RPS 1702 - Predicting response and recurrence in breast cancer

Author Block: F. Ramos Barra¹, V. A. Ohannesian², P. Q. Monteiro³, M. S. Dias⁴; ¹Brasilia/BR, ²São Paulo/BR, ³Salvador/BR, ⁴Goiania/BR
Purpose: This systematic review and meta-analysis aimed to evaluate the diagnostic accuracy of early, mid-treatment Contrast-Enhanced Ultrasound (CEUS) for predicting pathological Complete Response (pCR) in patients with primary, non-metastatic breast cancer undergoing Neoadjuvant Chemotherapy (NAC).
Methods or Background: Following PRISMA guidelines, we systematically searched PubMed, Embase, and Web of Science for studies assessing the predictive accuracy of CEUS after one or more cycles of NAC. The reference standard for pCR varied across studies (absence of invasive cancer or Miller-Payne Grade ≥4). A potential source of heterogeneity was addressed by selecting the best performing CEUS parameter from each study for analysis.
Results or Findings: Seven eligible studies, comprising 979 patients, were included in the bivariate meta-analysis. The overall diagnostic accuracy of early CEUS for predicting pCR was high, demonstrated by an Area Under the Summary Receiver Operating Characteristic Curve (AUROC) of 0.90 (95% CI: 0.87–0.93) and a Diagnostic Odds Ratio (DOR) of 25 (95% CI: 13, 48). The pooled sensitivity and specificity were 0.83 (95% CI: 0.74, 0.89) and 0.84 (95% CI: 0.76, 0.90), respectively, with low heterogeneity (I2=8.4%). Subgroup analysis indicated that assessment after 2nd or 3rd cycles offered slightly higher sensitivity (0.86; 95% CI: 0.76–0.96 versus 0.80; 95% CI: 0.68-0.92) than assessment at the end of the 1st cycle.
Conclusion: Early monitoring of NAC response in breast cancer using CEUS demonstrates high diagnostic accuracy for predicting pCR. This non-invasive tool is valuable for guiding mid-treatment decisions and tailoring subsequent therapeutic strategies for patients.
Limitations: Limitations included the variability in the pathological complete response definition used across studies (absence of invasive cancer or Miller-Payne Grade ≥4) and the methodological choice to utilize the best-performing CEUS parameter from each study to synthesize the findings.
Funding for this study: None
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: Ö. Ezmeci, S. Kul; Trabzon/TR
Purpose: Pathological complete response (pCR) after neoadjuvant chemotherapy (NAC) has gained importance as a favorable prognostic indicator in breast cancer (BC). Pretreatment imaging biomarkers may predict tumors likely to achieve pCR and support personalized NAC planning. This study investigated the role of pretreatment contrast-enhanced mammography (CEM) findings in predicting the NAC response in BC.
Methods or Background: This retrospective, ethics-approved single-center study (2021–2024) included BC patients treated with NAC who had pretreatment CEM and surgical pathology. Patients who did not complete NAC, did not undergo surgery or lacked histopathological results were excluded. Pre-treatment CEM images were evaluated for breast density, background enhancement (BPE), lesion size, shape, margin, internal enhancement, conspicuity, and presence of calcifications using BI-RADS lexicon. Tumor contrast-to-noise ratio (CNR) was calculated from early-phase recombined images using the formula: (mean tumor SI – mean background SI) / standard deviation of background SI). Tumor enhancement kinetics were assessed by visually comparing early and delayed phase images. Imaging parameters were correlated with response status. Chi-square and Mann–Whitney U tests were used (p<0.05).
Results or Findings: In 15 (26%) of the 58 cases pCR was achieved. The relationship between pre-treatment CEM parameters and the presence of pCR was documented in Table 1. Tumor visibility and CNR values were significantly higher in patients with pCR (p<0.049) (Fig 1). pCR rates were 0%, 24%, and 43% in tumors with type 1, 2, and 3 enhancement kinetics, respectively (p=0.025) (Fig 2). Other CEM parameters showed no correlation with pCR.
