Unveiling Risk Factors for Treatment Failure in Patients with Graves’ Disease: A Nationwide Cohort Study in Korea
Article information
Abstract
Background
Antithyroid drug (ATD) treatment is the preferred initial treatment for Graves’ disease (GD) in South Korea, despite higher treatment failure rates than radioactive iodine (RAI) therapy or thyroidectomy. This study aimed to evaluate the incidence of treatment failure associated with the primary modalities for GD treatment in real-world practice.
Methods
We included 452,001 patients diagnosed with GD between 2004 and 2020 from the Korean National Health Insurance Service-National Health Information Database. Treatment failure was defined as switching from ATD, RAI, or thyroidectomy treatments, and for ATD specifically, inability to discontinue medication for over 2 years.
Results
Mean age was 46.2 years, with females constituting 70.8%. Initial treatments for GD included ATDs (98.0%), thyroidectomy (1.3%), and RAI (0.7%), with a noted increment in ATD application from 96.2% in 2004 to 98.8% in 2020. During a median follow-up of 8.5 years, the treatment failure rates were 58.5% for ATDs, 21.3% for RAI, and 2.1% for thyroidectomy. Multivariate analysis indicated that the hazard ratio for treatment failure with ATD was 2.81 times higher than RAI. RAI treatments ≥10 mCi had 37% lower failure rates than doses <10 mCi.
Conclusion
ATDs are the most commonly used for GD in South Korea, followed by thyroidectomy and RAI. Although the risk of treatment failure for ATD is higher than that of RAI therapy, initial RAI treatment in South Korea is relatively limited compared to that in Western countries. Further studies are required to evaluate the cause of low initial RAI treatment rates in South Korea.
INTRODUCTION
Graves’ disease (GD) is the most common cause of thyrotoxicosis, with a lifetime prevalence of 2% in women and 0.5% in men [1]. GD, an autoimmune disease, is related with thyrotropin receptor stimulating autoantibodies (TRAb), which causes increased thyroid hormone synthesis and secretion [1,2]. Currently, antithyroid drugs (ATDs), radioactive iodine (RAI) therapy, and surgical thyroidectomy are used to inhibit thyroid hormone synthesis or destroy the thyroid structure. However, there is no targeted therapy against TRAb [3].
ATDs are relatively safe and accessible, making them the most prevailing initial treatment option for GD in most countries, including East Asia, Europe, and the United States [3-5]. In South Korea, ATDs are prescribed as the initial treatment for >90% of patients with GD [6]. However, ATD use is associated with high recurrence rate, approximately 30% to 70%, despite prolonged treatment duration of more than 12 to 18 months [7,8]. Definitive therapy such as RAI or thyroidectomy has a lower recurrence rate than ATDs [8]. RAI treatment destroys thyroid follicular cells through ionizing radiation, normalizing thyroid function in approximately 50% to 90% of patients within 3 to 12 months [9]. RAI is strongly recommended as the initial treatment by National Institute for Health and Care Excellence guidelines [10] and recommended in appropriate patients by European Thyroid Association and American Thyroid Association [11,12]. Surgical thyroidectomy rapidly controls thyrotoxicosis and is recommended in patients with thyroid malignancy or thyroid nodules. However, utilization of RAI and thyroidectomy as initial therapies has limits.
Untreated and sustained thyrotoxicosis is associated with arrhythmia, hypertension, heart failure, osteoporosis, mood changes, and death. Moreover, patients with recurrent GD have a higher risk of cardiovascular morbidity and mortality than those without GD [13]. Therefore, when selecting the initial treatment for GD, treatment effectiveness, along with quality of life, recurrence reduction, and underlying comorbidities, should be considered.
In this study, we evaluated the initial treatment failure rate according to each first-line treatment modality for GD using data from the largest retrospective cohort study in South Korea from 2004 to 2020. Moreover, we investigated trends in the initial treatment modalities for GD in South Korea.
METHODS
Data source
Data were extracted from the National Health Insurance Service (NHIS)-National Health Information Database (NHID) (number: NHIS-2022-1-208). Under the NHIS, individuals are required to have biennial medical examination at accredited hospitals as part of the public health insurance program. The associated dataset comprised comprehensive longitudinal health examination data, including socioeconomic and demographic information, smoking and alcohol consumption histories, physical activity, anthropometric measurements, and laboratory data. It also contains medical records classified according to the International Classification of Diseases, 10th Revision (ICD-10) codes and matched mortality data. Previous studies have extensively documented the utility and applications of the NHIS data [14,15].
