Clinical utility of TRAb reduction in predicting remission in pediatric Graves’ disease

Abstract.

Hyperthyroidism in childhood impairs quality of life, with Graves’ disease (GD) being the predominant cause. Although most cases can be managed with antithyroid drugs (ATD), recurrence remains challenging due to its frequency and unpredictability. Although thyrotropin receptor antibody (TRAb) has been proposed as a potential biomarker for predicting recurrence, the details, including the cut-off value, remain to be elucidated. We retrospectively analyzed 41 children (aged 2–15 yr) with GD and initiated ATD between 2005 and 2024 at two centers. Patients were classified into two groups, remission (n = 8) or non-remission (n = 33) based on the duration of euthyroidism for ≥ 24 mo after ATD discontinuation. TRAb levels were measured at the commencement of the therapy and every 6 mo afterward up to 24 mo. TRAb reduction rates (TRR) were significantly greater in the remission group at 6, 18, and 24 mo after treatment initiation. Among these, the 24-mo ROC analysis of TRR yielded the highest predictive value with an optimal cut-point of 93.2% (sensitivity 81.5%, specificity 85.7%). We concluded that the TRR is a potential marker to predict the remission of pediatric GD. Serial monitoring of TRAb would help to optimize the timing of ATD discontinuation.

Highlights

● TRAb reduction rate predicts remission in pediatric Graves’ disease.

● Early and sustained TRAb reduction is linked to long-term remission.

● A 93.2% TRAb reduction at 24 mo showed highest predictive value.

Introduction

Graves’ disease (GD) is the most common cause of hyperthyroidism (20–50 cases per 100,000 person-years) (1, 2), characterized by diffuse goiter and excessive thyroid hormone production. Although GD commonly develops between 30 and 50 yr of age, it can also present in childhood, adolescence, and young adulthood (1). Pediatric cases often exhibit distinct features, such as poor concentration, accelerated growth, and early puberty, which profoundly affect their social development and adaptation (3, 4). Disease recurrence is particularly problematic in pediatric patients. However, treatment resistance to antithyroid drugs (ATD), which is a primary option for the treatment of GD, is common (5). The recurrence rate of the disease among children is higher, and merely 28.8%–38.7% of the patients achieved remission for more than one year after cessation of ATD (6, 7). Furthermore, curable but more invasive therapies, such as surgical thyroidectomy and radioactive iodine treatment, are generally avoided in children and adolescents due to potential long-term adverse effects (5, 8). Therefore, identifying pediatric patients at high risk of relapse remains a critical problem.

The pathophysiology of GD is endocrinological autoimmunity, with binding of thyrotropin receptor antibody (TRAb) to the thyrotropin receptor being the primary cause (1). Accordingly, the TRAb levels are elevated at the onset of the disease and decrease over time with ATD therapy (8, 9). A marked reduction in TRAb titers is often used as an indicator of treatment efficacy, and negative seroconversion has been one of the criteria for discontinuing ATD (10). However, TRAb seroconversion does not generally warrant long-term remission of the disease, as relapse occurs in approximately 20%–34% of the patients who discontinued the ATD therapy even after TRAb becomes negative (8, 11). Although the usefulness of TRAb as a predictor of long-term remission has been considered limited, previous studies mainly focused on absolute values of TRAb rather than other indexes (8, 9), such as the extent of TRAb reduction with treatment. However, most studies on TRAb dynamics have focused on adults, and pediatric data remain limited (8, 12). Given the higher relapse rates, restricted therapeutic options, and significant developmental consequences of uncontrolled disease in children, identifying reliable predictors of remission is particularly important. Therefore, in this study, we aimed to determine whether the extent to which TRAb values decrease over a certain period after commencement of the ATD therapy would be useful for predicting long-term remission.

