J C E M

B r i e f

O N L I N E

R e p o r t E n d o c r i n e

C a r e

Antithyroid Drug-Induced Hematopoietic Damage:

A Retrospective Cohort Study of Agranulocytosis and
Pancytopenia Involving 50,385 Patients with Graves
Disease
Natsuko Watanabe, Hiroto Narimatsu, Jaeduk Yoshimura Noh,
Takuhiro Yamaguchi, Kazuhiko Kobayashi, Masahiro Kami, Yo Kunii,
Koji Mukasa, Kunihiko Ito, and Koichi Ito
Ito Hospital (N.W., J.Y.N., Y.K., K.M., Ku.I., Ko.I.), Tokyo 150-8308, Japan; Advanced Molecular
Epidemiology Research Institute (H.N.), Faculty of Medicine, Yamagata University, Yamagata 990-8560,
Japan; Biostatistics (T.Y.), Tohoku University Graduate School of Medicine, Sendai 980-8575, Japan;
Department of Hematology (K.K.), JR Tokyo General Hospital, Tokyo 151-8528, Japan; and Division of
Social Communication System for Advanced Clinical Research (M.K.), the Institute of Medical Science,
the University of Tokyo, Tokyo 108-8639, Japan
Context: Although antithyroid drug (ATD)-induced hematopoietic damage is a significant concern,
it has not been comprehensively investigated.
Objective: Our objective was to describe the clinical features of ATD-induced hematopoietic
damage.
Design and Setting: This was a retrospective cohort study in Tokyo, Japan.
Patients: Between January 1983 and December 2002, 50,385 patients at Ito Hospital were diagnosed with Graves disease. We retrospectively reviewed their medical, pathological, and laboratory records between January 1983 and December 2010.
Main Outcome: Incidence and clinical features of ATD-induced agranulocytosis and pancytopenia
were evaluated.
Results: Of 55 patients with documented hematopoietic damage, 50 had agranulocytosis and 5 had
pancytopenia. All of them received ATD, either methimazole (n 51) or propylthiouracil (n 4).
Median intervals between initiation of ATD therapy and the onset of agranulocytosis and pancytopenia were 69 d (range, 11233 d) and 41 d (range, 3297 d), respectively. Either anemia or
thrombocytopenia was also documented in seven of the 50 patients with agranulocytosis. Agranulocytosis was the first manifestation of hematopoietic damage in four of the five patents with
pancytopenia. Hematopoietic damage recovered with supportive measures including granulocyte
colony-stimulating factor (n 37), steroids (n 10), and other supportive measures (n 8) in 54
patients, whereas the remaining patient died of complications from infection. This study failed to
identify the risk factors for ATD-induced hematopoietic damage.
Conclusions: This study showed that ATD cause hematopoietic changes, which are occasionally
severe and potentially fatal. The pathogenesis of agranulocytosis and pancytopenia might overlap,
and additional studies are warranted to clarify this and to establish an optimal treatment strategy.
(J Clin Endocrinol Metab 97: E49 E53, 2012)

ISSN Print 0021-972X ISSN Online 1945-7197

Printed in U.S.A.
Copyright 2012 by The Endocrine Society
doi: 10.1210/jc.2011-2221 Received August 3, 2011. Accepted October 11, 2011.
First Published Online November 2, 2011

Abbreviations: ATD, Antithyroid drug; G-CSF, granulocyte colony-stimulating factor; GD,

Graves disease.

J Clin Endocrinol Metab, January 2012, 97(1):E49 E53

jcem.endojournals.org

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 December 2016. at 09:06 For personal use only. No other uses without permission. . All rights reserved.

E49

E50

Watanabe et al.

