Dynamic vitamin D trajectories and their prognostic value in breast cancer: a group-based trajectory modeling study

Breast cancer represents the most frequently diagnosed malignancy and a leading cause of cancer-related mortality among women globally, with substantial prognostic heterogeneity that underscores the urgent need for robust prognostic biomarkers to guide personalized clinical management. Although accumulating evidence has linked vitamin D status to breast cancer progression, most existing studies have relied on static single measurements of serum vitamin D, leading to inconsistent findings, while the prognostic implications of dynamic longitudinal changes in vitamin D levels during treatment and follow-up remain largely underexplored. In this retrospective cohort study, we enrolled 513 female patients with pathologically confirmed primary invasive breast cancer treated between 2020 and 2023, with a median follow-up duration of 38.0 months. Serial serum 25-hydroxyvitamin D measurements were obtained at three predefined time points: baseline, mid-treatment, and post-treatment follow-up. Group-based trajectory modeling (GBTM) was employed to identify distinct latent subgroups with homogeneous vitamin D change patterns. We then evaluated the association between these trajectories and event-free survival (EFS) using Kaplan-Meier analysis, log-rank test, and multivariable Cox regression, and further constructed a trajectory-based risk stratification model and a visual prognostic nomogram, whose performance was validated via concordance index (C-index), calibration curves, and decision curve analysis. We identified 6 distinct vitamin D trajectories, which were further categorized into three risk strata, with 64.9% of patients presenting with severe vitamin D deficiency at baseline. After multivariable adjustment for established clinical factors, dynamic vitamin D trajectories were confirmed as an independent prognostic factor for EFS: patients in the high-risk group (characterized by persistent deficiency or worsening vitamin D status) exhibited significantly worse survival outcomes compared with the low-risk group with consistently sufficient vitamin D levels (all pairwise p < 0.001). Landmark analysis further confirmed the robustness of these findings by ruling out potential survivorship bias. The integrated prognostic nomogram, which incorporated age, tumor stage, molecular subtype, and trajectory risk group, achieved a C-index greater than 0.7, with favorable calibration and substantial clinical net benefit, demonstrating superior discriminative ability compared with traditional prognostic models. In conclusion, dynamic serum vitamin D trajectories serve as a strong, independent prognostic indicator for breast cancer patients. Our findings highlight the clinical significance of longitudinal vitamin D monitoring, and the proposed risk stratification tool provides a practical approach to facilitate individualized prognostic assessment and optimized clinical management for breast cancer patients.