 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
ARTICLE |
Department of Pediatrics, University of California Davis, School of Medicine, Sacramento, California
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
METHODS. We prospectively studied children who had Graves’ disease and were treated with antithyroid medications. We compared children who achieved remission after 2 years with those who had persistent disease to determine which variables were associated with remission; multiple logistic regression and binary recursive partitioning analyses were used to evaluate interactions among predictive variables.
RESULTS. Of 51 children who completed the study, 15 (29%) achieved remission. Children who achieved remission had lower thyroid hormone concentrations at presentation than those with persistent disease (free thyroxine: 6.17 ± 3.10 vs 9.86 ± 7.54 ng/dL; total triiodothyronine: 431 ± 175 vs 561 ± 225 ng/dL). Children who achieved remission were also more likely to be euthyroid within 3 months of initiating propylthiouracil (82% vs 29%). Binary recursive partitioning analysis identified rapid achievement of euthyroid status after initiation of propylthiouracil, lower initial triiodothyronine, and older age as significant predictors of remission.
CONCLUSIONS. Thyroid hormone concentrations at diagnosis, age, and initial response to propylthiouracil can be used to stratify patients according to the likelihood of remission after 2 years of antithyroid medication use. These data provide a useful guide for clinical decision-making regarding Graves’ disease in children.
Key Words: hyperthyroidism • Graves’ disease • propylthiouracil
Abbreviations: T4— thyroxine • T3—triiodothyronine • SDS—SD score • TSI—thyroid-stimulating immunoglobulin • TBII—thyroid-binding inhibitory immunoglobulin • TPO—anti–thyroid peroxidase antibody • ANA—antinuclear antibody
The optimal approach to treatment of Graves’ disease in children has long been a subject of controversy. Available therapies, including antithyroid medications, radioactive iodine therapy, and thyroidectomy, all entail risks.1–7 Antithyroid medications are often favored as the initial treatment approach for children,8,9 in hopes of avoiding permanent hypothyroidism, which occurs frequently after either radioactive iodine therapy or surgery. Achieving remission with medical therapy, however, usually requires many years of treatment,10,11 and the risk for adverse reactions from antithyroid medications is relatively high (11%–22%).1,2,12
Many children who are initially treated with antithyroid medication eventually receive other treatments, most often radioablation.10,13–16 In some cases, alternative therapies are necessary because of adverse reactions to antithyroid medication, but many patients discontinue antithyroid mediations because of difficulties maintaining long-term compliance.10,13–15 Previous studies suggested that, of children who were treated with antithyroid medication, 
25% achieved remission with every 2 years of treatment.11,14 Probabilities of remission with long-term antithyroid therapy, however, are based on data from small numbers of patients and differ from adult data, which suggested that remission is unlikely with continued use of antithyroid medication beyond 18 months.17,18
Because only a minority of children who are treated with antithyroid medications eventually achieve remission with medical therapy alone, information regarding whether individual patients are likely to achieve remission within a relatively brief period using antithyroid medication would be useful in guiding therapy. We undertook this study to determine whether clinical characteristics or biochemical measures could be used to determine the likelihood of remission after 2 years on antithyroid medication.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Study Protocol
The study was approved by the institutional review board of each participating institution. After obtaining written informed consent from the patients’ parent or legal guardian, as well as assent for children who were older than 7 years, we recorded data regarding clinical features at presentation, including age, gender, ethnicity, family history of thyroid disease, height, weight, blood pressure, heart rate, goiter size, and presence or absence of exophthalmos (abnormal anterior protrusion of the eye determined by clinical judgment of the attending endocrinologist). We calculated the BMI and standardized the BMI for age and gender by calculating a BMI SD score (SDS): BMI SDS = (patient's BMI – mean BMI for age and gender)/SD for BMI. We assessed goiter size using 2 methods: (1) measurement of the longest length (diagonal) of right and left lobes and (2) physician's estimation of the percentage enlargement. Endocrinologists who participated in the study received standardized instructions on measurement of thyroid lobe length before the initiation of the study. Goiter size was classified according to the lobe length measurements (mean of right and left lobes) and estimated percentage enlargement as none, small, moderate, or large on the basis of standards described previously (Appendix).14
|
After the initial laboratory evaluation, children began treatment with propylthiouracil at a dosage of 5 to 7 mg/kg per day in 3 divided doses, up to a maximum of 450 mg/day. When the serum free T4 concentration declined to within the reference range, levothyroxine was added to the regimen according to the following age-based initial dosages: 1 to 5 years, 4 µg/kg per day; 6 to 12 years, 3 µg/kg per day; 
12 years, 2 µg/kg per day, up to a maximum dosage of 150 µg/day. The dosage of levothyroxine was adjusted to maintain the serum free T4 and thyrotropin concentrations within the reference range. The medical regimen described was chosen for the study protocol because this regimen was used by the majority of participating endocrinologists from the Organization of Pediatric Endocrinologists of Northern California at the time of initiation of the study. By using the most frequently used regimen, we hoped to parallel actual clinical practice as closely as possible.