Antithyroid Drug-Induced Hematopoietic Disorders

ntithyroid drugs (ATD) have been used as the standard therapy for Graves disease (GD) since their
initial development in 1941. These agents, however,
cause various adverse reactions, of which hematopoietic organ toxicity is particularly well known. Use of
methimazole or propylthiouracil causes agranulocytosis in 0.1 0.5% of treated patients, and some events
become life-threatening (1).
Pancytopenia also occurs in some patients administered ATD. Its incidence, however, is much lower than that
of agranulocytosis, with only 42 patients reported to date
(213). Eight of those patients died due to hemorrhage or
infection, suggesting that the clinical course of patients
with pancytopenia is more serious than those with agranulocytosis. However, information regarding the clinical
and pathological features of pancytopenia is still insufficient, and there is no consensus regarding either its diagnosis or treatment. Furthermore, although hematopoietic
disorders after administration of ATD are extremely important complications, to date, there have been almost no
studies that have dealt with agranulocytosis and pancytopenia in a comprehensive manner.
Ito Hospital is one of the largest hospitals in Japan
specializing in thyroid diseases. We conducted a retrospective cohort study investigating the incidence and clinical features of agranulocytosis and pancytopenia that occurred after administration of ATD in patients with GD
who were treated at Ito Hospital between 1983 and 2002.

J Clin Endocrinol Metab, January 2012, 97(1):E49 E53

FIG. 1. Of the 50,385 patients, 186 patients with GD treated with

ATD who did not develop agranulocytosis or pancytopenia were
randomly extracted to evaluate the incidence of development of
agranulocytosis and pancytopenia. The Kaplan-Meier method was
used to evaluate the incidence of agranulocytosis and pancytopenia in
the patients with GD treated with ATD.

due to the concerns of fetal teratogenesis or transmission of methimazole through breast milk (14, 15).
Diagnosis of ATD-induced agranulocytosis was made when
patients showed a granulocyte count of less than 500/l after
ATD administration (16). ATD-induced pancytopenia was diagnosed when, in addition to agranulocytosis, the patients
showed a hemoglobin concentration of less than 11.0 g/dl and a
platelet count of less than 100 109/liter. We defined resolution
of ATD-induced agranulocytosis or pancytopenia as recovery of
granulocyte counts to 500/l or higher.

Statistical analysis

Patients and Methods

SAS version 9.1.3 (SAS Institute Inc., Cary, NC) was used for
all statistical analyses.

Study patients
We conducted a retrospective cohort study between January
1983 and December 2002.
A total of 50,385 patients at Ito Hospital were diagnosed with
GD. We retrospectively reviewed their medical, pathological,
and laboratory records at Ito Hospital and the hospitals where
the patients were transferred between January 1983 and December 2010.
The patients who were newly treated with ATD routinely
underwent blood examinations including complete blood cell
counts once every 2 wk. The patients who developed agranulocytosis or pancytopenia routinely underwent blood examinations every day. These patients were managed in hospitals.
The median age of the patients was 35 yr (range, 1295 yr).
Of them, 9,391 (19%) were male and 40,992 (81%) were female.
Gender information was not available in the remaining two patients. This study was approved by the institutional review board
of Ito Hospital.

Management of GD and diagnosis of ATD-induced

agranulocytosis and pancytopenia
Methimazole is usually the first-line therapy for GD at Ito
Hospital, although pregnant women receive propylthiouracil

Results
Incidences of ATD-induced agranulocytosis and
pancytopenia
Diagnoses of ATD-induced agranulocytosis were established in 55 patients. Of those, five patients were diagnosed as having ATD-induced pancytopenia. The cumulative incidence of ATD-induced agranulocytosis and
pancytopenia at 100 and 150 d after initiation of ATD
were 0.28% (95% confidence interval 0.20 0.36%)
and 0.29% (95% confidence interval 0.21 0.37%),
respectively (Fig. 1).
Clinical courses of patients with ATD-induced
agranulocytosis and pancytopenia
The clinical characteristics of the 55 patients with
ATD-induced agranulocytosis and pancytopenia are
shown in Table 1. Median intervals between initiation of
ATD therapy and the onset of agranulocytosis and pan-

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 December 2016. at 09:06 For personal use only. No other uses without permission. . All rights reserved.

