Graves' disease is the most common cause of hyperthyroidism. In Korea, the overall incidence is 0.72 per 1,000 persons/year (males 0.40 and females 1.03), and its prevalence is 3.40/1,000 persons (males 2.09 and females 4.70) [1]. The treatment modalities involve antithyroid drugs (ATDs), radioactive iodine therapy (RAI), and total thyroidectomy. ATDs are used as an initial treatment for Graves' disease in 97.1% of patients in Korea and are generally well-tolerated agents [2]. ATDs can cause transient and minor adverse reactions such as benign leukopenia, urticaria, rash, arthralgia, and arthritis in 1% to 5% of patients, and agranulocytosis, severe toxic hepatitis, vasculitis, and lupus-like syndrome are very rare but fatal. Agranulocytosis, which is defined as a granulocyte count below 500/mm³, is a life-threatening adverse reaction of ATDs [3]. Reports on the characteristics of agranulocytosis are limited because its incidence is rare, occurring in 0.1% to 0.5% of patients [4,5]. Several studies have shown that ATD-induced agranulocytosis (AIA) occurs more frequently in elderly patients and within the first 3 months of treatment [6,7,8]. In Korea, the largest AIA study included 13 patients in a single center [9], and other studies consisted only of several case reports [10,11,12,13,14,15]. Therefore, in the present study, we reviewed data from AIA patients at four tertiary hospitals in Korea and investigated their clinical characteristics.

In the present study, which is to date the largest in Korea, we analyzed 54 AIA patients. AIA occurred within the first 3 months after initiation of ATDs in 76% of patients, but complications developed even after 3 years of treatment or in relapse cases treated with the same drugs in previously unaffected patients. Five patients with fever and mild neutropenia (ANC, 500 to 1,000/mm³) developed AIA within 5 days and their ANC and febrile symptoms were recovered after ATDs were discontinued. The patients treated with MMI showed lower ANC levels, but their recovery times from AIA did not differ.

ATDs are the common treatment modality for Graves' disease in Korea [2]. ATDs are associated with a variety of minor adverse effects including urticaria, arthralgia, and gastrointestinal upset in 1% to 5% of patients as well as life-threatening complications such as agranulocytosis and hepatitis. AIA risk factors are not well established because of its low incidence of occurrence. Recently, Nakamura et al. [7] analyzed 754 AIA patients, the largest study to date. The results showed that AIA tends to occur abruptly within 3 months after initiation of therapy, and elderly patients and females are more susceptible to AIA. In our study, 76% of patients were diagnosed with AIA within the first 90 days of treatment, similar to the observations in other published reports [7,9]. Importantly, AIA can occur even 3 years or more after the initiation of ATD therapy or in relapse patients with a previously uneventful course of the same drug therapy [4,16]. Although controversial, several studies have reported that elderly patients are susceptible to agranulocytosis [6,7], and have a higher rate of death [17]. In our study, the mean age of AIA patients and fatal cases were 38 years and 41 years, respectively. The mean onset age of Graves' disease was 44 years according the Medicare claims data provided by the Health Insurance Review and Assessment Service in Korea [1]. Therefore, it was difficult to make conclusions regarding the use of age as a risk factor of AIA. Other risk factors for AIA include the types and dose of ATDs. Several studies reported that patients with low-dose MMI therapy showed a lower risk of agranulocytosis compared with a high-dose MMI or PTU [4,6,18,19]. We could not determine an association with the dose of drugs because various dosages were initiated in patients with AIA. In our study, 72% patients of AIA were given MMI. Although, it is difficult to make conclusions on the basis of this study, it is considered that MMI is a much more common prescription than PTU; for example, from 2007 to 2011, MMI was prescribed to 54.8% with Graves' disease, whereas PTU was prescribed to 42.9% patients with Graves' disease in Korea [1]. Patients taking MMI showed more severe agranulocytosis, but the recovery time was not different than in patients taking PTU. There are no studies directly comparing PTU and MMI; thus, it is difficult to interpret the meaning of more severe agranulocytosis in patients taking MMI. Five febrile patients with mild neutropenia (ANC, 500 to 1,000/mm³) developed agranulocytosis within 5 days. During the same study period, two patients with mild neutropenia without symptoms of fever or sore throat were taking MMI for 5 and 42 months. One patient who developed AIA 5 months after the initiation of drug therapy showed recovered ANC levels 3 months later without any treatment change and another patient had persistent mild neutropenia without any symptoms and signs despite continuing MMI. Therefore, cessation of ATDs should be considered in symptomatic neutropenic patients.