Conclusion: This study suggests that CEM-derived tumor enhancement intensity and kinetics may serve as markers in predicting NAC response. Previous radiomics-based studies also support these findings (2-4).
Limitations: Small cohort and lack of inter-reader analysis.
Funding for this study: No funding was received.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Approved by the Ethics Committee of Karadeniz Technical University, Faculty of Medicine (approval no: 2024/280, year 2025)

Author Block: C. Bellini¹, G. Bicchierai¹, F. Amato², C. Maiello¹, F. Di Naro¹, D. De Benedetto¹, S. Vidali¹, V. Miele¹, J. Nori¹; ¹Florence/IT, ²Agrigento/IT
Purpose: To evaluate which contrast-enhanced mammography (CEM) acquisition protocol—early, delayed, or combined—is most accurate in assessing pathological complete response (pCR) after neoadjuvant chemotherapy (NAC) in breast cancer.
Methods or Background: This retrospective study included 202 women who underwent CEM after NAC and before surgery, with both early and delayed acquisitions. Pathological tumor status (ypT) at surgery was the gold standard, with pCR defined as ypT0 or ypTis. Two breast radiologists reviewed images in consensus. Early, delayed, and combined image sets were assessed in randomized sessions. Radiologic response was categorized per RECIST criteria and then dichotomized as complete response vs residual disease. Sensitivity, specificity, predictive values, accuracy, and ROC-AUCs were calculated and compared.
Results or Findings: Sensitivity was higher for combined (86.6%) and delayed (85.7%) acquisitions than for early images (68.1%), whereas specificity was superior in early acquisitions (63.9%). Accuracy was 66.3% for early, 71.3% for delayed, and 69.8% for combined CEM. NPV improved from early (58.2%) to delayed (71.2%) and combined acquisitions (70.4%). ROC curve comparisons showed no statistically significant differences between protocols. Delayed acquisitions reclassified 57.9% of early false negatives as true positives and corrected 13.3% of early false positives but converted 28.3% of early true negatives into false positives. A relevant proportion of false positives in both protocols corresponded to residual in situ carcinoma at histology.
Conclusion: Delayed acquisitions improve sensitivity and reduce false negatives in NAC response assessment with CEM, although at the cost of decreased specificity due to overestimation of residual disease, particularly in cases with in situ carcinoma.
The combined protocol enhances overall performance but does not significantly differ from early or delayed acquisitions alone.
Limitations: Limitations include the retrospective, single-center design, consensus readings without interobserver analysis, absence of comparison with MRI.
Funding for this study: No funding
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Comitato Etico Area Vasta Centro 16.251

Author Block: G. Barcaroli, F. Galati, F. Shakki Katouli, R. Maroncelli, F. Cicciarelli, M. Pasculli, F. Pediconi, C. Catalano; Rome/IT
Purpose: Neoadjuvant therapy (NAT) in breast cancer enables tumor downstaging and response evaluation. Although magnetic resonance imaging (MRI) is the gold standard, its limitations have prompted growing interest in alternatives such as contrast-enhanced mammography (CEM).
Methods or Background: This review was conducted according to PRISMA guidelines. We evaluated CEM accuracy in predicting pathological complete response after NAT, using surgery as reference standard. A comprehensive search was conducted in PubMed/Medline, Embase, Cochrane, and Web of Science. Two reviewers screened studies, extracted data, and evaluated risk of bias with QUADAS-2 tool. A bivariate random-effects model estimated diagnostic odds ratio (DOR), sensitivity, specificity, and likelihood ratios. Summary receiver operating characteristic (sROC) curves were generated, and heterogeneity was quantified using Higgins’I². Publication bias was assessed via Deek’s funnel plot.