J Clin Endocrinol Metab, January 2012, 97(1):E49 E53

TABLE 1.

jcem.endojournals.org

E51

Characteristics of patients with agranulocytosis and pancytopenia

Variables
No. of patients
Age (yr)
Sex (male/female)
ATD (methimazole/propylthiouracil)
Interval between initiation of ATD therapy and onset
of agranulocytosis/pancytopenia (d)
Daily dose at onset of agranulocytosis (mg/d)
Methimazole
Propylthiouracil
Cumulative dose at onset of agranulocytosis (mg)
Methimazole
Propylthiouracil
Laboratory findings at diagnosis of GD
White blood cell count (l1)
Granulocyte count (l1)
Hemoglobin (g/dl)
Platelet count (10 109/liter)
Total bilirubin (mg/dl)
Aspartate aminotransferase (IU/liter)
Alanine aminotransferase (IU/liter)
Blood urea nitrogen (mg/dl)
Creatinine (mg/dl)
Laboratory findings at onset of agranulocytosis/pancytopenia
White blood cell count (l1)
Granulocyte count (l1)
Hemoglobin (g/dl)
Platelet count (10 109/liter)
Febrile episode (38.0 C) n (%)
Documented infection n (%)
Initial treatment of agranulocytosis/pancytopenia (n)
G-CSF
Steroids
Supportive care
Patient fatalities n (%)

Patients with
agranulocytosis
alone
50
41 (1274)
3/47
46/4
69 (11233)

Patients with
pancytopenia
5
48 (38 67)
0/5
5/0
41 (3297)

30 (530)
300 (100 300)

20 (10 60)
NA

P value
0.12
0.57
0.51
0.42
0.55

954 (2532700)
9200 (8100 15000)

1568 (1200 2109)

NA

0.09

5200 (3200 8500)

2600 (1600 6300)
12.8 (10.114.2)
233 (89 339)
0.5 (0.32.1)
29 (14 91)
32 (16 179)
16 (10 25)
0.6 (0.31.0)

5900 (3700 6600)

3100 (2400 3500)
13.2 (10.6 14.6)
201 (188 244)
0.6 (0.4 0.7)
27 (26 43)
37 (2571)
13 (10 18)
0.5 (0.4 0.7)

0.79
0.46
0.62
0.43
0.75
0.73
0.41
0.52
0.25

1300 (400 4500)

18 (0 416)
12.2 (8.514.7)
250 (61 471)
46 (92)
37 (74)

900 (200 2700)

4 (0 108)
9.3 (9.111.0)
89 (36 100)
5 (100)
4 (80)

35
7
8
0

2
3
0
1 (20)

Unless indicated otherwise, results are shown as median (range). NA, Not applicable.

cytopenia were 69 d (range, 11233 d) and 41 d (range,

3297 d), respectively.
Of the 50 patients with agranulocytosis alone, seven
also showed either a hemoglobin concentration of less
than 11.0 g/dl or a platelet count of less than 100 109/
liter at the onset but not both and, hence, were not labeled
as pancytopenia. Of the five patents with pancytopenia,
development of agranulocytosis preceded the development of pancytopenia in four of them. One patient displayed thrombocytopenia at the onset of agranulocytosis.
Outcomes of the patients with ATD-induced agranulocytosis and pancytopenia are shown in Table 1 and Supplemental Table 1 (published on The Endocrine Societys
Journals Online web site at http://jcem.endojournals.org).
All of the patients with agranulocytosis alone were successfully treated with granulocyte colony-stimulating factor (G-CSF) (n 35), steroids (n 7), or supportive care
alone (n 8). The recovery periods with the different
therapies were 7 d (range, 222 d), 9 d (511 d), and 9 d

