Abstract
OBJECTIVE: To address the overestimation of levothyroxine (L-T4) doses in conventional weight-based regimens for individuals who are overweight and obese, this study aimed to identify the most predictive body weight metrics and establish an optimized dosing model for accurate thyroid-stimulating hormone (TSH) suppression following total thyroidectomy in differentiated thyroid carcinoma (DTC). METHODS: This retrospective study included 385 patients with DTC treated at our institution between October 2019 and December 2024. Patients were stratified by TSH targets (A1: <0.1 mIU/L; A2: 0.1-0.5 mIU/L; A3: 0.5-2.0 mIU/L) and body mass index (BMI) according to Chinese criteria (normal: <24 kg/m²; overweight: 24-27.9 kg/m²; obesity: ≥28 kg/m²). Linear regression analysis was used to analyze correlations between the final stable L-T4 dose and weight metrics, including total body weight, adjusted body weight, lean body weight, ideal body weight, and body surface area, followed by model validation. Model performance was internally validated using a hold-out method. Efficacy was estimated as the accuracy of the model-predicted dose compared with the actual dose required when a patient first achieved their TSH target within the first postoperative year. RESULTS: The baseline characteristics showed no significant intergroup differences (P>0.05). Postoperative TSH levels varied significantly according to BMI (P<0.05). Patients with higher BMI required higher total L-T4 doses (µg/d) (P<0.001) but lower weight-adjusted doses (µg/kg/d) (P<0.001). Adjusted body weight best predicted L-T4 dose for patients with BMI ≤ 23.9 kg/m(2), while lean body weight was optimal for those with BMI≥24.0 kg/m(2). The new model achieved a significantly higher rate of accurate initial dose prediction compared with that via empirical dosing (68.0% vs. 30.2%, P<0.001). CONCLUSION: The BMI-stratified L-T4 dosing formula based on optimized body weight metrics demonstrated improved accuracy, expediting TSH suppression and reducing adverse events.