Endocrine Abstracts

Objectives: Graves’ disease (GD) is characterized by variable disease severity at diagnosis and variable disease activity during follow-up. Antithyroid drugs (ATD) such as carbimazole (CMZ), methimazole or propylthiouracil are the first-line treatment. Current guidelines recommend a range of ATD starting dose according to GD severity, followed by dose titration until euthyroidism is reached and a maintenance dose for a specific patient is established. As thyroid hormones have a strong positive chronotropic effect, heart rate (HR) turns out to be a useful clinical marker to monitor thyroid activity under pharmacotherapy. The aim of this study was to extend recently developed clinically practical pharmacometrics (PMX) computer model characterizing individual disease activity and describing the relation between free thyroxine (FT4) and tachycardia in pediatric patients with GD with various GD severity under CMZ monotherapy to adults with or without receiving beta blockers.

Methods: Retrospectively collected clinical (resting HR during consultation) and laboratory data from adults with GD during the first 120 days of treatment at University Hospital Basel, Switzerland, were analyzed. PMX computer model was developed in the non-linear mixed effects modeling framework consisting of differential equations linking FT4 kinetics with HR dynamics, accounting for inter-individual variability, and incorporating clinically relevant individual patient characteristics. GD severity groups were defined based on FT4 at diagnosis according to current guidelines.

Results: Data from 37 adults with GD with at least 2 HR measurements (68% female, median age 42 [IQR 35, 55] years, and 25 mild, 8 moderate, and 4 severe GD at diagnosis, 54% receiving beta blockers) with 203 FT4 measurements and 129 HR measurements, 122 paired measurements, were analyzed and used for model development. At diagnosis, patients showed a median FT4 of 38.3 [IQR 32.3, 54.1] pmol/l, and a median HR of 94 [IQR 81, 118] bpm. Final PMX model accurately predicted individual disease dynamics for each GD patient during the first 120 days of treatment accounting for clinically relevant covariate effects, such as age, gender, and GD severity.

Conclusion: Developed clinically practical PMX computer model predicts individual HR and disease dynamics under CMZ monotherapy with or without receiving beta blockers. This computer model is expected to facilitate personalized pharmacotherapy and has the potential to mitigate risk for under- or overdosing of ATD in patients with GD. Prospective randomized validation trials are warranted to further validate and fine-tune computer-supported personalized dosing in GD patients.