Methods

Study design and subjects

This retrospective cohort study included patients diagnosed with GD before 15 yr of age who were treated at Toyama University Hospital or the National Center for Child Health and Development. Patients were followed from January 2005 to December 2024. GD was defined as a set of thyrotoxic conditions due to hyperthyroidism in the presence of TRAb or thyroid-stimulating antibodies (TSAb). All patients initially received ATD therapy. Methimazole or propylthiouracil was selected at the discretion of the attending physician. The initial dose and tapering schedule were not standardized and were determined based on clinical judgment. Patients with neonatal GD resulting from transplacentally acquired antibodies were excluded. Patients were also excluded if they discontinued ATD therapy within 6 mo, had less than 36 mo of follow-up since starting ATD therapy, or had incomplete clinical data (Fig. 1).

Fig. 1.

Flow diagram showing the recruitment process for participants and reasons for exclusion in this study.

Data collection and thyroid function assessment

Data on patient demographics, clinical history, laboratory results, and therapeutic interventions were extracted from electronic medical records. Among the various thyroid-related autoantibodies, we selected TRAb as a biomarker because it is most frequently used in clinical practice and has been measured longitudinally at our institutions. TRAb values were primarily measured using an electrochemiluminescence immunoassay with the ECLusys TRAb kit, a third-generation assay with a cut-off value of < 2.0 IU/L (Roche Diagnostics, Germany). In a subset of patients, TRAb was measured with the DYNOtest TRAb Human kit, a second-generation assay with a cut-off value of < 1.0 IU/L (Yamasa, Tokyo, Japan). The values obtained via the above two measurement methods are strongly correlated (correlation coefficient: 0.979) (13), so they were conveniently treated as equivalent in this study. TSH, free T3, and free T4 were assessed using electrochemiluminescence immunoassays (ECLusys TSH, FT3 III, FT4 IV, Roche Diagnostics, Germany), and the reference ranges, based on the test kits employed, were 0.61–4.23 mIU/L (post-harmonization), 2.30–4.50 pg/mL, and 0.90–1.70 ng/dL, respectively. Because our dataset included both pre- and post-harmonization TSH values, direct comparison of serial TSH levels was not appropriate. Hyperthyroidism was defined as a TSH below the lower limit of detection with elevated free T3 and free T4 levels.

Definition of remission/non-remission of Graves’ disease

We defined the remission group as those who discontinued ATD within 36 mo of initiation and maintained normal thyroid function for at least 24 mo thereafter.

The non-remission group included those who were unable to discontinue ATD within 36 mo of initiation, transitioned to non-oral treatments, including surgery or radiation therapy, or experienced hyperthyroidism within 24 mo after ATD discontinuation (Fig. 1).

Statistical analysis

The primary outcome of this study was remission status in patients with GD. Continuous variables were expressed as medians with interquartile ranges and compared using the Mann–Whitney U test. Categorical variables were compared using the chi-square test. Values below the lower limit of detection (LOD) were imputed as LOD/2, in accordance with standard practice. TRAb levels and their reduction rates (TRR) were evaluated at 6, 12, 18, and 24 mo after ATD initiation. TRAb values obtained after ATD discontinuation were excluded from the analysis. The TRR was calculated as follows: [(TRAb at pretreatment − TRAb at timepoint) / TRAb at pretreatment] × 100 (%). Receiver operating characteristic (ROC) curves and area under the curve (AUC) values with their 95% confidence intervals (CI) were used to assess the predictive performance of TRAb metrics for remission. The optimal cutoff value was determined using the Youden Index. To further examine whether combining TRR and absolute TRAb values could improve predictive performance, we constructed a simple risk score based on ROC-derived cutoffs. TRR and TRAb were each dichotomized according to their respective optimal cutoff values, and the resulting binary variables were summed to generate a combined score. The discriminatory ability of this score was evaluated by ROC analysis and compared with that of TRR or TRAb alone using the DeLong test.

All statistical analyses were performed using R version 4.3.1, and ROC curves were generated using GraphPad Prism version 10.3.1 (GraphPad Software, La Jolla, CA, USA). A p-value of < 0.05 was considered statistically significant.

