. Dindelegan, S. Nietz, F. Malherbe, A. Kothari, R. Dave, O. Gentilini, B. Güllüoglu, M. Hahn, G. Karadeniz Cakmak, M. Lux, S. Potter, I. Rubio, M. Smidt, W. Weber, N. Krawczyk, K. Jursik, A. Kaiser, A. Körner, M. Köpke, N. Tauber, S. Lukac, M. Banys-Paluchowski, MELODY study group. Melody: A prospective non-interventional multicenter cohort study to evaluate different imaging-guided methods for localization of malignant breast lesions (Eubreast-4 / iBra-net, Nct 05559411) [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS5-08-17.

Abstract Background: Invasive lobular carcinoma (ILC) is the second most common breast cancer (BC) subtype, accounting for up to 15% of cases. Although several biologic differences between ILC and invasive carcinoma of no special type (NST) have been described, most data are derived from primary tumors. In contrast, the biology of metastatic ILC remains poorly defined. This study aimed to investigate the unique genomic, transcriptomic, and immune landscapes of metastatic ILC using a multi-omic approach. Methods: This real-world study included patients with confirmed NST or pure ILC who had a biopsy of a metastatic lesion or a breast lesion in the setting of known metastatic disease. Tumors underwent whole exome sequencing (NGS 592, NextSeq; WES, NovaSeq), whole transcriptome sequencing (WTS, NovaSeq) (Caris Life Sciences, Phoenix, AZ). Demographic and baseline molecular features were compared across histologic subtypes. Tumor mutational burden (TMB; high ≥10 mut/MB) was assessed by NGS. Immune cell populations were estimated by RNAseq deconvolution (quanTIseq). Pathway enrichment analysis was determined by GSEA. Overall survival (OS) was defined from the date of biopsy to death from any cause using Kaplan-Meier estimates. Statistical tests included chi-square and Mann-Whitney U with multiple testing correction (q&amp;lt;0.05). Results: Among 2,651 metastatic BC samples (ILC n = 608; NST n = 2,043), patients with ILC were older (median 67 vs 63 yrs) and more likely white (81.5% ILC vs 73.1% NST), all p&amp;lt;0.05. By IHC, ILC were more frequently HR+/HER2- (83.6% vs 55.9%), less commonly triple-negative (11.9% vs 29.8%) or HER2+ (4.6% vs 14.3%), all p&amp;lt;0.05. PAM50 subtypes differed between ILC and NST: luminal A (Lum A) (31.1% vs 13.7%) and luminal B (Lum B) (44.9% vs 43.8%) intrinsic subtypes were enriched in ILC, while NST was more frequently basal (1.2% vs 22.4%) and HER2-enriched (19.8% vs 22.2%), all p&amp;lt;0.05. Within luminal BC, Lum B was more common than Lum A in both histologies, but Lum A was proportionally more frequent in ILC (ILC: 40.3% Lum A vs 59.7% Lum B; NST: 23.4% Lum A vs 76.6% Lum B), p&amp;lt;0.0001. Comparative multi-omic analysis focused on the 922 HR+/HER2- tumors (ILC n = 275; NST n = 647). Compared to NST, ILC had higher frequency of CDH1 (87.3% vs 6.1%), PIK3CA (52.7% vs 38.5%), FOXA1 (9.0% vs 3.4%), ERBB2 (7.1% vs 2.0%), ARID1A, and NF1 (both q &amp;lt; 0.05) mutations and TMB high (17.7% vs 9.9%), but lower frequency of TP53 (15.7% vs 33.3%), ESR1 (10.9% vs 18.1%) and GATA3 (2.3% vs 14.9%) mutations along with FGFR1 amplifications (15.3% vs 6.7%), all q&amp;lt;0.05. ILC had higher expression of androgen receptor (91.2% vs 83.5%, q&amp;lt;0.05) and higher MAPK activation score (-0.81 vs -1.19, q&amp;lt;0.05), consistent with a luminal, hormone-driven phenotype. NST tumors demonstrated enrichment of pathways involved in the cell cycle and metabolic activity, including E2F targets, G2M checkpoint, MYC targets, mTORC1 signaling, glycolysis, and unfolded protein response (NES: 1.5-2.7, FDR&amp;lt;0.25). ILC tumors had higher infiltration of B cells, dendritic cells, neutrophils, NK cells and M2 macrophages, but lower M1 macrophages, all q&amp;lt;0.05. ILC had differential expression of immune checkpoint genes (upregulation: CD274 (PD-L1), PDCD1 (PD-1), TNFSF14, CEACAM1, CD160, fold change (FC): 1.2-1.5; down: HAVCR2 (TIM-3): FC: 1.2), all q&amp;lt;0.05. Median OS (mOS) was similar between luminal ILC and NST (33.8 vs 35.4 months (mo); HR 1.0, 95% CI 0.91-1.3). No mOS difference was seen between Lum A ILC and NST (47.4 vs 50.5 mo; HR 1.0, 0.84-1.43), while Lum B ILC had worse mOS than NST (27.4 vs 35.3 mo; HR 1.3, 1.1-1.6). Conclusions: This study represents the largest cohort to date with multi-omic characterization of metastatic ILC, revealing distinct genomic, transcriptomic, and immune features compared to NST. These findings highlight the unique biology of ILC and may inform future subtype-specific therapeutic strategies. Citation Format: G. Nader-Marta, S. Kumar Deshmukh, K. Fanucci, P. Tarantino, S. Schnitt, A. Lee, S. Oesterreich, S. Wu, J. Xiu, P. Advani, M. Lustberg, N. Lin, S. Tolaney, E. Mayer, O. Metzger, G. Sledge Jr., R. Jeselsohn. The Genomic, Transcriptomic, and Immune Hallmarks of Metastatic Lobular Breast Cancer [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PD9-07.

