Prediction of pathological complete response to neoadjuvant chemotherapy for invasive breast cancers based on longitudinal ultrasound and superb microvascular imaging: a single-center retrospective study

Yanling Zuo, Yongtao Zhan, Jie Zhou, Haoming Xia, Tao Li, Fan Zhou, Chunyue Luo, Huiai Zeng, Yingjia LiPeerJ 2025 · DOI

TL;DR

This study introduces a novel ultrasound-based approach to predict pathological complete response (pCR) in breast cancer patients undergoing neoadjuvant chemotherapy (NAC). By analyzing dynamic changes in tumor size and vascularity using conventional ultrasound (US) and superb microvascular imaging (SMI), the researchers developed a predictive model that significantly outperforms individual parameters, offering a promising tool for personalized treatment adjustments.

What Problem Does This Paper Address?

Predicting how well a breast cancer patient will respond to neoadjuvant chemotherapy (NAC) is crucial for tailoring treatment and improving outcomes. Pathological complete response (pCR), meaning no residual invasive cancer after treatment, is a key indicator of long-term survival. However, accurately identifying patients likely to achieve pCR early in the treatment course remains a challenge. While MRI is a sensitive imaging modality, its accessibility is limited. Ultrasound (US) offers a more convenient, cost-effective, and radiation-free alternative. This paper aims to leverage advanced US techniques, specifically superb microvascular imaging (SMI), to provide early and accurate predictions of pCR, thereby enabling timely treatment modifications and avoiding unnecessary interventions.

Methods

The researchers conducted a single-center retrospective study involving 107 patients with invasive breast cancer who underwent NAC. They analyzed ultrasound (US) and superb microvascular imaging (SMI) data collected before NAC and after two cycles of treatment.

Key imaging parameters analyzed included:

  • Conventional US: Longest diameter (LD) and shortest diameter (SD) of the lesion, and the presence of an echogenic rind.
  • Superb Microvascular Imaging (SMI):
    • Color SMI (cSMI): Vascular index (VI), which quantifies the ratio of Doppler signal to the total lesion area.
  • Monochrome SMI (mSMI): Lesion scope (comparing mSMI to grayscale US) and its expansion.

The study calculated the change (Δ) in LD, SD, and VI after two cycles of NAC. Multivariate logistic regression, employing a forward selection and Wald test, was used to identify the most significant predictors of pCR. The predictive performance of the model was evaluated using receiver operating characteristic (ROC) curves and area under the curve (AUC) analysis, with internal validation performed through bootstrap resampling. A nomogram was also developed to visually represent the predictive model.

Datasets

The study utilized data from 107 biopsy-proven invasive breast cancer patients treated at the Affiliated Cancer Hospital and Institute of Guangzhou Medical University. This constitutes a single-center retrospective dataset. No new benchmark datasets were introduced.

Key Insights and Ideas

  • Dynamic Changes Matter More: While baseline tumor characteristics (like ER/PR negativity and HER2 positivity) were associated with pCR, the changes in tumor size and vascularity after just two cycles of NAC proved to be more powerful predictors.
  • Shortest Diameter (SD) is a Key Player: Both the baseline SD and, more importantly, the rate of change in SD (ASD) emerged as strong independent predictors of pCR. This suggests that how uniformly the tumor shrinks is a critical indicator of treatment response.
  • Vascularity Changes are Predictive: The rate of change in vascular index (AVI) also demonstrated significant predictive power, highlighting the role of microvascular response to chemotherapy.
  • SMI's Added Value: The study leveraged SMI, a technique that enhances visualization of microvasculature without contrast agents, to capture subtle changes in tumor vascularity that might be missed by conventional US.
  • A Powerful Combined Model: A multivariate model incorporating ASD (%), SD, and AVI (%) achieved a remarkably high AUC of 0.963, significantly outperforming any single parameter. This suggests a synergistic effect of these parameters.

Key Results

  • Baseline Predictors: Patients achieving pCR were more likely to be ER-negative (P < 0.001), PR-negative (P = 0.005), HER2-positive (P < 0.001), and have echogenic rinds (P = 0.029). They also had significantly smaller baseline LD and SD (P = 0.001 and P = 0.003, respectively).
  • Post-NAC Predictors: After two cycles of NAC, the pCR group showed a higher proportion of non-expanded mSMI scope and disappearance of echogenic rinds (P < 0.001 and P = 0.002). Crucially, the rates of change in LD, SD, and VI were significantly higher in the pCR group (all P < 0.001).
  • Multivariate Model Performance: The combined model using ASD (%), SD, and AVI (%) achieved an AUC of 0.963. The individual AUCs for ASD (%) and AVI (%) were also strong at 0.906 and 0.855, respectively.
  • Internal Validation: Bootstrap resampling confirmed the model's excellent discrimination and calibration, indicating its robustness.

Limitations and Future Work

The authors acknowledge several limitations:

  • Single-Center Retrospective Study: This limits generalizability, and a larger, multi-center prospective study is needed for validation.
  • Lack of Molecular Subtype Analysis: While molecular subtypes are known to influence NAC response, a detailed stratified analysis was not performed.
  • 2D Imaging Limitations: The reliance on 2D ultrasound makes it challenging to consistently select the exact same section for VI measurement across different treatment cycles.
  • No Volumetric Imaging: Future research should incorporate volumetric imaging to provide a more comprehensive assessment of microvascular changes.

The authors suggest that future work should focus on validating this model with larger datasets and exploring volumetric imaging techniques to further enhance predictive accuracy. The developed nomogram offers a practical tool for clinical application, but further validation is essential before widespread adoption.

Summary updated

Abstract

The baseline status of US and SMI, as well as the longitudinal changes, demonstrated good predictive performance for pCR in BC following NAC.
BibTeX
@inproceedings{zuo2025prediction,
  author = {Yanling Zuo and Yongtao Zhan and Jie Zhou and Haoming Xia and Tao Li and Fan Zhou and Chunyue Luo and Huiai Zeng and Yingjia Li},
  title = {{Prediction of pathological complete response to neoadjuvant chemotherapy for invasive breast cancers based on longitudinal ultrasound and superb microvascular imaging: a single-center retrospective study}},
  booktitle = {PeerJ},
  year = {2025},
  doi = {10.7717/peerj.20171},
}
Figure