This study was approved by the Institutional Review Board of Korea University Anam Hospital (IRB number: 2021AN0339). The NHIS database was anonymized and de-identified and did not require prior informed consent to be included in the analysis.
Study population
We selected 641,203 patients who were diagnosed with GD between January 1, 2002 and June 30, 2020. The index date was based on ICD-10 codes (E05) and the day of prescription of ATDs, RAI, or thyroidectomy. After excluding individuals diagnosed with GD in 2002 to 2003 (n=109,019), those with follow-up durations <6 months (n=11,130), aged <18 years (n=21,230), previously diagnosed with thyroid cancer (ICD-10 C73, n=43,479), and had undergone treatment with RAI or thyroidectomy for conditions other than GD (n=3,300), a cohort of 465,683 patients was identified for further study. After excluding 13,682 patients diagnosed with thyroid cancer during the follow-up period, we analyzed 452,001 patients with GD. Within this cohort, 442,990 (98.0%) patients received initial treatment with ATDs (ATD group), 2,981 (0.7%) with RAI (RAI group), and 6,030 (1.3%) with thyroidectomy (thyroidectomy group) (Fig. 1).
Schematic study design. RAI, radioactive iodine; ATD, antithyroid drug; PY, person-year.
Definitions of GD treatment and outcomes
The ATD group included patients who received antithyroid medications (propylthiouracil, methimazole, and carbimazole) two or more times for >90 days. Patients in the RAI group with a thyrotoxicosis diagnosis (ICD-10 E05) were treated with a single dose of at least 5 mCi within 6 months. The thyroidectomy group included patients who underwent thyroidectomy within 6 months.
The index date was defined as the first day on which a patient began the initial treatment for GD, irrespective of whether the treatment involved ATDs, RAI, or thyroidectomy. Each patient was followed-up from the index date until the earliest occurrence of any of the following events: treatment failure, death, loss of follow-up, or the date of the last data collection in the cohort (December 31, 2020). The definition of treatment failure differs according to the initial treatment modality. In the ATD group, treatment failure was defined under several conditions: (1) patients undergoing RAI or thyroidectomy after initiation of ATDs; (2) patients restarting ATDs after a cessation period >6 months after initial ATD discontinuation; and (3) patients prescribed ATDs for >2 years. In the RAI group, treatment failure was delineated under the following conditions: (1) patients receiving additional RAI >5 mCi after the initial RAI; (2) patients undergoing thyroidectomy after RAI treatment; (3) patients who continued ATDs for >6 months following RAI treatment; and (4) patients initiating ATDs therapy >6 months after receiving RAI treatment. In the thyroidectomy group, treatment failure was defined under the following conditions: (1) patients receiving RAI with ≥5 mCi; (2) patients undergoing total thyroidectomy; and (3) patients restarting ATD therapy more than 6 months after thyroidectomy.
Definitions of covariates
We used specific criteria for previous diseases. Hypertension was defined as blood pressure ≥140/90 mm Hg or prescribed antihypertensive medication under ICD-10 codes I10–I15. Diabetes was defined as a fasting plasma glucose level ≥126 mg/dL or the current use of prescribed glucose-lowering agents under ICD-10 codes E10–E14. Dyslipidemia was defined as a total cholesterol level ≥240 mg/dL or prescription of lipid-lowering agents under the ICD-10 code E78. Cardiovascular disease was defined as hospitalization or ≥2 outpatient visits with ICD-10 codes for conditions such as ischemic heart disease (I20–I25), stroke (I60–I69), or heart failure (I42–I43 or I50). Atrial fibrillation was defined as hospitalization or ≥2 outpatient visits with ICD-10 code I48. This study also examined several lifestyle determinants including smoking status (non-smoker, ex-smoker, and current smoker), alcohol consumption (none, ≤2 times per week, or ≥3 times per week), and socioeconomic status (SES), which was divided into tertiles (first, second, and third tertiles).