Pilomatrixoma is a rare benign skin neoplasm originating from the hair follicle matrix and is exceptionally uncommon in breast tissue, particularly in male patients. Due to overlapping clinical and ultrasonographic features, it is frequently misdiagnosed as breast cancer. This study aims to highlight pilomatrixoma as an important differential diagnosis of breast masses and to emphasize the role of histopathological confirmation. We report two male patients with breast lesions. A 40-year-old male presented with an infected breast lesion, and a 70-year-old male presented with a palpable breast mass. Both patients underwent clinical evaluation, ultrasonography, and tru-cut biopsy. Following histopathological confirmation of pilomatrixoma, elective surgical excision was performed. Tru-cut biopsy findings in both cases were consistent with pilomatrixoma. Ultrasonographic evaluation suggested suspicious features, leading to BI-RADS 4–5 categorization and initial concern for malignancy. Complete surgical excision with clear margins was achieved in both patients, and no complications or recurrences were observed. Pilomatrixoma is a rare benign tumor of the skin that can closely mimic breast cancer, especially on imaging. Despite its typical occurrence in the head, neck, and upper extremities and predominance in young females, it should be considered in the differential diagnosis of male breast masses. Histopathological confirmation via tru-cut biopsy is essential for accurate diagnosis, and surgical excision with clear margins is curative.

Abstract Introduction: Recent randomized controlled trials (RCTs) of breast MRI combined with mammography among high-risk women resulted in fewer interval breast cancer (BC) cases and earlier cancer detection compared to mammography alone. Despite this finding, supplemental breast MRI is underutilized. We aimed to identify factors associated with the receipt of supplemental breast MRI in a racially/ethnically diverse population of high-risk women. Methods: We conducted a retrospective cohort study among high-risk women undergoing screening mammography at Columbia University Irving Medical Center (CUIMC) in New York, NY, from 2007-2024. BC risk categories were classified based upon ICD-9/10 diagnostic codes or findings from breast radiology/pathology reports: 1) at high risk for BC, 2) family history of BC, 3) extremely dense breasts, 4) high-risk breast lesion (atypical hyperplasia [AH] or lobular carcinoma in situ [LCIS]), and 5) genetic predisposition for BC. Patients with a diagnosis of BC prior to 2007 were excluded. We collected information on demographic and clinical characteristics in the electronic health record (EHR), including age, race, ethnicity, primary health insurance, body mass index, mammographic density (BIRADS classification), benign breast biopsies, and family history of BC. Our primary outcome was receipt of supplemental breast MRI based upon radiology reports in the EHR. A multivariable logistic regression model was used to determine the association between breast MRI receipt and demographic and clinical characteristics. Results: Of the 54,122 women included in our analysis, the mean age was 58 years (SD, 10), including 29.0% non-Hispanic White women, 34.0% Hispanic, 12.0% non-Hispanic Black, 3.9% Asian, and 5.1% other or multiracial. In terms of BC risk factors, 4.5% of the women had extremely dense breasts, 16.0% had a family history of BC, 2.0% had AH or LCIS, and 1.6% had a genetic predisposition for BC. Overall, only 3.6% of women received a supplemental breast MRI. In multivariable analysis, younger age was significantly associated with breast MRI receipt, with women aged 25-34 demonstrating higher odds of undergoing supplemental breast MRI (odds ratio [OR]=2.59; 95% confidence interval [CI]=1.77-3.72) compared to those aged 45-54. Compared to non-Hispanic White women, Asian and Hispanic women had significantly lower odds of receiving a breast MRI (OR=0.72, 95% CI=0.63-0.82 and OR=0.66, 95% CI=0.51-0.84, respectively). Having extremely dense vs. less dense breasts on mammography (BIRADS D vs. B) was associated with higher breast MRI uptake (OR=1.99, 95% CI=1.35-2.92). Compared to women with a diagnostic code for at high risk for BC, those in other risk categories had significantly higher odds of receiving a screening breast MRI: family history