SABCS China Voice | Prof. Mao Xiaoyun: A Nomogram Model Based on Mammographic Imaging to Accurately Predict NAC Efficacy

Editor’s Note The 48th San Antonio Breast Cancer Symposium (SABCS) was held in San Antonio, USA, from December 9 to 12, 2025. A retrospective study led by Professor Mao Xiaoyun from the First Affiliated Hospital of China Medical University, with principal investigators Wang Shijin and Wei Hongrui, was presented as a poster at the meeting (Abstract No. P4-05-18). This study developed and validated a nomogram model based on pre-treatment mammographic imaging features combined with clinicopathological factors to predict the probability of pathological complete response (pCR) after neoadjuvant chemotherapy (NAC) in patients with breast cancer. The model demonstrated excellent predictive performance and calibration, and decision curve analysis showed a significant net clinical benefit across a wide range of threshold probabilities. These findings provide a powerful tool for the pre-treatment prediction of NAC efficacy and support individualized treatment decision-making. Oncology Frontier invited Professor Mao Xiaoyun to introduce the study and Professor Xu Yan from Daping Hospital of the Army Medical University to provide expert commentary.

Study Overview

Background Breast cancer remains one of the most common malignancies and a leading cause of cancer-related death among women worldwide, posing a serious threat to women’s health. Neoadjuvant chemotherapy has become a core component of comprehensive breast cancer management, playing a crucial role in tumor downstaging, increasing breast-conserving surgery rates, identifying high-risk patients, and improving long-term outcomes. Accurately predicting a patient’s response to NAC is essential for optimizing individualized treatment strategies, reducing unnecessary toxicity from ineffective therapy, avoiding waste of medical resources, and improving quality of life.

Currently, NAC response is assessed using clinical examination, ultrasound, mammography, and MRI, but the predictive value of any single modality or indicator is limited. Mammography, as a routine and widely available tool for breast cancer screening and diagnosis, provides density information that is closely related to tumor cellularity, stromal components, and vascular distribution, and may indirectly reflect tumor chemosensitivity.

Based on this rationale, the present study innovatively proposed a quantitative imaging indicator, the “primary tumor-to-glandular density ratio,” derived from mammographic images. By integrating this novel parameter with established clinicopathological factors, the authors aimed to construct and validate a precise and practical nomogram model for predicting NAC efficacy, thereby providing robust evidence to support pre-treatment individualized therapeutic planning.

Methods A total of 387 patients with breast cancer who received NAC between January 2020 and January 2024 were retrospectively included. Clinical and pathological data, along with pre-NAC mammographic images, were collected. Image processing was performed using 3D Slicer software (https://www.slicer.org/), with semi-automatic segmentation of the tumor region and normal glandular tissue to calculate the gray-scale density ratio between the two. Based on the Miller–Payne grading system after NAC, patients were classified into pCR and non-pCR groups. Univariate analysis, multivariate logistic regression, and LASSO regression were used to identify predictive factors, which were then incorporated into predictive models and visualized as nomograms. Model performance, accuracy, and clinical utility were evaluated using receiver operating characteristic curves, calibration plots, decision curve analysis, and bootstrap resampling.

Results Among the 387 included patients, 184 achieved pCR. Multivariate analysis identified the primary tumor-to-glandular density ratio, estrogen receptor (ER) status, HER2 status, and Ki-67 index as independent predictors of pCR. Model 3, which incorporated the density ratio together with ER, HER2, and Ki-67, demonstrated the best performance, with an area under the curve (AUC) of 0.852 (95% CI: 0.813–0.890). After bootstrap validation, the AUC remained 0.850 and the concordance index was 0.8503. Calibration curves showed excellent agreement between predicted probabilities and observed pCR rates, and the Hosmer–Lemeshow test indicated good model fit. Decision curve analysis revealed that the model provided significant net clinical benefit across a wide threshold probability range of 0.05 to 0.95.

Based on Model 3, a nomogram was constructed that translates the regression coefficients of the four predictors into an intuitive scoring system. Clinicians can calculate an individual patient’s total score and corresponding pCR probability using readily available variables. In addition, an interactive web-based application was developed (https://hongruiwei.shinyapps.io/dynnomapp/), allowing users to input patient data via sliders to generate personalized pCR predictions with 95% confidence intervals, further enhancing clinical usability.

Expert Commentary

Professor Xu Yan This nomogram study, which integrates pre-treatment mammographic imaging features with clinicopathological indicators, demonstrates notable innovation and strong clinical potential. By leveraging routine and widely accessible mammographic images, the authors introduced and quantified the novel parameter of the tumor-to-glandular density ratio. This metric indirectly reflects the microstructural and biological characteristics of the tumor by comparing the density of the cancerous lesion with that of normal glandular tissue. When combined with established biomarkers such as ER, HER2, and Ki-67, it forms a comprehensive and accurate predictive framework.

Importantly, the tumor-to-glandular density ratio, as a key independent predictor, showed a negative association with pCR. Its inclusion significantly improved the model’s discriminative ability, increasing the AUC from 0.769 to 0.852, which represents a substantial enhancement in predictive performance. The development of an online interactive tool further facilitates clinical implementation and real-time decision support, representing a critical step toward translating research findings into practice. Future studies are encouraged to explore the underlying molecular mechanisms linking the density ratio to NAC response, incorporate additional radiomic features and emerging molecular markers, and conduct prospective and external validation across different molecular subtypes, with the ultimate goal of establishing a highly efficient clinical tool to guide individualized NAC decision-making.

Professor Xu Yan

Director, Department of Breast and Thyroid Surgery Daping Hospital, Army Medical University (Army Specialty Medical Center)

Professor Mao Xiaoyun

Chief Physician, Doctoral Supervisor Administrative Deputy Director, Department of Breast Surgery The First Affiliated Hospital of China Medical University

Wei Hongrui