Results or Findings: 13 studies, including 582 patients, met inclusion criteria. The pooled analysis showed a DOR of 8.60 (95%,CI:3.29–22.52) with substantial heterogeneity (I²=70.06%). Pooled sensitivity and specificity were 72% (95%,CI:63%–79%) and 83% (95%,CI:68%–92%), respectively, with an sROC AUC of 0.82 (95%,CI:0.78–0.85). Excluding the study contributing most to heterogeneity improved the DOR to 9.76 (95%,CI:6.00–15.89) and reduced I² to 0.00%.
Conclusion: CEM demonstrates good diagnostic accuracy for assessing NAT response, performing comparably to MRI.
Limitations: Retrospective study, limited number of studies included
Funding for this study: None
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: L. Lin, J. Zheng, C. You, Y. Gu; Shanghai/CN
Purpose: Neoadjuvant chemotherapy combined with anti-programmed cell death protein-1 (PD-1) immunomodulators (NACI) has been proven to improve outcomes in triple-negative breast cancer (TNBC). However, the role of MRI in predicting therapeutic response remains underexplored. This study aims to investigate the utility of MRI for early response prediction and longitudinal assessment of reactive angiomatous changes in TNBC patients undergoing NACI therapy.
Methods or Background: This secondary analysis pooled data from the CamRelief and NeoTENNIS RCTs, among others, investigating NACI at our center (2020-2023). We analyzed baseline clinicopathological characteristics, MRI features (including BI-RADS descriptors and whole-tumor ADC histogram metrics), and response outcomes. Independent predictors of pathological complete response (pCR) were identified via univariate and multivariate logistic regression to develop a prognostic model. Longitudinal evolution of reactive angiomas during NACI was assessed with non-parametric and chi-square tests.
Results or Findings: A total of 61 TNBC patients undergoing NACI were enrolled. Univariate analysis identified four independent predictors of pCR. While elevated Ki-67 index (p<0.001) and CD8-T cell infiltration (p=0.004) reflected baseline pathological predictors, type III kinetic curves (p=0.004) and reduced median_ADC value (p=0.021) of MRI provided unique insights with pCR. Moreover, the combined multi-dimensional model achieved a high prediction accuracy, got AUC of 0.89. Serial MRI monitoring identified transient reactive changes characterized by multiple small masses emerging (mean 8.82±0.67mm) during NACI unrelated to response, peaking at post-cycle-2 MRI, followed by progressive regression. One was histopathologically confirmed as benign reactive angiomatous remodeling.
Conclusion: MRI played a dual role in managing NACI for TNBC. An integrative model of MRI biomarkers and clinicopathology superiorly predicted pCR, whereas reactive angiomatous remodeling was transient.
Limitations: Small sample size and and paucity of quantitative features on MRI.
Funding for this study: Special Research Fund for Clinical Studies of Innovative Drugs after Market Launch (WKZX2024CX103301)
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the Ethics Committee of the Institutional Review Board (IRB) of our institution. All patients provided written informed consent before enrollment.

Author Block: C. Yanardağ, M. G. Akpınar, G. Durhan, F. Demirkazık; Ankara/TR
Purpose: This study evaluates whether pre-treatment breast Magnetic Resonance Imaging (MRI)-derived radiomic features, alone and combined with radiologic and pathologic variables, can predict pathologic complete response (pCR) before neoadjuvant therapy (NAT) in breast cancer.