of BC (OR=7.14, 95% CI=6.36-8.02), genetic predisposition to BC (OR=26.6, 95% CI=22.30-31.70), and high-risk breast lesions (OR=28.1, 95% CI=23.70-33.30). We did not observe differences in breast MRI use among Black vs. White women or based upon health insurance status (private vs. public). Conclusion: In a racially/ethnically diverse cohort of women at high risk for BC, only 3.6% received a supplemental breast MRI. Utilization varied significantly by age, race/ethnicity, and specific BC risk factors. These findings highlight persistent gaps in the uptake of guideline-concordant supplemental BC screening. We plan to study the impact of breast MRI use on early detection of local disease and screen-detected vs. interval BC. Future research should investigate patient-, provider- and system-level barriers to adoption of supplemental breast MRI among high-risk women to promote equitable access to advanced screening modalities. Citation Format: A. Michel, J. Zahlan, K. D. Crew. Factors Associated with Receipt of Magnetic Resonance Imaging for Breast Cancer Screening Among Racially/Ethnically Diverse High-Risk Women [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS1-06-04.

Abstract Introduction: Vacuum-assisted excision (VAE) of breast lesions is a technique used for diagnostic and therapeutic purposes and is performed on an outpatient basis, with local anesthesia and image guidance. Currently, VAE is used in the management of benign lesions and lesions of uncertain malignant potential (B3 lesions). More recently, there has been interest in the application of VAE for the percutaneous treatment of small breast cancers with the aim of reducing morbidity and aggressive surgical treatment of breast cancers detected by screening programs. The way conventional VAE is performed, histopathological assessment of resection margins is not possible. Obtaining free margins after a breast cancer resection is a primary objective in the surgical treatment of this disease. If VAE could ensure free margins and absence of residual tumor in the resection cavity, this would also be a method safely used for minimally invasive treatment, providing an effective percutaneous treatment of early breast cancers. Objectives: evaluate the effectiveness of vacuum-assisted excision associated with percutaneous sampling of cavity margins for the complete resection of breast cancers ensuring the absence of residual disease in surgical pathology. Methods: Inclusion criteria are women with unifocal lesions smaller than 1.5 cm, Category 4 or 5 ACR BI-RADS™, identified by screening or clinical alteration. Non-inclusion criteria are multifocal and multicentric breast cancers, breast cancers associated with diffuse and extensive microcalcifications. Sensitivity, specificity, accuracy, positive predictive value, negative predictive value, false negative and false positive rate of vacuum-assisted excision associated with percutaneous sampling of cavity margins for the complete resection of breast cancers will be calculated. The data collected will also encompass demographics, characteristics of the lesion, and information regarding the biopsy, surgical pathology, and surgical procedures, as well as data on side effects, patient acceptance, cosmetic results and patients’ experience during VAE (BreastQ questionnaire). Ethics and dissemination: Ethics approval has been obtained by the Brazilian National Research Ethics Commission (CONEP) 61781922.5.0000.5109 and is registered in Brazilian Registry of Clinical Trials (https://ensaiosclinicos.gov.br/), identifier U1111-1301-4235. Participants will provide written informed consent, and researchers will follow institutional guidelines for data collection and management. Citation Format: H. L. Couto, B. A. Coelho, T. C. Oliveira, B. F. Ricardo, P. H. Toppa, D. d. Pires, S. d. Ferreira, L. B. Oliveira, A. C. Mendonça, R. G. Saliba, T. P. Moraes, P. C. Soares, C. A. Padua, G. F. Cunha Júnior, M. S. Castilho, L. L. Dominguez, J. S. Oliveira, A. C. Oliveira, D. R. Siqueira, B. A. Pires, A. Mattar, G. d. Silva Junior, M. Antonini, E. C. Pessoa, F. M. Reis, BreastMit Collaborative Research Group. Vacuum Assisted Excision (VAE): A single-step approach to the diagnosis and percutaneous treatment of Early Breast Cancer (THE VAE BReast 01 TRIAL) [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS5-08-02.