Ethics statement

This study was approved by the Ethics Committees of all participating institutions and the Ethics Committee of the University of Toyama (No. R2024158). All procedures complied with the ethical standards of the relevant national and institutional committees on research involving human participants and with the principles outlined in the Helsinki Declaration of 1975, as revised in 2008, and other national regulations and guidelines. Given the retrospective nature of the study and the use of de-identified data, the requirement for written informed consent was waived by the IRB. Instead, an opt-out method was employed, allowing participants and/or their legal guardians to decline participation. Information about the study—including its objectives, methods, and the right to withdraw from the study without any consequences—was publicly available on the Toyama University Hospital and the National Center for Child Health and Development website. Patients and/or their guardians were given sufficient opportunity to opt out if they did not wish for their medical data to be included in the study.

Results

Patient selection and pretreatment clinical characteristics

Of the 101 patients diagnosed with GD during the study period, 60 were excluded from the final analysis because of insufficient disease duration (< 36 mo at the time of analysis, n = 12), transfer to other hospitals (n = 8), or missing TRAb data at the required time points (n = 40) (Fig. 1). A total of 41 patients were included in the final analysis. Among them, 36 (87.8%) were female, and the median age at onset was 11 yr and 10 mo. Remission was achieved in 8 patients (19.5%), whereas 33 patients (80.5%) did not achieve remission. Of the 33 patients in the non-remission group, 26 (78.8%) did not meet the criteria for discontinuing ATD therapy even after 36 mo of treatment, and 7 (21.2%) experienced relapses within 24 mo after ATD discontinuation. Four patients with trisomy 21 were included in the final analysis, and all of them were classified into the non-remission group. No significant differences were observed between the remission and non-remission groups in terms of sex distribution, family history of thyroid disease, age at diagnosis, initial serum free T3 and free T4 levels, free T3/free T4 ratio, thyroid gland volume at diagnosis, or baseline TRAb levels (Table 1).

Table 1. Characteristics of the study population.

In both groups, all patients initially received thiamazole as the first-line ATD. The median (P25–P75) initial thiamazole dose (mg) was 20 (15–20) in the remission group and 20 (15–20) in the non-remission group, with no significant difference between the groups. In the non-remission group, one patient was switched from thiamazole to propylthiouracil due to myositis, whereas all other patients continued treatment with thiamazole throughout the course of ATD therapy. The proportions of patients who remained on ATD therapy were 100% in both groups at 6 and 12 mo; 87.5% in the remission group and 93.3% in the non-remission group at 18 mo; and 62.5% in the remission group and 96.7% in the non-remission group at 24 mo. During the observation period, one patient in the non-remission group underwent surgical treatment.

TRAb dynamics and predictive performance for remission

The absolute median TRAb values did not differ significantly between the two groups at any time point. In contrast, the TRR values were significantly higher in the remission group at 6, 18, and 24 mo after ATD initiation (Table 2). ROC curve analyses for predicting remission showed that the AUC for absolute TRAb values reached its maximum at 24 mo (AUC = 0.78). In contrast, for TRR, the highest AUC was also observed at 24 mo (AUC = 0.87) (Table 3). At 24 mo, the optimal TRR cut-off value was 93.2%, yielding a sensitivity of 81.5% and specificity of 85.7% (Fig. 2). This result indicates that a greater reduction in TRAb levels over 24 mo strongly predicts remission in pediatric GD.

Table 2. . Comparison of TRAb values and % reduction.

Table 3. Comparison of AUCs for predicting remission of Graves’ disease.

Fig. 2.

ROC curve for the TRAb reduction rate at 24 mo after initiation of therapy. The TRAb reduction rate: [(TRAb at pretreatment − TRAb at time point) / TRAb at pretreatment] × 100 (%).

To explore the utility of a combined metric, we constructed a simple risk score using the cutoff values corresponding to the highest AUC for each index (TRR at 24 mo: 93.2%; absolute TRAb at 24 mo: 1.9 IU/L). The combined score yielded an AUC of 0.81 (sensitivity 81.5%, specificity 85.7%). However, based on the DeLong test, its discriminatory performance was not statistically superior to that of TRR or absolute TRAb value.

Discussion

To the best of our knowledge, this is the first study to demonstrate the TRAb reduction rate as a potential predictor of remission in pediatric GD.