Methods or Background: We retrospectively included 154 patients who underwent pre-treatment breast MRI, received NAT, and subsequently had surgery at our center (2016-2024). Patients were split 70/30 into training (n=109) and test (n=45) sets. Pathologic variables (estrogen receptor, progesterone receptor, HER2 status, and histologic grade) were obtained from pre-treatment biopsies. Radiologic variables included tumor size, distribution, shape, and axillary lymph-node status. Variables showed balanced distributions between training and test sets (p>0.05). Tumors were manually volumetrically segmented on early post-contrast fat-suppressed T1-weighted images with syngo.via. We extracted 1,691 radiomic features, retained 652 after applying an intraclass correlation coefficient threshold of 0.90 (one reader, repeated segmentations, 38-case subset). Radiomic features underwent elastic-net selection in the training set with cross-validation, yielding nine features. Pure radiomic, radiologic, and pathologic, as well as combined (radiomics-radiologic, radiomics-pathologic, and radiomics-radiologic-pathologic) models were developed using a Support Vector Machine (SVM). The area under the curve (AUC), accuracy, sensitivity, and specificity were calculated for both sets.
Results or Findings: A total of 42.9% had pCR. Test-set AUCs: radiomics 0.563; radiologic 0.558; pathologic 0.704; radiomics+radiologic 0.557; radiomics+pathologic 0.662; radiomics-radiologic-pathologic 0.607. The pathologic model had the highest AUC. Specificities: radiomics 0.846; radiologic 0.807; pathologic 0.692; radiomics+radiologic 0.846; radiomics+pathologic 0.539; radiomics-radiologic-pathologic 0.654. Radiomics-only and radiomics+radiologic models showed higher specificity.
Conclusion: Radiomics alone offered modest discrimination, but adding pathologic variables improved performance. Radiomic models have higher specificity, useful for ruling out pCR. Combining radiologic and pathologic findings before treatment can help with risk stratification and customizing therapy prior to NAT.
Limitations: A modest sample size, retrospective, single-center design.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This study received approval from the Hacettepe University Health Sciences Research Ethics Committee (decision no. SBA 24/492) and was conducted in accordance with the Declaration of Helsinki.

Author Block: E. Tallamona¹, R. Gioco², C. Giuseppe², P. Francesco²; ¹Gela/IT, ²Misterbianco/IT
Purpose: This study aims to identify radiological features along with clinical and pathological predictors that influence the choice of mastectomy in non-metastatic breast cancer patients who demonstrate a pathological partial response (pPR) after neoadjuvant systemic therapy (NACT), despite presenting a favorable tumor-to-breast ratio that could allow breast-conserving surgery (BCS).
The goal is to better understand factors that lead to over-treatment and to prove how radiologists could help to reach a more evidence-based surgical approach.
Methods or Background: This retrospective single-center study included 165 female patients treated with NACT followed by surgery between 2021 and 2023 at Humanitas Istituto Clinico Catanese in Catania. Only patients with residual cancer burden class I or II were included. Patients with RCB III, metastatic disease, BRCA mutations, or male gender were excluded. For a subgroup of 26 patients, a tumor-to-breast volume ratio was calculated using mammographic measurements before and after NACT to determine suitability for BCS. Statistical analysis was performed to identify correlations between radiological measurement, clinical characteristics and surgical choice (BCS vs mastectomy).
Results or Findings: Among 165 patients with partial pathological response (pPR), 43% underwent BCS and 57% underwent mastectomy. Mastectomy was more frequently performed in younger and premenopausal patients, those with multifocal or larger tumors, and higher clinical staging at diagnosis. In the subgroup with a post-NACT favorable tumor-to-breast ratio, 38.5% still underwent mastectomy.
Conclusion: Despite partial tumor response and anatomical feasibility, many patients undergo mastectomy due to factors beyond tumor biology. A multidisciplinary, patient-centered approach, integrating predictive tools like tumor-to-breast ratio analysis and shared decision-making, may promote a more conservative and personalized surgical strategy leading to reduce unnecessary mastectomies and enhance patient outcomes and satisfaction
Limitations: Main limitation include the single-center design and limited sample size
Funding for this study: No funding was required for this study
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: R. Timóteo, A. Laborde, Y. Forghani, N. Loução, A. Cardoso, M. Correia, J. Santinha, T. Marques, P. Gouveia; Lisboa/PT
Purpose: Breast cancer (BC) is the second most common cancer worldwide, and breast conservative surgery (BCS) is its primary treatment. Accurate tumor localization is essential to achieve negative margins and reduce recurrence, yet current methods are invasive and often imprecise, leading to re-excision. Digital twins (DTs) are emerging in healthcare as dynamic virtual replicas of patients, integrating both anatomy and physiology, and serving as a powerful tool for treatment, decision-making, and outcome prediction. By providing a non-invasive and precise approach to tumor localization, DTs may enhance surgical accuracy, reduce re-excision rates, and ultimately improve patient outcomes.