(4 21 d), respectively. Patients treated with G-CSF recovered earlier than those who did not receive it [7 d (222 d)
vs. 9 d (4 21 d) (P 0.004)].
All of the five patients with pancytopenia developed
febrile episodes, and four of them developed documented
infections including tonsillitis (n 4) and pneumonia (n
1). Pancytopenia was initially treated with G-CSF (n 2)
and dexamethasone 4 6 mg/d (n 3). In four of the patients, pancytopenia subsided within 14 d of onset of therapy; two patients treated with G-CSF recovered after 9
and 10 d, and two patients treated with dexamethasone
recovered 8 and 10 d after the onset. There were no differences in recovery periods between the two therapies.
Pancytopenia was refractory in the remaining patient
treated with dexamethasone (Supplemental Table 1). This
patient developed fatal gastrointestinal hemorrhage and
received 2000 mg antithymocyte Ig (38.8 mg/kg d) for 4 d
but died due to pneumonia 102 d after the onset of pancytopenia (Supplemental Fig. 1). Two patients developed

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 December 2016. at 09:06 For personal use only. No other uses without permission. . All rights reserved.

E52

Watanabe et al.

Antithyroid Drug-Induced Hematopoietic Disorders

transient decreases in neutrophil count after recovery. No

adverse effects of G-CSF or steroids were documented in
any of the 55 patients.
Pathological features
None of the patients with agranulocytosis underwent
bone marrow examination. Two patients with pancytopenia underwent bone marrow aspiration at the onset of
pancytopenia. In these patients, bone marrow samples
showed severe hypoplasia, similar to that seen in severe
aplastic anemia. (Supplemental Fig. 2).
Risk factors of ATD-induced agranulocytosis and
pancytopenia
In the probable risk factors of ATD-induced agranulocytosis, no significant factors were selected in the multivariate model (Supplemental Methods and Supplemental
Table 2). The characteristics of patients with GD and without agranulocytosis or pancytopenia are shown in Supplemental Table 3. There were no differences in the occurrence of minor adverse effects between the two types of
ATD. In this study, the risk factors of ATD-induced pancytopenia could not be definitively determined (Table 1).

Discussion
This study elucidated the clinical features of hematopoietic disorders that occur after the administration of ATD.
Of the 50,385 patients analyzed in the present study, five
patients developed pancytopenia after treatment for GD.
The data (not shown) as of March 2005 showed that
77.7% of the patients at Ito Hospital who had been diagnosed with GD were administered ATD. Using this estimation, the incidence of pancytopenia in our study group
can be estimated to have been approximately 0.01%. In
addition, the incidence of agranulocytosis was approximately 0.3% in this study (Fig. 1), which is in agreement
with that reported in the published literature (1).
In four of the patients in this study, pancytopenia was
preceded by agranulocytosis. In addition, agranulocytosis
and pancytopenia showed no manifest differences in their
intervals of onset from the start of ATD administration.
These facts indicate the possibility that agranulocytosis
and pancytopenia in these cases could belong to the same
disease entity, there probably being some overlap in their
pathogenesis, and a difference only in their severity. This
would mean that, as has been indicated by earlier basic
research (17, 18), ATD damage hematopoietic stem cells.
The thinking to date has been that an autoimmune mechanism that targets an antigen on neutrophils is involved in
the onset of agranulocytosis due to ATD (19). However,
if that hypothesis were valid, then erythrocytes and plate-