Statistical analysis
Normally distributed continuous data are presented as mean±standard deviation. Non-normally distributed continuous data are presented as median and interquartile range (IQR). Categorical data are presented as frequencies and percentages. All incidence rates are expressed as the number of cases per 1,000 person-years. The risk of treatment failure associated with the initial treatment modalities is presented as a hazard ratio (HR) and the corresponding 95% confidence interval (CI) using the Cox proportional hazard regression model. Adjusted HRs for various risk factors were estimated from models that incorporated variables such as age, sex, SES, comorbidities (hypertension, diabetes mellitus, dyslipidemia, cardiovascular disease, and atrial fibrillation), and health check-up variables (body mass index, systolic blood pressure, total cholesterol, fasting plasma glucose, smoking status, and alcohol consumption status). To evaluate the occurrence of treatment failure according to the initial dose of RAI treatment, we subdivided it into two categories: <10 and ≥10 mCi. Association between the initial dose of RAI treatment and treatment failure using a restricted cubic spline regression model. Data were fitted by a restricted cubic spline Cox proportional hazards regression model, and the model was conducted with 3 knots at the tertiles of dose (reference is median). Solid lines indicate HRs, and shadow shape indicate 95% CIs. The rug above x-axis is the density of the population along the initial dose of RAI treatment. All statistical analyses were performed using the SAS Enterprise Guide version 7.1 (SAS Institute Inc., Cary, NC, USA). Two-sided P values <0.05 were considered statistically significant.
RESULTS
Baseline characteristics of the study population
In the 452,001 patients with GD, the initial treatments were ATDs (n=442,990, 98.0%), thyroidectomy (n=6,030, 1.3%), and RAI (n=2,981, 0.7%) (Table 1). The prescription rate of the initial ATD treatment exhibited a stepwise increase from 96.2% in 2004 to 98.8% in 2020 (Supplemental Table S1). Meanwhile, the average prescription rates of RAI and thyroidectomy gradually decreased from 1.4% to 0.4% and 2.5% to 0.8%, respectively. The median duration of ATDs at initial treatment was 2.5 years (IQR, 1.1 to 5.8). Additionally, the median initial treatment dose of RAI was 10 mCi (IQR, 10 to 15). The mean age of the patients was 46.2±14.5 years and 70.8% was female. Both the mean age and proportion of females increased incrementally across the ATD, RAI, and thyroidectomy treatment groups. No significant differences were observed in the SES between the groups. The median follow-up duration for total patients with GD was 8.5 years (IQR, 4.5 to 12.5). The median follow-up duration differed significantly across the treatment groups; ATD at 8.5 years (IQR, 4.5 to 12.5), RAI at 10.5 years (IQR, 5.9 to 14.3), and thyroidectomy at 12.3 years (IQR, 8.4 to 14.6). The proportion of patients with a history of hypertension and cardiovascular disease was significantly lower in the ATD group than that in the RAI and thyroidectomy groups. However, the ATD group had a higher proportion of patients who smoked and consumed alcohol than the RAI and thyroidectomy groups.
Baseline Characteristics of Patients with Graves’ Disease according to Initial Treatment
Treatment failure risk in patients with GD according to initial treatment
Over a median follow-up period of 8.5 years, the proportion of treatment failure according to the initial treatment option was 58.5% (n=259,361), 21.3% (n=636), and 2.1% (n=126) in the ATD, RAI, and thyroidectomy groups, respectively (Fig. 1). We evaluated the clinical factors related to treatment failure using multivariate Cox regression analysis (Table 2, Fig. 2). After adjusting for age, sex, SES, comorbidities, and health check-up variables, the HR of treatment failure in the ATD group was 2.81 (95% CI, 2.51 to 3.14), compared with that in the RAI group (P<0.001) (Table 2). However, the thyroidectomy group has a lower risk of treatment failure compared to the RAI group (HR, 0.08; 95% CI, 0.06 to 0.11; P<0.001). We further analyzed the risk factors associated with treatment failure in each treatment group (Fig. 2). In the ATD group, younger age, male sex, current smoking status, and non-drinking were associated with treatment failure. In the RAI group, patients treated with ≥10 mCi had a lower treatment failure rate (HR, 0.63; 95% CI, 0.49 to 0.81) compared to those receiving <10 mCi. Univariate analysis revealed patients treated with ≥10 mCi RAI had 19% reduced treatment failure compared to those who treated with <10 mCi (Supplemental Table S2). According to the restricted cubic splines for the adjusted HR of treatment failure (Fig. 3), the dose of RAI associated with the lowest risk of treatment failure was 10 to 15 mCi. In the thyroidectomy group, low SES and current smoking status significantly increased the rate of treatment failure.