Methods or Background: Four breast cancer patients proposed for BCS were selected between January and September 2025. Patient-specific DTs were built from annotated supine breast MRI and automatically aligned to the patient in real-time with a 3D surface scanning system in the operating room (OR). DTs were visualized through an Augmented Reality (AR) interface that was co-designed with breast surgeons to ensure suitability for the OR environment. Tumor marking with AR was compared to carbon tattoo marking and intraoperative findings.
Results or Findings: Results reveal a high degree of tumor localization accuracy in most surgeries, with fast setup time and tumor marking. Usability and workload questionnaires highlighted our system’s excellent user experience.
Conclusion: Integrating DTs in AR environments represents the next era of breast cancer surgery, offering surgeons a new paradigm for interacting with patient-specific data, enhancing their performance in the OR, and leading to better surgical outcomes.
Limitations: Future work should consider a clinical study with a larger patient cohort, accuracy measurement through rate of positive margins, non-rigid registration of DTs with volumetric correction, and biomechanical simulation capabilities.
Funding for this study: This research is part of the Health for Portugal (HfP), funded by Agendas Mobilizadoras para a 5 Inovação Empresarial - Plano de Recuperação e Resiliências Português (PRR). A special thanks to the Breast Unit at Champalimaud Clinical Center for their collaboration, and to all of those who participated in this study, to contribute to the evolution of health technologies and consequent patient care.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This study was approved by the ethics committee of Champalimaud Foundation.

Author Block: A. G. Azzarito, M. C. Torrione, A. Delli Pizzi, M. F. Savina, L. Di Pietrantonio, A. Di Credico, A. Figorilli, M. Muzi, M. Caulo; Chieti/IT
Purpose: To evaluate the local recurrence rate of pure ductal carcinoma in situ (DCIS) after surgery, with particular focus on the impact of adjuvant radiotherapy and clinicopathological factors, in order to explore the potential for more personalized management strategies.
Methods or Background: We retrospectively analyzed 403 patients with histologically confirmed pure DCIS treated with breast-conserving surgery or mastectomy between 2014 and 2023. Imaging and pathological variables were assessed. Recurrence rates were correlated with tumor grade, radiotherapy, and other clinicopathological parameters. Statistical analyses included Kaplan–Meier survival estimates and multivariate Cox regression.
Results or Findings: Among 417 lesions, 21 ipsilateral recurrences (5%) were observed, of which 57% were non-invasive and 38% invasive. Recurrences occurred predominantly in patients not treated with adjuvant radiotherapy (52%) and in those with breast-conserving surgery. All recurrent tumors were estrogen receptor–positive, yet none received endocrine therapy. Tumor grade and size were not significantly associated with recurrence. No distant metastases or disease-related deaths were recorded.
Conclusion: Recurrence after treatment for pure DCIS is relatively uncommon and often non-invasive. Traditional prognostic markers such as tumor grade and size did not predict recurrence. These findings support a more individualized approach to treatment, where selected patients may avoid overtreatment. Further studies incorporating molecular profiling are warranted to refine risk stratification.
Limitations: This study is limited by its retrospective design and by the relatively short follow-up in some patients, which may not capture late recurrences. Moreover, the absence of molecular or genomic profiling restricts the ability to refine risk stratification.
Funding for this study: None
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: