Neoadjuvant chemotherapy (NAC) is increasingly used for high-risk breast cancer to achieve pathologic complete response (pCR), an indicator of event-free survival and favorable survival outcomes. Integrating MRI and Ki-67 biomarker analysis into predictive models offers a promising approach to optimize NAC response assessment and guide personalized treatment strategies. This study evaluates the validity of combined MRI and Ki-67 metrics for predicting pCR. We conducted a systematic review and meta-analysis following Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines, including studies on NAC-treated breast cancer patients assessed by MRI and Ki-67. The predictive models were evaluated based on key parameters, including MRI-based tumor size reduction and Ki-67 levels, with outcomes measured by area under the receiver operating characteristic curve (AUC) and calibration metrics. Findings across ten studies consistently show that high Ki-67 levels and significant tumor size reduction on MRI are predictive of pCR, achieving AUCs near 0.90. The analysis highlighted that models integrating MRI with Ki-67 metrics outperformed single-modality approaches, showing enhanced predictive accuracy and calibration. However, high heterogeneity (I² = 77%) was noted, suggesting variability in imaging and Ki-67 assessment protocols across studies. Thisstudy underscores the combined utility of MRI and Ki-67 for the non-invasive prediction of pCR, offering both structural and biological insights into tumor responsiveness. The results align with prior research, affirming the role of Radiomic-clinicopathological models in providing a more comprehensive assessment compared to individual markers. Further refinement of imaging and biomarker protocols could improve model reproducibility and applicability. Our findings highlight the robust predictive accuracy of MRI-Ki-67 integrated models for assessing pCR, marking a significant step toward personalized cancer care. Future studies should focus on refining these models with additional biomarkers and standardized protocols, facilitating their integration into routine clinical oncology to enhance treatment decision-making and patient outcomes.
Read full abstract