# A prognostic nomogram model for non-complete remission following initial radioiodine therapy in Graves’ hyperthyroidism

## TL;DR

This study creates a predictive model to identify patients likely to have incomplete remission after initial radioiodine therapy for Graves’ hyperthyroidism.

## Contribution

A validated nomogram model is developed for personalized prediction of non-complete remission after radioiodine therapy in Graves’ disease.

## Key findings

- Thyroid mass, RAI uptake, effective half-life, and FT3 reduction are independent predictors of non-complete remission.
- The nomogram model achieved high AUC values (0.919 in training, 0.901 in validation) for predicting remission outcomes.
- Calibration and decision curve analysis confirmed the model's reliability and clinical utility.

## Abstract

Radioiodine (RAI) therapy, while established for Graves’ hyperthyroidism (GH), exhibits variable efficacy (50-80% cure rates), with non-complete remission (NCR) necessitating retreatment. In the study, we aimed to identify independent predictors of NCR and develop a validated nomogram for personalized RAI outcome prediction.

Data from 285 GH patients undergoing initial RAI therapy were retrospectively analyzed and randomly allocated into training (n=199) and validation (n=86) cohorts at a 7:3 ratio. Univariate followed by multivariate logistic regression identified independent predictors of NCR in the training cohort. These variables informed the construction of a prognostic nomogram model, subsequently verified in the validation cohort through calibration, receiver operating characteristic (ROC) curve, and decision curve analysis (DCA) to assess model reliability, discriminative ability, and clinical utility.

Thyroid mass (TM), 24-hour RAI uptake (RAIU24h), effective half-life (Teff), and free triiodothyronine reduction at 1-month post-therapy (ΔFT3) were independent predictors. The prognostic nomogram integrating these variables exhibited superior discriminative performance in both training (AUC = 0.919) and validation cohorts (AUC = 0.901). Calibration curves confirmed high fidelity between predicted and observed NCR probabilities. DCA demonstrated significant clinical net benefit across threshold probabilities.

TM, RAIU24h, Teff, and ΔFT3 are critical determinants of RAI efficacy in GH. The validated nomogram enables precise NCR risk stratification, facilitating optimized activity prescription to improve remission rates.