Enrolled patients were seen monthly for the initial 3 months and quarterly thereafter during a 2-year period. At each visit, we measured goiter size as described previously and sent blood samples to the central laboratory for measurement of serum free T4, total T3, and thyrotropin concentrations. A complete blood count and measurement of serum liver enzymes were repeated at the 1- and 3-month visits with subsequent additional measurements made at the discretion of the treating physician.
Children who developed adverse reactions to propylthiouracil (rash, arthralgias/arthritis, elevated liver enzymes, neutropenia) were switched to methimazole and continued to be followed in the study, unless the treating physician believed that the adverse reaction was of sufficient severity to warrant avoidance of all antithyroid medications. The dosage and frequency of administration of methimazole for these patients was chosen at the discretion of the attending physician.
After 2 years of therapy, both propylthiouracil and levothyroxine were discontinued and serum free T4, total T3, and thyrotropin concentrations were measured at 2 weeks, 1 month, 3 months, 6 months, and 12 months after discontinuation of medications. Children who continued to be euthyroid 12 months after the discontinuation of medications were considered to have achieved remission.
Sample-size determinations for the study were based on the observed differences between groups and SD values for free T4 and total T3 concentrations from our previous retrospective study.14 We assumed an acceptable probability of type 1 error to be .05 and an acceptable probability of type 2 error to be .20. Children who achieved remission after 2 years of treatment with antithyroid medications in this study were compared with those who had persistent disease at 2 years. For univariate comparisons, we used Student's t test for continuous variables and the 
2 test for categorical variables. To determine which variables were independent predictors of early remission and to assess interactions among predictive variables, we also compared the 2 groups using 2 separate multivariable methods: multiple logistic regression and binary recursive partitioning. We used these 2 different multivariable methods to help verify the selection of predictive variables.
Binary recursive partitioning is a nonparametric analytic technique that is used to classify patients on the basis of the presence of predictive variables, using a tree-like structure with decision "nodes."19 Recursive partitioning analysis may be preferable to multiple logistic regression when the objective is derivation of a highly sensitive clinical prediction rule.20 Recursive partitioning allows for the inclusion of patients with missing predictors by substituting "surrogate" variables that contain information similar to that contained in the missing variables. Interactions between variables are also considered automatically during the process of recursive partitioning. These and other attributes of recursive partitioning may make this technique superior to logistic regression in accurately identifying patients with the outcome of interest.19 For these reasons, we considered the recursive partitioning analysis to be the primary multivariable analysis.
Variables that were chosen for inclusion in the multivariate analyses were those with significant univariate associations with remission status in this study, as well as those found in previous studies to be significant predictors of remission (age, gender, goiter size, BMI, thyroid hormone concentrations, and initial response to antithyroid medication). Total T3 was not included in the logistic regression analysis because of collinearity with free T4. Univariate comparisons and logistic regression analysis were performed by using Stata 8.0 (Stata Corp, College Station, TX).
We performed binary recursive partitioning using Answer Tree 3.0 (SPSS Inc, Chicago, IL). Candidate variables that were selected for the recursive partitioning analysis were identical to those entered into the logistic regression analysis with the exception that both total T3 and free T4 were considered because recursive partitioning analysis automatically chooses a candidate variable from among those that are collinear. Cutoff values for the conversion of continuous to dichotomous variables are also chosen automatically in the recursive partitioning program to separate optimally the data into risk profiles for the outcome of interest.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
Of the 51 children who completed the study, 15 (29%) achieved remission after 2 years of antithyroid medication (Table 2) . In univariate analyses, there were no significant differences in age, gender, or family history of thyroid disease between children who achieved remission within 2 years and those who did not. There were no significant differences in the frequency of exophthalmos, mean BMI, or BMI SDS between the 2 groups. Of note, however, when the population was divided into groups on the basis of ethnicity, patients who achieved remission in the white group (n = 25) tended to have higher BMI SDSs (remission group BMI SDS: –0.1 ± 0.8; persistent disease group BMI SDS: –0.7 ± 1.1; P = .17), whereas patients of other ethnicities (n = 26) showed an opposite trend (remission group BMI SDS: –0.4 ± 1.9; persistent disease group BMI SDS: –0.1 ± 1.1; P = .66).
