Background: Auto-antibody testing is recommended for Graves' Disease (GD). The thyroid-stimulating-immunoglobulin (TSI) bridge method is designed to provide specificity for stimulatory antibodies to the TSH receptor which may translate to differences in performance in diagnosis of GD compared to TSH-receptor-antibody (TRAb) assays that don't distinguish between TSH-receptor antibody subclasses. The objective of this study was to prospectively compare the performance of a TSI assay to a TRAb assay for diagnosis of GD.
Methods: A total of 158 non-pregnant patients with new-onset hyperthyroidism were recruited into the study by endocrinologists. Final diagnosis of GD or non-GD was made by endocrinologists after clinical assessment and diagnostic work-up (TSH, free T4, TRAb and imaging). TSI results were blinded. Sensitivity/specificity of TRAb and TSI were determined; receiver operating characteristic (ROC) curve analysis was used to compare overall accuracy and optimal diagnostic thresholds for each assay.
Results: Complete assessment was available for 131/158 patients (95 GD and 36 non-GD). A strong correlation between TRAb and TSI existed (r = 0.92 (0.89-0.94), p < 0.0001). There was no significant difference between tests, area under the curve (AUC) (0.935 vs 0.929, p = 0.86) using manufacturer's recommended thresholds. Both assays had sensitivity around 88 % and specificity 80-90 %. The ROC-determined threshold for 95 % GD specificity was 3.63 IU/L (TRAb) and 0.98 IU/L (TSI) with corresponding sensitivities 77.5 % and 86.2 %. Assuming nuclear tracer thyroid scanning for results below 95 % specificity, 14 (14.7 %) GD patients would require a scan after TRAb, and 7 (7.4 %) after TSI.
Conclusions: TRAb and TSI performed similarly in the work up of hyperthyroidism; an optimized specificity threshold might permit TSI testing to reduce nuclear tracer thyroid scan requirements, without losing diagnostic sensitivity.
Keywords: Anti-thyroid antibody; Graves’ disease; Hyperthyroidism; Thyroid autoimmunity.
Copyright © 2025. Published by Elsevier Inc.