Baseline TSH within reference range and incident thyroid dysfunction in cancer patients: a retrospective cohort study

Whether baseline thyrotropin (TSH) within the reference range predicts thyroid dysfunction in cancer patients remains unclear. We evaluated this association in hospitalized cancer patients with normal baseline thyroid function.

This retrospective cohort study (October 2020-September 2025) included 532 cancer patients with normal baseline thyroid function and ≥ 1 follow-up test. Baseline TSH was modeled using restricted cubic splines and categorized as low (< 1.02 mIU/L), optimal (1.02-2.00 mIU/L), or high (> 2.00 mIU/L). Primary outcome was overall thyroid dysfunction; secondary outcomes included clinical dysfunction and mortality. Multivariable Cox models and 365-day restricted mean survival time (RMST) analyses were performed.

Overall dysfunction occurred in 175 patients (32.9%). Baseline TSH showed a U-shaped association with overall thyroid dysfunction (P for nonlinearity < 0.001), with the lowest risk at 1.02-2.00 mIU/L. Compared with optimal TSH, low TSH (HR 1.90, 95% CI 1.19-3.03) and high TSH (HR 2.22, 95% CI 1.55-3.18) independently predicted increased risk. RMST analysis confirmed 52.5 and 64.6 fewer dysfunction-free days for low and high groups, respectively (both P < 0.001). High TSH predicted clinical dysfunction (HR 1.71, 95% CI 1.03-2.81); low TSH did not. No association was found with mortality.

In hospitalized cancer patients, baseline TSH within the reference range showed a U-shaped association with incident thyroid dysfunction, with lowest risk at 1.02-2.00 mIU/L. Risk stratification within the conventional reference range may inform personalized monitoring strategies during anticancer treatment.