Serum thyroid hormone concentrations and breast cancer risk: an extended cohort study with time-varying exposure analysis

Purpose Single-baseline categorization of biochemical exposures that fluctuate within individuals can misclassify cumulative exposure, yet most cohort studies of thyroid function and breast cancer rely on a single measurement and self-reported outcomes. We examined serum thyroid-stimulating hormone (TSH) and free thyroxine (FT4) in relation to breast cancer risk in a large Korean cohort, comparing static baseline with pre-specified time-varying exposure modeling, and tested whether previously reported associations from this cohort were reproduced. Methods In 162,702 women aged ≥ 20 years in the Kangbuk Samsung Health Study (2011–2024), incident breast cancer (ICD-10 C50 or D05) was ascertained by Korea Central Cancer Registry linkage. We fit progressively adjusted static Cox models and time-varying Cox models that updated TSH and FT4 at repeated visits, with pre-specified subgroup (menopausal status, body mass index) and sensitivity analyses. Results Over a mean 6.94 years, 1,841 incident breast cancers occurred. For euthyroid-range TSH, static and time-varying models gave concordant inverse estimates (highest vs lowest tertile: fully adjusted hazard ratio [HR] 0.93, 95% confidence interval [CI] 0.81–1.07; time-varying 0.90, 0.80–1.02). The inverse trend was stronger in postmenopausal women (HR 0.69, 0.47–1.02) and for invasive-only disease (HR 0.88, 0.76–1.01). A previously reported positive elevated-FT4 association was not reproduced and was unstable across static and time-varying models (0.62 vs 1.05; 12 events). Conclusion Euthyroid TSH showed a consistent but attenuated inverse association with breast cancer risk, whereas a prior elevated-FT4 signal was not replicated. Repeated-measurement designs matter most for episodic biochemical exposures; subtype-stratified replication is needed to test the postmenopausal TSH signal.