|
In contrast to the clinical measures, thyroid hormone concentrations were significantly associated with remission status in univariate analyses. Both serum free T4 and serum total T3 concentrations were significantly lower at the time of presentation in the remission group. In addition, children in the remission group were much more likely to achieve euthyroid status within the first 3 months of treatment with propylthiouracil. Eighty-two percent of children in the remission group were euthyroid within the first 3 months of treatment, compared with only 29% of children in the nonremission group. The mean dosage of propylthiouracil in the remission group (6.0 ± 3.9 mg/kg) was not significantly different from that of the nonremission group (5.5 ± 2.1 mg/kg; P = .62).
At the time of presentation, 98% of patients had positive tests for TBIIs, but only 51% had positive tests for TSIs (Table 1). TPO antibodies were also detected frequently (87%). Thirty percent of children had positive tests for ANAs before treatment with propylthiouracil. There were no significant differences in the frequency of positive thyroid antibody tests between the remission group and the nonremission group (Table 2). The measured levels of thyroid antibodies likewise were not different between the 2 groups.
In the multivariable logistic regression analysis, only the initial response to propylthiouracil treatment maintained a significant association with remission after adjustment for the other variables (Table 3). There was a trend toward a significant association with remission for age, with older age associated with greater likelihood of remission (P = .06). In the binary recursive partitioning analysis, the initial response to propylthiouracil was also the most important predictive variable (Fig 1). Age and the initial total T3 concentration were also significantly associated with remission in this analysis, with older age and lower total T3 indicating a higher likelihood of remission. When stratified according to these variables, children who were euthyroid within 3 months of initiating treatment with propylthiouracil and who were older (>14.6 years of age) had the highest likelihood of achieving remission (83%). Those who were not euthyroid within 3 months of initiating treatment and who had high initial T3 concentrations (>383 ng/dL) had the lowest likelihood of achieving remission (5%).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Previous studies suggest that 25% of children achieve remission with every 2 years of treatment with antithyroid medications.10,11,14 In theory, therefore, most medically treated patients should eventually achieve remission, provided that medications are continued for sufficient periods. Probabilities of remission with long-term antithyroid medication use, however, are largely statistical predictions, based on data from relatively few patients, and may therefore be less accurate than short-term data.11 In addition, data from adult studies contrast with the more limited data from pediatric studies and suggest that prolonged treatment with antithyroid medications beyond 18 months is not associated with higher remission rates.17,18
Regardless of whether long-term treatment with antithyroid medications can improve remission rates, descriptive data indicate that many patients eventually receive alternative therapies. In previous retrospective studies, 25% to 66% of patients who initially were treated with antithyroid medications discontinued medications before achieving remission and were eventually treated with either radioactive iodine or, less commonly, surgery.10,11,13–16,23 Over time, progressively fewer patients continued medical therapy, with 14% to 18% per year opting for alternative treatment.14 Although some patients discontinue medications because of adverse reactions, these patients generally account for a minority of the total.10,13–16 Poor compliance and personal preference seem to be more frequent reasons for these changes in treatment modality.11
Previous retrospective studies evaluated factors that were associated with time to achieve remission in children with hyperthyroidism, but the results were variable and conflicting. Older age, lower initial thyroid hormone concentrations, a more rapid response to antithyroid medications, smaller goiter size at presentation, a decrease in goiter size during treatment, and higher initial BMI all have been associated with increased likelihood of remission in some studies but not in others.13,14,24–26 In a previous retrospective study by our group using multivariable analysis, smaller goiter size at presentation and higher initial BMI SDS were found to be independent predictors of early remission.14 Because this was a retrospective study, however, the methods for measuring thyroid hormone concentrations and thyroid antibody levels varied, and goiter size was not measured in a standardized manner. Therefore, we could not exclude the possibility that variations in these measures may have biased the results in favor or against finding an association with early remission.
Because this study was done prospectively, using standardized methods, the results differ in some respects from earlier reports. In agreement with some previous studies, older age, lower initial thyroid hormone concentrations, and a more rapid initial response to antithyroid medications were important predictors of early remission. In contrast to previous retrospective studies, however, initial goiter size was similar in the remission and nonremission groups. The reason for this difference is unclear. It is possible that clinicians tend to overestimate goiter size in children who are known to have very high thyroid hormone concentrations or more pronounced symptoms of hyperthyroidism, leading to bias when medical charts are reviewed retrospectively. In addition, in this study, we did not detect an association between BMI SDS and likelihood of remission. The reason for this difference is also unclear but may reflect underlying differences in the ethnic makeup of the population. Notably, when data were analyzed for white patients only, the differences in BMI SDSs between the remission group and the group with persistent disease were similar to those documented in our previous retrospective study.14 Ethnicity was not recorded in the previous retrospective study, but it is possible that the ethnic makeup of that population may have differed from that in this study, accounting for the observed variation in predictive value of BMI SDS.
