Screening optimal candidates with operable, early-stage triple-negative breast cancer benefitting from capecitabine maintenance: A post-hoc analysis of the SYSUCC-001 study

To explore whether specific clinicopathological covariates are predictive for a benefit from capecitabine maintenance in early-stage triple-negative breast cancer (TNBC) in the SYSUCC-001 phase III clinical trial. Candidate covariates included age, menstrual status, type of surgery, postoperative chemotherapy regimen, Ki-67 percentage, histologic grade, primary tumor size, lymphovascular invasion, node status, and capecitabine medication. Their nonlinear effects were modeled by restricted cubic spline. The primary endpoint was disease-free survival (DFS). A survival prediction model was constructed using Cox proportional hazards regression analysis. All 434 participants (306 in development cohort and 128 in validation cohort) were analyzed. The estimated 5-year DFS in development and validation cohorts were 77.8 % (95 % CI, 72.9%–82.7 %) and 78.2 % (95 % CI, 70.9%–85.5 %), respectively. Age and node status had significant nonlinear effects on DFS. The prediction model constructed using four covariates (node status, lymphovascular invasion, capecitabine maintenance, and age) demonstrated satisfactory calibration and fair discrimination ability, with C-index of 0.722 (95 % CI, 0.662–0.781) and 0.764 (95 % CI, 0.668–0.859) in development and validation cohorts, respectively. Moreover, patient classification was conducted according to their risk scores calculated using our model, in which, notable survival benefits were reported in low-risk subpopulations. An easy-to-use online calculator for predicting benefit of capecitabine maintenance was also designed. The evidence-based prediction model can be readily assessed at baseline, which might help decision making in clinical practice and optimize patient stratification, especially for those with low-risk, capecitabine maintenance might be a potential strategy in the early-disease setting. •Individualized prediction model to quantify the clinical benefit of capecitabine maintenance in TNBC is still lacking.•We reanalyzed data of SYSUCC-001 trial and developed the SYSU-001 prediction model integrating four routine clinic variables.•This study underscored the nonlinear effects of different factors in routine oncology practice (age and node status).•Our SYSU-001 prediction model may be a potential tool to quantify personal benefit from adding capecitabine therapy in early-stage TNBC.