J Clin Endocrinol Metab, January 2012, 97(1):E49 E53

lets would remain normal, which is contrary to our present

observations. The pathogenic factors involved in erythropoietic organ disorders caused by ATD are diverse and
require further research. Analyses of the pathological findings of bone marrow of a greater number of cases may
prove effective in elucidating these pathogenic factors.
The prognosis of agranulocytosis after treatment has
improved due to advances in supportive therapy, including antibiotics, blood transfusions, G-CSF, etc. (20). In
fact, all of the patients included in the present study who
were diagnosed with agranulocytosis were successfully
treated. On the other hand, there are insufficient data regarding the treatment results for pancytopenia. Because
one of two patients who developed hematopoietic disorders that were equivalent to severe aplastic anemia died, it
indicates the possibility that the prognosis of severe pancytopenia is poor. GD itself is a benign disease. In clinical practice, because the use of ATD is common, physicians should be aware that this often-prescribed
therapy for a benign disease could itself lead to life-threatening complications.
Treatment of agranulocytosis with G-CSF seems to
yield good results (1). On the other hand, there is extremely limited information regarding the treatment of
pancytopenia. Treatments for pancytopenia include supportive therapy using blood transfusions, G-CSF administration, etc., and immunosuppression centered on
administration of antithymocyte globulin. However, because the risk of infection with antithymocyte globulin is
high, it must be used with great care. For patients with mild
pancytopenia, it is probably appropriate to administer
supportive therapy and monitor the clinical course as was
done with three patients with comparatively mild pancytopenia in this study. On the other hand, one of our
patients with severe pancytopenia who received only
supportive therapy also recovered. This indicates that immunosuppressive therapy may not be essential even for
severe cases. In this regard, treatment of ATD-induced
pancytopenia differs from treatment of severe aplastic
anemia. Further research is needed to clarify which patients require immunosuppressive therapy.
In this study, no significant risk factors for the onset of
agranulocytosis after ATD were identified (Supplemental
Table 2); both methimazole and propylthiouracil had similar incidences of development of agranulocytosis. Our
analyses also failed to elucidate any risk factors for pancytopenia, but that was probably inevitable in view of the
small number of patients who developed pancytopenia in
this study (Table 1). Thus, identification of risk factors for
the onset of hematopoietic damage proved unsuccessful.
This is in agreement with current consensus, because it
is thought that immunological responses that cannot be

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 December 2016. at 09:06 For personal use only. No other uses without permission. . All rights reserved.

J Clin Endocrinol Metab, January 2012, 97(1):E49 E53

predicted on the basis of other background factors are

involved in the onset of agranulocytosis (1). Other factors
than those that have been investigated to date, such as
genetic predisposition, will probably need to be taken into
consideration for future identification of risk factors for
agranulocytosis.
Unfortunately, our study has a number of limitations.
A major limitation of our study is the fact that it was a
retrospective study. However, prospective clinical studies
are very difficult to carry out in the case of diseases that
occur at a low incidence, such as the hematopoietic disorders caused by ATD. It will be necessary to apply other
investigative techniques, such as creation of an adverse
event database, to obtain more information. Optimal
treatment methods can then be developed on the basis of
those data.

jcem.endojournals.org

7.

8.

9.

10.

11.

12.

Acknowledgments
We thank Dr. Hiraku Mori (Showa University Fujigaoka Hospital, Tokyo, Japan) for providing us with clinical data and bone
marrow findings.
Address all correspondence and requests for reprints to:
Jaeduk Yoshimura Noh, M.D., Ph.D., Ito Hospital, 4-3-6
Jingumae, Shibuya-ku, Tokyo 150-8308, Japan. E-mail:
h-yoshimura@ito-hospital.jp.
Disclosure Summary: The authors have nothing to disclose.

13.

14.

15.

16.

References
1. Cooper DS 2005 Antithyroid drugs. N Engl J Med 352:905917
2. Biswas N, Ahn YH, Goldman JM, Schwartz JM 1991 Aplastic anemia associated with antithyroid drugs. Am J Med Sci 301:190 194
3. Escobar-Morreale HF, Bravo P, Garca-Robles R, Garca-Larana J,
de la Calle H, Sancho JM 1997 Methimazole-induced severe aplastic
anemia: unsuccessful treatment with recombinant human granulocyte-monocyte colony-stimulating factor. Thyroid 7:6770
4. Jakucs J, Pocsay G 2006 Successful treatment of methimazole-induced severe aplastic anemia with recombinant human granulocyte
colony-stimulating factor and high-dosage steroids. J Endocrinol
Invest 29:74 77
5. Levine B, Rosenberg DV 1954 Aplastic anemia during the treatment
of hyperthyroidism with tapazole. Ann Intern Med 41:844 848
6. Lopez-Karpovitch X, Ulloa-Aguirre A, von Eiff C, Hurtado-Monroy R, Alanis A 1992 Treatment of methimazole-induced severe
aplastic anemia with recombinant human granulocyte-monocyte

17.