The multivariate analyses in this prospective study likewise differ from previous analyses using retrospectively collected data.14 In this prospective study, the initial response to antithyroid medication, initial thyroid hormone concentrations, and age were the most important predictive variables. In contrast, in our previous retrospective study, BMI SDS and goiter size were the most important predictors in the multivariate analysis. This discrepancy likely reflects the use of standardized methods for measurement of goiter size and the use of a single laboratory for measurement of thyroid hormone concentrations in this study. These methods greatly limited sources of potential bias that are unavoidable in retrospective studies but might bias the results either in favor of or against finding associations between variables.
A notable finding in this study is the relatively low rate of detection of thyrotropin receptor autoantibodies by TSI bioassay, despite detection of thyrotropin receptor autoantibodies by TBII assay in almost all enrolled patients. Some previous studies demonstrated a high prevalence (>90%) of both TSIs and TBIIs in children with Graves’ disease,27–29 whereas others showed a lower rate of detection of TSIs compared with TBIIs.30 In our previous retrospective study of children with Graves’ disease, TSIs were detected in 76% of patients, but no data were collected regarding TBII measurements.14 The reason for the observed differences among studies is unclear. Thyrotropin receptor–blocking antibodies may be involved but are thought to be uncommon in patients with Graves’ disease and clinical hyperthyroidism.28 It is possible that the lower rate of detection of TSIs among patients in this study may reflect relatively milder disease, although TSI detection rates were similar in the remission group and persistent disease group. Lower rates of detection of TSIs may also indicate lower sensitivity of the TSI assay compared with the TBII method.
An additional notable finding in this study is the high frequency of detection of ANAs before treatment with antithyroid medications. Similar findings were reported in a previous study of adults with Graves’ disease,31 as well as a previous pediatric study.13 The current data reinforce these findings and suggest that detection of ANAs need not be taken as an indication of adverse reaction to antithyroid medication.
This study has limitations. The study population was relatively small compared with previous retrospective studies, and we were therefore limited in our ability to detect associations of smaller magnitude. In addition, the small sample size makes it difficult to estimate precisely the likelihood of early remission given various combinations of predictive variables. These variables could be evaluated with greater precision in a larger study. In addition, several patients were lost to follow-up or dropped out of the study before study completion. In this lost population, the initial free T4 concentrations tended to be higher than those of the study group. Because initial thyroid hormone concentrations are important predictors of early remission, it is possible that the study population overrepresented patients with a greater previous probability of achieving early remission. In addition, although we attempted to exclude patients with Hashitoxicosis from the study, there is substantial overlap in both clinical features and laboratory measures between patients with Hashitoxicosis and those with Graves’ disease,32 and it is possible that rare patients with Hashitoxicosis may have been included. Inclusion of these patients seems relatively unlikely, however, because none of the study subjects developed hypothyroidism and almost all patients had evidence of thyrotropin receptor antibodies.
An additional limitation of the study is that goiter size was assessed using manual measurements, rather than ultrasound, which may be more precise. Ultrasound measurements of goiter size were not included in the study protocol so that the variables assessed would be those that are routinely and readily available to clinicians. It is possible, however, that goiter size would have been a more important predictor of outcome, if measured more precisely by ultrasound. Finally, patients in the study were followed for only 1 year after discontinuation of antithyroid medications. It is possible that some patients experienced a relapse of the disease after the 1-year follow-up period; however, data from previous studies indicated that the frequency of such late relapses is very low.11
|
CONCLUSIONS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
We gratefully acknowledge the generous assistance of Quest Diagnostics in conducting the assays for thyroid hormone concentrations and thyroid antibody levels. We are also grateful to Dr Nathan Kuppermann for his helpful assistance with the statistical analyses.
|
FOOTNOTES |
|---|
Address correspondence to Nicole S. Glaser, MD, Department of Pediatrics, University of California, Davis, School of Medicine, 2516 Stockton Blvd, Sacramento, CA 95817. E-mail: nsglaser{at}ucdavis.edu
The authors have indicated they have no financial relationships relevant to this article to disclose.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Read all P3Rs
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2008 American Academy of Pediatrics. All rights reserved. |
||
|
||