Eighty percent of patients (43/54) with agranulocytosis presented with fever and sore throat due to nasopharyngeal bacterial infection. Broad-spectrum antibiotic therapy is essential with cessation of the causative drug for the treatment of agranulocytosis [16]. The usefulness of G-CSF or GM-CSF has been controversial; several reports showed their effectiveness in shortening the recovery time of agranulocytosis and reducing morbidity and mortality [20,21,22]. Because this study was performed retrospectively, we could not determine the effects of GM-CSF on shortening the recovery time from AIA. Agranulocytosis is an autoimmune-mediated response as demonstrated by anti-granulocyte and anti-neutrophil cytoplasmic antibodies present in immunofluorescence and cytotoxicity assays [23,24]. Therefore, steroid therapy is occasionally employed for the treatment of AIA, although the effects are questionable.

Development of sepsis increases mortality when the occurrence of AIA is not recognized and therapy is delayed, and mortality was higher in elderly patients [17]. Since AIA develops suddenly and prevention is difficult, providing adequate information to patients is important. Patients need to understand the early signs and symptoms of AIA and should visit the hospital immediately if they occur. The WBC and ANC of patients should be monitored once symptoms occur and the ATDs discontinued immediately until test results are available, especially for elderly patients. In the present study, 50% of the patients recovered 6 days after ATDs were discontinued, but the risk factors for delayed recovery could not be determined. The four fatality cases developed sepsis 4 to 7 days after fever, but the risk factors for fatality could not be determined due to the small number of patients.

In most cases, agranulocytosis occurs abruptly with very low incidence, and various guidelines exist for the treatment of Graves' disease related to regular monitoring of WBC counts [5,25,26,27]. Nakamura et al. [7] reported that the time interval between the last day with normal WBC levels and onset of agranulocytosis was 33 days in most patients, although some patients showed a gradual decline in granulocyte count; therefore, routine WBC measurement is not recommended. However, Tajiri et al. [3] suggested that asymptomatic patients can be detected through routine CBC monitoring and rescued by discontinuing ATD therapy, and regular monitoring of WBC every 2 weeks for the first 2 months of therapy was recommended in Japan. In our study, six patients without any symptoms were diagnosed with agranulocytosis using routine WBC tests, and WBC levels were low compared with the normal range at 1 month prior to the onset of agranulocytosis. Therefore, although routine monitoring of WBC counts has not been considered cost-effective in all patients, but monitoring WBC for 3 months after the initiation of ATDs can be helpful for predicting and detecting agranulocytosis in asymptomatic patients. If low or declining WBC levels are observed in patients receiving ATD therapy, WBC counts should be monitored to prevent the development of more severe agranulocytosis.

There were several limitations to this study. First, we were unable to show the cumulative incidence of AIA according to duration of taking ATD. We could only estimate the incidence at one center (Samsung Medical Center), which was 0.12%. This incidence is in agreement with other published reports [4,5]. Another limitation of the study is that we did not determine changes in WBC before AIA development, which is a limitation of retrospective studies. In total, 67% patients (16/24) showed mild leukocytopenia within 1 month before AIA onset. Infrequently, patients showed a tendency to have declining leukocyte counts before the onset of AIA. Therefore, CBC monitoring in the early treatment phase could be beneficial. Third, we could not determine the cause of the difference in ANC levels between patients taking MMI and PTU. A study with a larger number of patients is necessary to determine the cause of the difference.

In conclusion, agranulocytosis tends to occur within 3 months after starting ATDs, with the majority of patients experience fever and sore throat. In clinical practice, physicians should be aware of AIA and inform patients of the side effects of ATDs, and consider routine CBC monitoring during the early treatment period. Patients with fever and mild neutropenia were regarded as having agranulocytosis and had to discontinue taking ATDs and undergo close follow-up.