Predicting recurrence risk after neoadjuvant therapy in breast cancer: a single-center retrospective cohort study integrating residual cancer burden and immune microenvironment features
4 affiliations
- 1Physical Examination Center, Wang Jiang Hospital, Sichuan University
- 2Cancer Center, West China Hospital, Sichuan University
- 3Pharmacy Department, Chengdu Eighth People’s Hospital
- 4General Inpatient Department, West China Health Care Hospital, Sichuan University
Abstract
Background: Residual cancer burden (RCB) after neoadjuvant therapy (NAT) provides prognostic information, but its combination with immune microenvironment features for recurrence prediction remains underexplored.
Methods: This single-center retrospective cohort study included 420 consecutive patients with breast cancer who underwent neoadjuvant therapy followed by definitive surgery between January 2015 and June 2024. Residual cancer burden (RCB) and post-neoadjuvant immune features, including tumor-infiltrating lymphocytes (TILs), CD8, FOXP3, and the proportion of CD163-positive among CD68-positive macrophages, were assessed. Patients were stratified by calendar era and randomly partitioned within strata into a training cohort (n=294) and an internal hold-out validation cohort (n=126). A LASSO-penalized Cox workflow was used for predictor selection, followed by unpenalized Cox refitting. Model performance was evaluated using C-statistics, time-specific Brier scores, calibration, bootstrap internal validation, and 5-year time-dependent decision curve analysis.
Results: During a median follow-up of 61 months, 126 first recurrence events were observed (88 training and 38 validation). The final combined model retained RCB category, triple-negative breast cancer (TNBC), CD8 status, and CD163-positive macrophage proportion. The combined model achieved C-statistics of 0.79 (95% CI 0.75-0.83) in the training cohort and 0.75 (95% CI 0.69-0.81) in the internal validation cohort, with corresponding 5-year Brier scores of 0.142 and 0.158. Bootstrap internal validation yielded an optimism-corrected C-statistic of 0.772 and calibration slope of 0.92. The combined model provided the greatest net benefit among the evaluated models over 5-year threshold probabilities of approximately 15%-45%.
Conclusions: Integration of RCB with TNBC status and post-neoadjuvant CD8 and CD163-positive macrophage features may provide incremental prognostic information for recurrence after neoadjuvant therapy. The model showed acceptable internal discrimination and calibration, but external validation is required before clinical implementation or treatment-directed use.
Related
- A multimodal deep learning radiomics nomogram integrating clinicopathological features for predicting pathological complete response after neoadjuvant therapy in triple-negative breast cancer: a multicenter retrospective study
- A nomogram integrating YTHDC2 expression and clinicopathological parameters for predicting endometrial cancer recurrence: a multicenter retrospective study
- Association of neoadjuvant therapy with locoregional recurrence according to recurrence definitions in breast cancer: a retrospective cohort study
- Transformer-Based Deep Learning Integrating Multi-Regional MRI and Clinical Features for Predicting Biochemical Recurrence After Radical Prostatectomy: A Multicenter Study.
- Integrating Systemic Inflammation, Nutrition and Skeletal Muscle into a Composite Score (the HALP–Sarcopenia Combined Index) for Preoperative Risk Stratification of Lymph-Node Metastasis and Recurrence in Endometrial Carcinoma: A Single-Centre Retrospective Study
BibTeX
@article{zhao2026predicting,
title = {{Predicting recurrence risk after neoadjuvant therapy in breast cancer: a single-center retrospective cohort study integrating residual cancer burden and immune microenvironment features}},
author = {Zhihua Zhao and Rui Deng and Liangliang Bai and Xiaogang Duan and Jingyuan Yi},
journal = {Frontiers in Oncology},
year = {2026},
month = oct,
doi = {10.3389/fonc.2026.1925326},
url = {https://www.frontiersin.org/articles/10.3389/fonc.2026.1925326/full},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Predicting recurrence risk after neoadjuvant therapy in breast cancer: a single-center retrospective cohort study integrating residual cancer burden and immune microenvironment features
AU - Zhihua Zhao
AU - Rui Deng
AU - Liangliang Bai
AU - Xiaogang Duan
AU - Jingyuan Yi
PY - 2026
DA - 2026/10/14
JO - Frontiers in Oncology
DO - 10.3389/fonc.2026.1925326
UR - https://www.frontiersin.org/articles/10.3389/fonc.2026.1925326/full
AB - Background: Residual cancer burden (RCB) after neoadjuvant therapy (NAT) provides prognostic information, but its combination with immune microenvironment features for recurrence prediction remains underexplored. Methods: This single-center retrospective cohort study included 420 consecutive patients with breast cancer who underwent neoadjuvant therapy followed by definitive surgery between January 2015 and June 2024. Residual cancer burden (RCB) and post-neoadjuvant immune features, including tumor-infiltrating lymphocytes (TILs), CD8, FOXP3, and the proportion of CD163-positive among CD68-positive macrophages, were assessed. Patients were stratified by calendar era and randomly partitioned within strata into a training cohort (n=294) and an internal hold-out validation cohort (n=126). A LASSO-penalized Cox workflow was used for predictor selection, followed by unpenalized Cox refitting. Model performance was evaluated using C-statistics, time-specific Brier scores, calibration, bootstrap internal validation, and 5-year time-dependent decision curve analysis. Results: During a median follow-up of 61 months, 126 first recurrence events were observed (88 training and 38 validation). The final combined model retained RCB category, triple-negative breast cancer (TNBC), CD8 status, and CD163-positive macrophage proportion. The combined model achieved C-statistics of 0.79 (95% CI 0.75-0.83) in the training cohort and 0.75 (95% CI 0.69-0.81) in the internal validation cohort, with corresponding 5-year Brier scores of 0.142 and 0.158. Bootstrap internal validation yielded an optimism-corrected C-statistic of 0.772 and calibration slope of 0.92. The combined model provided the greatest net benefit among the evaluated models over 5-year threshold probabilities of approximately 15%-45%. Conclusions: Integration of RCB with TNBC status and post-neoadjuvant CD8 and CD163-positive macrophage features may provide incremental prognostic information for recurrence after neoadjuvant therapy. The model showed acceptable internal discrimination and calibration, but external validation is required before clinical implementation or treatment-directed use.
SN - 2234-943X
ER -