In this study, no pretreatment variables showed significant differences between the remission and non-remission groups, including pretreatment TRAb levels, suggesting that predicting remission prior to therapy remains extremely challenging. Previous reports on pediatric GD have proposed several potential prognostic indicators, such as lower thyrotropin-binding inhibitory immunoglobulin or free T3 levels at onset, older age at diagnosis, and the absence of marked goiter (14,15,16). However, these markers are inconsistent across studies, and no reliable pretreatment biomarker has been established. Accordingly, unless novel prognostic factors for GD are identified, therapeutic outcomes cannot be predicted before treatment.

In contrast to pretreatment findings, the TRR was significantly higher in the remission group at 6, 18, and 24 mo during ATD therapy. Furthermore, the AUC of the TRR reached its maximum value of 0.87 in 24 mo, which was sufficient to distinguish GD remission from non-remission. In adult GD, the TRR has been suggested as a biomarker for predicting remission in a few studies (17, 18). Our study extends this finding to children, in whom the risk of relapse is high, and treatment options are limited. Given that a minimum treatment duration of 1–2 yr of ATD therapy is recommended for pediatric patients (8), the predictive utility of the 24-mo TRR is clinically meaningful for optimizing the timing of ATD discontinuation and avoiding unnecessarily prolonged treatment.

On the other hand, patients with a TRR < 93.2% at 24 mo may be less likely to achieve remission within three years or successfully discontinue ATD therapy. This suggests that extended ATD treatment could be considered in such cases, with attention to adherence and alternative options. Supporting this approach, one study reported that long-term ATD therapy in pediatric GD increased cumulative remission rates over a five-year treatment period without a notable rise in adverse events (19). Another study demonstrated that prolonged administration of carbimazole, a precursor of methimazole, increased cumulative remission rates to over 40% when continued for 8–10 yr (20). Given the low remission rate of 19% in our cohort, long-term ATD therapy may represent a practical option for patients with low TRR values.

A notable limitation of TRR as a predictive marker is that patients with low baseline TRAb values may display low TRR values despite substantial improvement. In our cohort, two patients in the remission group had baseline TRAb values < 10; therefore, their TRR values were relatively low. These cases indicate that relying solely on TRR may underestimate the potential for remission, and that concurrent evaluation of absolute TRAb values is advisable. However, five patients in the non-remission group also had baseline TRAb < 10 and achieved TRAb-negativity at 24 mo, yet did not attain remission. Furthermore, the combined risk analysis incorporating both TRR and absolute TRAb values did not demonstrate superior discriminatory performance compared with TRR alone. Establishing reliable predictive methods for patients with low initial TRAb values remains an important challenge for future research.

Although elucidating the underlying mechanisms is beyond the scope of this study, one possible explanation is the involvement of unidentified autoantibodies, as reported in other autoimmune disorders (21). Such autoantibodies may lead to persistently low TRAb levels from pretreatment through ATD therapy. Future studies that account for relevant background factors will be necessary to clarify these immunological features.

This study has some limitations. First, the small cohort size precluded multivariable logistic regression analysis. Second, TRAb values from different assay generations were analyzed, and treating their absolute values as equivalent is inherently limited. Variability between assays may have influenced the results. A further limitation is the heterogeneity in treatment regimens inherent to the retrospective design, as both ATD dosage and duration were determined by individual physicians, and medication adherence was not systematically assessed, potentially influencing treatment outcomes. Moreover, relapse beyond two years after ATD withdrawal has been reported, with about 15% occurring after this period (22); thus, the durability of long-term remission remains an important question. To validate the usefulness of the TRR, a further prospective and multicenter cohort study with long-term follow-up, standardized protocols, and systematic assessment of adherence is needed.

In conclusion, the TRR appears to be a valuable predictor of remission in pediatric GD. Serial monitoring TRAb reduction over time may facilitate individualized decisions regarding the optimal timing of ATD withdrawal and overall treatment duration.

Conflict of interests

CI reports patent royalties from Juno Therapeutics, research funds, and advisory fees from CURED Inc. The other authors declare no conflicts of interest associated with this manuscript.