18.

19.

20.

E53

colony-stimulating factor and glucocorticosteroids. Acta Haematol

87:148 150
Mezquita P, Luna V, Munoz-Torres M, Torres-Vela E, Lopez-Rodriguez F, Callejas JL, Escobar-Jimenez F 1998 Methimazole-induced aplastic anemia in third exposure: successful treatment with
recombinant human granulocyte colony-stimulating factor. Thyroid 8:791794
Moreb J, Shemesh O, Shilo S, Manor C, Hershko C 1983 Transient
methimazole-induced bone marrow aplasia: in vitro evidence for a
humoral mechanism of bone marrow suppression. Acta Haematol
69:127131
Pearce SH 2004 Spontaneous reporting of adverse reactions to carbimazole and propylthiouracil in the UK. Clin Endocrinol (Oxf)
61:589 594
Richardson JS, Sarkany I, Campbell CD 1954 Fatal case of marrow
aplasia after treatment with carbimazole; the effect of ACTH. Br
Med J 1:364 365
Michallet M, Sobh M, Morisset S, Kraghel S, Nicolini FE, Thomas
X, Bienvenu AL, Picot S, Nicolle MC, Vanhems P 2011 Risk factors
for invasive aspergillosis in acute myeloid leukemia patients prophylactically treated with posaconazole. Med Mycol 49:681 687
Yamamoto A, Katayama Y, Tomiyama K, Hosoai H, Hirata F,
Kimura F, Fujita K, Yasuda H 2004 Methimazole-induced aplastic
anemia caused by hypocellular bone marrow with plasmacytosis.
Thyroid 14:231235
Wilcox G, Wong R, Elliot PJ, Topliss DJ, Stockigt JR 2003 Recovery
From Carbimazole-Induced Aplastic Anemia. Int J Endocrinol
Metab 1:41 43
Abalovich M, Amino N, Barbour LA, Cobin RH, De Groot LJ,
Glinoer D, Mandel SJ, Stagnaro-Green A 2007 Management of thyroid dysfunction during pregnancy and postpartum: an Endocrine
Society Clinical Practice Guideline. J Clin Endocrinol Metab 92:S1
S47
Momotani N, Yamashita R, Makino F, Noh JY, Ishikawa N, Ito K
2000 Thyroid function in wholly breast-feeding infants whose
mothers take high doses of propylthiouracil. Clin Endocrinol (Oxf)
53:177181
Solomon DH 1986 Treatment of Graves hyperthyroidism. In: Ingbar SH, Braverman LE, eds. Werners the thyroid. 5th ed. Philadelphia: JB Lippincott; 9871014
Fibbe WE, Claas FH, Van der Star-Dijkstra W, Schaafsma MR,
Meyboom RH, Falkenburg JH 1986 Agranulocytosis induced by
propylthiouracil: evidence of a drug dependent antibody reacting
with granulocytes, monocytes and haematopoietic progenitor cells.
Br J Haematol 64:363373
Douer D, Eisenstein Z 1988 Methimazole-induced agranulocytosis:
growth inhibition of myeloid progenitor cells by the patients serum.
Eur J Haematol 40:9194
Akamizu T, Ozaki S, Hiratani H, Uesugi H, Sobajima J, Hataya Y,
Kanamoto N, Saijo M, Hattori Y, Moriyama K, Ohmori K, Nakao
K 2002 Drug-induced neutropenia associated with anti-neutrophil
cytoplasmic antibodies (ANCA): possible involvement of complement in granulocyte cytotoxicity. Clin Exp Immunol 127:9298
Andre`s E, Kurtz JE, Perrin AE, Dufour P, Schlienger JL, Maloisel F
2001 Haematopoietic growth factor in antithyroid-drug-induced
agranulocytosis. QJM 94:423 428

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 December 2016. at 09:06 For personal use only. No other uses without permission. . All rights reserved.