Risk of Treatment Failure according to Initial Treatments in Grave’s Disease Patients
Multivariate analysis on risk factors associated with treatment failure according to initial treatment in Graves’ disease patients. These results were analyzed in patients with all available data included in models (290,231 in the antithyroid drug [ATD] group, 1,761 in the radioactive iodine [RAI] group, and 3,700 in the thyroidectomy group). Bold values indicated statistically significant factors. CI, confidence interval.
Association between initial dose of radioactive iodine (RAI) treatment and treatment failure using a restricted cubic spline regression model. Graphs show hazard ratios for treatment failure according to the initial dose of RAI treatment adjusted for age, sex, socioeconomic status, hypertension, diabetes mellitus, dyslipidemia, cardiovascular disease, atrial fibrillation, body mass index, smoking status, and alcohol consumption. Data were fitted by a restricted cubic spline Cox proportional hazards regression model, and the model was conducted with 3 knots at the tertiles of dose (reference is median). Solid lines indicate hazard ratios, and shadow shape indicate 95% confidence intervals (CIs). The rug above x-axis is the density of the population along the initial dose of RAI treatment.
DISCUSSION
This large-scale longitudinal study elucidated the treatment patterns and corresponding failure rates associated with each treatment modality in patients with GD. From 2004 to 2020, ATDs were the most prescribed treatment for GD (98%), and the prescription rate gradually increased. However, the treatment selection rates for thyroidectomy and RAI decreased by 1.3% and 0.7%, respectively. After a median of 8.5 years of follow-up, 58.5%, 21.3%, and 2.1% of the patients in the ATD, RAI, and thyroidectomy groups, respectively, experienced recurrence.
ATDs are the preferred initial treatment for GD globally, with approximately 93.7% of patients with hyperthyroidism in South Korea were prescribed ATDs [6]. This treatment approach is also predominant in Europe, the Asia-Pacific, and the Middle East [4,16,17]. In the United States, RAI therapy was the preferred first-treatment modality for GD; however, ATDs therapy has recently emerged as the predominant choice [5]. Taking ATDs is efficient and accessible, with relatively mild side effects, such as pruritus and skin rashes, which can be controlled with antihistamines [1]. However, the recurrence rate remains a major concern. Approximately half of the patients experienced recurrence even after 12 to 18 months of treatment [18]. Our results revealed that the treatment failure rate of ATDs was 58.5% (140 person-per 1,000 years). Several factors may have contributed to GD recurrence. Smoking has been identified as a dose-dependent risk factor for GD onset and recurrence. The underlying mechanisms are hypothesized to involve the activation of the sympathetic nervous system and modulation of immune responses [19,20]. We also demonstrated that smoking increased the recurrence risk by 20%. Fluctuation in the TRAb levels is another factor. The pathogenesis of GD is primarily associated with TRAb, and the normalization of thyroid function and TRAb levels can lead to remission. However, even after TRAb normalization has been achieved through 12 to 18 months of ATD treatment, TRAb levels are prone to fluctuation. Elevated TRAb levels can potentially lead to GD recurrence. Consequently, extended treatment duration may enhance the likelihood of achieving remission. In our study, despite an increase in the use of ATDs, the proportion of patients who continued long-term therapy—defined as treatment more than 2 years—remained consistently around 50%, with no significant increasing trends observed. Several studies have demonstrated that remission rates can be increased by up to 85% in patients who undergo ATD therapy for more than 5 years, without a significant increase in side effects [21,22]. Additionally, chronic inflammation and oxidative stress are important factors contributing to GD progression [23,24]. In our study, metabolic diseases were associated with recurrence among patients receiving ATDs. Additional studies are warranted to elucidate the underlying biological mechanisms contributing to this association.
