Risk of recurrence at the time of withdrawal of short- or long-term methimazole therapy in patients with Graves' hyperthyroidism: a randomized trial and a risk-scoring model.

Purpose: In Graves' disease, administration of low-dose methimazole for more than 60 months induces higher remission rates compared with the conventional duration of 12-18 months. However, the risk of recurrence and its predictors beyond 48 months of drug withdrawal are not known. The aims of this study were to determine the risk of recurrence during 84 months after withdrawal of short- or long-term methimazole therapy and a risk stratification for recurrence of hyperthyroidism.

Methods: A total of 258 patients were treated with methimazole for a median of 18 months and randomized to discontinuation of the drug(conventional short-term group; n = 128) or continuation of the treatment up to 60-120 months(long-term group; n = 130). Patients were followed for 84 months after methimazole withdrawal. Cox proportional hazards modeling was performed to identify factors associated with relapse and develop a risk-scoring model at the time of discontinuing the treatment.

Results: Hyperthyroidism recurred in 67 of 120(56%) of conventionally-treated patients versus 20 of 118(17%) of those who received long-term methimazole treatment, p < 0.001. Age, sex, goiter grade, triiodothyronine, thyrotropin, and thyrotropin receptor antibodies were significant predictors of recurrence in both "conventional" and "long-term" groups but free thyroxine just in the "long-term" group. The risk-scoring model had a good discrimination power (optimism corrected c-index = 0.78,95%CI = 0.73-0.82) with a range of 0-14 and sensitivity of 86% and specificity of 62% at the risk-score of eight.

Conclusion: A relapse-free state was achieved in 83% of patients with Graves' hyperthyroidism 84 months after cessation of long-term methimazole treatment which could be predicted by some significant predictors in a simple risk-scoring system.