Predicting Recurrence in HR+/HER2- Early Breast Cancer

Overview

Hormone receptor-positive/human epidermal growth factor receptor 2-negative (HR+/HER2-) breast cancer constitutes approximately 70% of all breast cancer cases. Although early-stage patients generally have favorable outcomes following standard surgery and adjuvant endocrine therapy, long-term follow-up data reveal a distinct "bimodal" or "long-tail" recurrence pattern, with risks persisting for decades. Recent landmark trials (e.g., NATALEE, MonarchE) have established that combining CDK4/6 inhibitors with endocrine therapy significantly improves invasive disease-free survival (iDFS) in high-risk populations. However, the stringent enrollment criteria of these randomized controlled trials may not fully capture the heterogeneity of real-world patients. Reliance on binary cut-off values (e.g., nodal status alone) risks misclassifying biologically high-risk individuals with low anatomical burden, leading to either undertreatment or overtreatment. There is an urgent clinical need for a multidimensional, individualized risk assessment tool to guide escalated therapy decisions.

Detailed Description Background and Rationale Hormone receptor-positive/human epidermal growth factor receptor 2-negative (HR+/HER2-) breast cancer constitutes approximately 70% of all breast cancer cases. Although early-stage patients generally have favorable outcomes following standard surgery and adjuvant endocrine therapy, long-term follow-up data reveal a distinct "bimodal" or "long-tail" recurrence pattern, with risks persisting for decades. Recent landmark trials (e.g., NATALEE, MonarchE) have established that combining CDK4/6 inhibitors with endocrine therapy significantly improves invasive disease-free survival (iDFS) in high-risk populations. However, the stringent enrollment criteria of these randomized controlled trials may not fully capture the heterogeneity of real-world patients. Reliance on binary cut-off values (e.g., nodal status alone) risks misclassifying biologically high-risk individuals with low anatomical burden, leading to either undertreatment or overtreatment. There is an urgent clinical need for a multidimensional, individualized risk assessment tool to guide escalated therapy decisions. Objective To develop and internally validate a nomogram integrating clinicopathological and lifestyle factors for predicting disease-free survival (DFS) in patients with HR+/HER2- early breast cancer, and to evaluate its clinical utility using decision curve analysis (DCA). Study Design This is a single-center, retrospective cohort study based on real-world data. The study protocol has been approved by the Ethics Committee of Liaoning Cancer Hospital \& Institute, and the requirement for informed consent was waived due to the retrospective nature of the study. All data will be de-identified prior to analysis to protect patient confidentiality. Study Population Patients with histologically confirmed HR+/HER2- early-stage breast cancer who received surgical treatment at Liaoning Cancer Hospital \& Institute between July 2014 and June 2024 will be enrolled. Data Collection and Variables Demographic, clinicopathological, and treatment-related variables will be extracted from the electronic medical record system. Variables to be collected include: Demographic characteristics: age, menopausal status Pathological features: tumor size (T stage), lymph node status (N stage), histological grade, Ki-67 proliferation index, lymphovascular invasion (LVI) Treatment-related data: surgical procedure (breast-conserving surgery vs. mastectomy), chemotherapy (yes/no, regimen), endocrine therapy type (tamoxifen, aromatase inhibitor, etc.), radiotherapy (yes/no) Lifestyle factors: duration of breastfeeding (months) Follow-up data: recurrence events (locoregional, distant), second primary cancers, vital status, and date of last follow-up The primary endpoint is disease-free survival (DFS), defined as time from surgery to first recurrence (locoregional or distant), second primary cancer, or death from any cause. Quality Assurance and Data Validation A comprehensive quality assurance plan will be implemented to ensure data accuracy and completeness: Data Verification: All data entered into the electronic case report form (eCRF) will be independently double-checked by two trained research assistants. Discrepancies will be resolved by consensus or by referral to a third reviewer. Range and Consistency Checks: Data will be validated against predefined rules (e.g., date consistency: diagnosis date ≤ surgery date ≤ last follow-up date; value ranges: Ki-67 0-100%, age \>18 years). Automated logic checks will be programmed into the data entry system to flag outliers or inconsistencies. Source Data Verification: A random sample of 10% of patient records will be selected for source data verification against original medical records (paper or electronic) to assess accuracy and completeness. This will be performed by an independent data monitor not involved in primary data extraction. Data Dictionary: A comprehensive data dictionary will be developed, containing detailed descriptions of each variable, including variable name, definition, source (e.g., pathology report, operative note), coding guidelines (e.g., AJCC 8th edition for TNM staging), permissible values, and normal ranges where applicable. This dictionary will be maintained and updated throughout the study. Standard Operating Procedures (SOPs) The following SOPs will govern study operations: Patient Screening and Enrollment: SOP for identifying eligible patients from hospital databases and reviewing medical records. Data Extraction and Entry: SOP detailing procedures for manual data abstraction, coding, and entry into the secure study database. Data Management and Security: SOP for data storage, backup, access control, and de-identification to ensure compliance with institutional and national privacy regulations. Adverse Event Reporting: Although this is a retrospective observational study with no intervention, any serious adverse events identified during data collection that require reporting will follow institutional guidelines. Change Management: Any modifications to the protocol, data collection procedures, or analysis plan will be documented with version control and approved by the principal investigator. Sample Size Assessment The sample size for this study is determined by the availability of eligible patients diagnosed during the 10-year study period (July 2014 - June 2024). Based on institutional case volume, an estimated total of approximately 500-600 patients is expected. Statistical Analysis Plan All statistical analyses will be performed using R software (version 4.1.0 or higher) and SPSS (version 24.0). Descriptive Analysis: Baseline characteristics will be summarized using frequencies and percentages for categorical variables, and means (with standard deviations) or medians (with interquartile ranges) for continuous variables, as appropriate. Comparisons between training and validation cohorts will be performed using chi-square tests (or Fisher's exact test) for categorical variables and t-tests (or Mann-Whitney U tests) for continuous variables. Variable Screening and Model Development: Univariate Cox proportional hazards regression will be performed in the training cohort to screen potential predictors. Variables with P \< 0.10 will be considered for inclusion in the multivariate model to avoid excluding potentially important factors. Multivariate Cox regression with backward stepwise selection (based on Akaike Information Criterion, AIC) will be used to identify independent prognostic factors. Hazard ratios (HR) and 95% confidence intervals (CI) will be reported. A nomogram will be constructed based on the final multivariate Cox model, assigning points to each predictor proportional to its regression coefficient. Model Validation: Discrimination: Model performance will be assessed using the concordance index (C-index) and time-dependent receiver operating characteristic (ROC) curve analysis with area under the curve (AUC) at 3 and 5 years. Calibration: Calibration plots will be generated comparing predicted versus observed survival probabilities at 3 and 5 years, using 1000 bootstrap resamples to estimate optimism. Clinical Utility: Decision curve analysis (DCA) will be performed to evaluate the net clinical benefit of using the nomogram across a range of threshold probabilities for guiding escalated therapy decisions (e.g., recommending CDK4/6 inhibitors), compared to "treat-all" or "treat-none" strategies. Sensitivity Analyses: Subgroup analyses may be performed based on menopausal status or nodal status to explore model performance in clinically relevant subgroups. All statistical tests will be two-sided, and a P-value \< 0.05 will be considered statistically significant unless otherwise specified. Study Status The study is currently in the protocol stage and has not yet begun data collection. Institutional ethics approval has been obtained. Data extraction and analysis are expected to commence upon registration.