Despite the safety and efficacy of ATDs administration, thyroidectomy and RAI are recommended as the first-line treatment for specific conditions such as recurrent hyperthyroidism, thyroid cancer, toxic nodular goiter, and cardiovascular instability. Thyroidectomy is beneficial for patients with large goiters or those who may have thyroid cancer [11]. Additionally, thyroidectomy is associated with a lower risk of recurrence compared to RAI or ATDs. Furthermore, health care expenses for 10 years post-treatment were also lower for patients who underwent thyroidectomy compared to those who received ATDs and RAI [25]. RAI destroys thyroid follicular cells through ionizing radiation; thus, 50% to 90% of patients with GD can achieve normalized thyroid function within 3 to 12 months [9]. Nevertheless, selection rate of RAI for treating GD in South Korea declined between 2002 and 2020, with only 0.4% of patients being prescribed RAI in 2020. RAI treatment relieves thyrotoxicosis within weeks and is recommended for patients with comorbidities that increase the surgical risk or require more rapid disease control. For treating GD, a sufficient RAI activity should be prescribed, with a mean dose of 10 to 15 mCi [12]. In our study, ≥10 mCi RAI treatment significantly reduced the recurrence rate compared to doses <10 mCi (HR, 0.63; 95% CI, 0.49 to 0.81). However, increasing the dose to >15 mCi did not confer additional benefits compared with doses ≤15 mCi. These results may be attributable to recurrence-related conditions such as the presence of a goiter or longer disease duration. A major concern in prescribing RAI is the potential for cancer development. In a recent meta-analysis of 479,452 patients with GD [26], the overall cancer risk in patients who received RAI was not different from that in those who did not receive RAI therapy, except for thyroid cancer. However, our previous study demonstrated that the overall cancer risk, including thyroid cancer, in patients with GD who received RAI therapy compared to those who did not, were not significant (adjusted HR, 0.92; 95% CI, 0.81 to 1.04) [27]. Definitive therapy-induced lifelong hypothyroidism is another factor of the low prescription rate. In our study, of the 2,981 patients who received initial RAI therapy, 71.1% (n=2,119) were prescribed levothyroxine after RAI therapy. Levothyroxine treatment is associated with an impaired sense of well-being, quality of life, high body mass index, and hyperlipidemia compared to normal controls or patients with euthyroid GD [28]. However, in a 14- to 21-year follow-up of patients randomly assigned to receive ATDs, RAI, or surgery, no differences in quality of life were observed between the three treatment groups [29].
This study had several limitations. First, we could not evaluate the serum thyroid function test values, serum TRAb levels, free thyroxine levels, and goiter size, which are remarkably related with recurrence rate. However, we extensively investigated other risk factors for recurrence such as smoking, age, and sex, and adjusted for these variables in the multivariate analysis. Second, the diagnosis of GD was based on the NHID dataset, and transient thyrotoxicosis or toxic adenoma was included. To overcome this, we strictly defined GD as patients who had a claim for GD and at least 3 months of ATD therapy. Although we strictly defined GD patients as those with a GD claim code who had used ATDs for at least 3 months, a limitation of our cohort is that transient thyrotoxicosis or toxic adenoma patients were not completely excluded. Third, our study included only Korean men and women; therefore, our findings should be confirmed in an independent population. Fourth, because of the nature of the claims data, there could be a discrepancy between the actual disease and the recorded claims, and we could not evaluate the causal relationship and underlying mechanisms. Lastly, our data did not include an analysis of treatment failure rate according to the different levels of medical facilities. Despite these limitations, our study has several strengths. We used the Korean NHID database and evaluated the overall treatment pattern of GD and the failure rate according to treatment modalities using a nationwide health insurance database for a relatively long period.
In conclusion, ATDs are the predominant initial treatment modalities for GD in South Korea, followed by thyroidectomy and RAI. The recurrence rate of GD is notably high, exceeding 50%, in patients initially treated with ATDs. Despite its effectiveness in controlling GD and low recurrence rate, RAI is used in only 0.7% of patients with GD. Further investigation is warranted to elucidate the factors contributing to the minimal utilization of RAI as an initial treatment option in South Korea.
Supplementary Material
Supplemental Table S1.
Distribution of First Treatment for Graves’ Disease Patients by Year of Diagnosis
Supplemental Table S2.
Univariable Analysis on Risk Factors Associated with Treatment Failure among Graves’ Disease Patients
Notes
CONFLICTS OF INTEREST
No potential conflict of interest relevant to this article was reported.
ACKNOWLEDGMENTS
This research was supported by a grant from the Korean Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health and Welfare, South Korea (grant number: HC21C0078). The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication.
AUTHOR CONTRIBUTIONS
Conception or design: K.J.K.(Kyeong Jin Kim), J.C., S.G.K. Acquisition, analysis, or interpretation of data: J.A.K., K.J.K.(Kyeong Jin Kim), J.C., K.J.K.(Kyoung Jin Kim), S.G.K. Drafting the work or revising: J.A.K., K.J.K.(Kyeong Jin Kim), E.S., J.H.Y., N.H.K.(Nam Hoon Kim), H.J.Y., J.A.S. Final approval of the manuscript: N.H.K.(Nan Hee Kim), K.M.C., S.H.B., S.G.K.