Predicting the Risk of Recurrence Before the Start of Antithyroid Drug Therapy in Patients With Graves' Hyperthyroidism

Treatment options of Graves' hyperthyroidism should be tailored to the needs of individual patients, taken into account recurrence risk, co-morbidities, and personal preferences. This study aimed to construct a prediction model to calculate recurrence risk after a course of antithyroid drugs, based on clinical and genetic parameters prior to the start of treatment. Consecutive, untreated patients with a first episode of Graves' hyperthyroidism were included in a prospective, multicenter, observational study. Before starting antithyroid drugs, clinical parameters and blood samples were collected. Antithyroid drugs were withdrawn after 1 year, and patients were followed for a further 2 years. Clinical and genetic markers that are independently associated with recurrent hyperthyroidism, and a multimarker prediction model for recurrence. Thirty-seven percent of 178 patients had recurrent Graves' hyperthyroidism within 2 years after antithyroid drug withdrawal. In Cox regression, lower age, higher serum free T4, higher serum thyrotropin-binding inhibitor immunoglobulin, larger goiter sizes at diagnosis, PTPN22 C/T polymorphism, and HLA subtypes DQB1*02, DQA1*05, and DRB1*03 were independent predictors for recurrence. Two simplified predictive models were constructed based on HRs of the multivariate model, called the Graves' Recurrent Events After Therapy (GREAT) score for clinical markers and the GREAT+ score for the combination of clinical and genetic markers. The GREAT and GREAT+ scores were divided into classes according to recurrence rates. Higher recurrence rates were observed in GREAT score class III (68%) compared with class II (44%) or class I (16%). The GREAT+ score showed much higher rates of recurrence in class IV+ (84%), compared with class III+ (49%), class II+ (21%), and class I+ (4%).The largest benefit of the GREAT+ score is in GREAT score class II: addition of genotypes changes recurrence risk (and thereby management) in 38%. Our prediction model based on simple clinical assessment, supplemented with genotyping at intermediate risk, can be of great value in individualized treatment of newly diagnosed patients with Graves' hyperthyroidism in routine clinical practice.