Abstract Background: Breast cancer (BC) chemoprevention with selective estrogen receptor modulators (SERMs) and aromatase inhibitors (AIs) remains underutilized among high-risk women. Improving knowledge about BC risk and chemoprevention among high-risk patients and their healthcare providers may enhance informed decision-making about this BC risk reduction strategy. Methods: We conducted a cluster randomized controlled trial to evaluate the effectiveness of web-based patient and provider decision support, RealRisks and BNAV, to improve chemoprevention informed choice among women with atypical hyperplasia (AH) or lobular carcinoma in situ (LCIS). Thirty-one sites, including academic and community practices across the U.S., were randomized to standard patient educational materials alone or combined with RealRisks and BNAV. A total of 412 patients and 210 healthcare providers were enrolled. Patient surveys were administered at baseline, 6 and 12 months. The primary outcome at 6 months was chemoprevention informed choice, defined as adequate chemoprevention knowledge and attitudes congruent with decision. Secondary endpoints included BC risk perceptions, worry, chemoprevention knowledge, decision conflict and chemoprevention decision. Assuming 10% loss to follow up (LTFU), roughly equal accrual across sites, and an intraclass coefficient (ICC) of 0.02, we had 90% power to detect a 15% absolute difference in informed choice if the event rate in the control arm was 10%. Results: Among 24 evaluable sites, 14 were randomized to the intervention arm and 10 to the control arm. Among 210 enrolled providers, median age was 46 (range, 28-83) and 78% were female; 75% physicians and 25% advanced practice providers/nurses; 53% medical oncologists, 29% surgeons, and 10% gynecologists/family practitioners/internists; and 83% had previously prescribed chemoprevention. Among 412 enrolled patients, median age was 53 (range, 35-74), including 77% White, 8% Black, and 10% Asian race, and 14% Hispanic ethnicity. In terms of BC risk factors, 77% had AH and 23% LCIS, 34% had a first-degree family history of BC, and 65% had heterogeneously or extremely dense breasts on mammography. Thirty percent were unevaluable for the primary endpoint (1% ineligible, 26% LTFU, 3% missing data). Among 288 evaluable patients (148 at intervention sites, 140 at control sites), the proportion of patients exercising informed choice at 6 months in the intervention and control arms was 35% vs. 27%, respectively (odds ratio [OR]=1.44, 95% confidence interval [CI]=0.84-2.46, p=0.19; observed ICC=0.006). Compared to patients in the control arm at 6 months, those in the intervention arm were more likely to have accurate BC risk perceptions (OR=1.80, 95% CI=1.11-2.91) and adequate chemoprevention knowledge (OR=1.62, 95% CI=1.00-2.61). We observed no significant differences in BC worry, decision conflict, or chemoprevention decision between study arms. At 6 months, 49% of women in the control arm and 51% in the intervention arm decided to take a SERM or AI for BC chemoprevention (p=0.58), a rate higher than anticipated. Conclusions: We did not observe a significant increase in informed choice among patients and providers assigned to chemoprevention decision support compared to standard educational materials alone. Among women with AH or LCIS seen mainly by breast specialists who prescribed chemoprevention, about half decided to take a SERM or AI. Therefore, relatively high chemoprevention uptake may be achieved among women with high-risk breast lesions managed by breast specialists. Chemoprevention decision support may improve decision antecedents, such as accurate BC risk perceptions and chemoprevention knowledge, but may have a limited role in changing health behaviors. Citation Format: K. D. Crew, G. L. Anderson, K. B. Arnold, A. Michel, M. T. DeLucie, C. W. Law, S. Pruthi, A. C. Sandoval Leon, R. Shirley, M. T. Grosse Perdekamp, S. V. Colonna, S. L. Krisher, T. King, L. D. Yee, T. J. Ballinger, C. Braun-Inglis, D. A. Mangino, K. B. Wisinski, C. A. DeYoung, M. Ross, J. D. Floyd, A. Kaster, L. VanderWalde, T. Saphner, C. Zarwan, S. Lo, C. Graham, A. K. Conlin, K. J. Yost, D. M. Agnese, C. Jernigan, D. L. Hershman, M. L. Neuhouser, J. A. Zell, B. Arun, R. Kukafka. Making informed choices on incorporating chemoprevention into care (MiCHOICE): Cluster randomized controlled trial of decision support for breast cancer chemoprevention, SWOG 1904 [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS3-01-15.