Case–control study of risk factors of myelodysplastic
syndromes according to World Health Organization
classification in a Chinese population
Ling Lv,1 Guowei Lin,2 Xiang Gao,1 Cuie Wu,3 Junmin Dai,3 Yongchen Yang,4 Hejian Zou,1
Hengjuan Sun,4 Minghua Gu,1 Xi Chen,5 Hua Fu,3 and Liming Bao4,6
Risk factors of mydelodysplastic syndromes (MDS) remain largely unknown. We conducted a hospital-based
case–control study consisting of 403 newly diagnosed MDS patients according to World Health Organization
classification and 806 individually gender and age-matched patient controls from 27 major hospitals in
Shanghai, China, to examine relation of lifestyle, environmental, and occupational factors to risk of MDS. The
study showed that all MDS (all subtypes combined) risk factors included anti tuberculosis drugs [odds ratio
(OR)adj 5 3.15; 95% confidence interval (CI) 5 1.22–8.12] as an independent risk factor, benzene (ORadj 5 3.73;
95% CI 5 1.32–10.51), hair dye use (OR 5 1.46; 95% CI 5 1.03–2.07), new building and renovations (OR 5 1.69;
95% CI 5 1.11–2.00), pesticides (OR 5 2.16; 95% CI 5 1.22–3.82), and herbicides (OR 5 5.33; 95% CI 5 1.41–
20.10) as relative risk factors. Risk factors of MDS subtype refractory cytopenia with multiple dysplasia
(RCMD) were benzene (ORadj 5 5.99; 95% CI 5 1.19–30.16) and gasoline (ORadj 5 11.44; 95% CI 5 1.31–100.03)
as independent risk factors, and traditional Chinese medicines (OR 5 2.17; 95% CI 5 1.15–4.07), pesticides
(OR 5 2.92; 95% CI 5 1.37–6.25), and herbicides (OR 5 12.00; 95% CI 5 1.44–99.67) as relative risk factors.
Smoking tobacco was significantly associated with refractory anemia with excess of blasts (RAEB) (ORadj 5
2.43; 95% CI 5 1.02–5.77). Education is shown as an independent protective factor against all MDS (ORadj 5
0.90; 95% CI 5 0.83–0.99) and RCMD (ORadj 5 0.89; 95% CI 5 0.79–0.99). These findings suggest that multiple
modifiable behavioral, environmental, and occupational factors play a role in MDS etiology, and various MDS
subtypes may have different susceptibility. Am. J. Hematol. 86:163–169, 2011. VVC 2010 Wiley-Liss, Inc.
Introduction
Myelodysplastic syndromes (MDS) represent a heteroge-
neous group of neoplastic clonal stem cell disorders
characterized by clinical presentations of anemia, thrombo-
cytopenia, and leucopenia. MDS may be categorized into
subtypes according to histological, immunological, and
genetic characteristics. MDS was usually diagnosed by
French–American–British (FAB) classification with subtypes
including refractory anemia (RA), RA with ringed sidero-
blasts (RARS), RA with excess of blasts (RAEB), RAEB in
transformation (RAEB-T), and chronic myelomonocytic
leukemia (CMML) [1]. Since its publication in 2001, World
Health Organization (WHO) classification for MDS has become
widely adopted [2]. In the WHO MDS system, blast cutoff is
less than 20% compared to 30% in FAB system. Additional
WHO MDS subtypes include refractory cytopenia with multiple
dysplasia (RCMD), MDS with isolated del(5q), and MDS
unclassifiable (MDS-u) and the WHO MDS system does not
include CMML [3].
Secondary MDS is usually resulted from radiation and
chemotherapy. Little is known about the etiology of primary
or de novo MDS. Most previous studies on MDS risk fac-
tors focused on FAB MDS [4–6]. Here, we report a large
hospital-based case–control study of 403 WHO MDS cases
and 806 age and sex-matched controls in a Chinese popu-
lation to assess effects of lifestyle, environmental, and
occupational factors on MDS development.
Patients and Methods
Patients and controls. A total of 403 adult de novo MDS patients
were enrolled to the Shanghai Hematology Collaboration Study at 27
major hospitals in Shanghai, China, between June 2003 and November
2006. The patients were at least 18 years old, and MDS diagnoses
were made according to WHO classification [2]. Each MDS case was
matched with two controls on age (±5 years) and gender. The controls
were recruited during the same period at the hospital where the case
was enrolled. The controls were hospitalized patients for conditions not
related to hematological diseases. The most common diseases in the
control group were respiratory infections, acute gastrointestinal infections,
genitourinary diseases, cardiovascular diseases, and trauma. Those with
current or a history of any malignant or nonmalignant hematological dis-
eases were excluded from the control group. Diagnosis of tuberculosis
(TB), were based on medical history, physical examination, chest X-ray,
tuberculin skin test and confirmed by positive for Mycobacterium tubercu-
losis. The cases and controls were from the same regions and had simi-
lar occupations (Tables I and II). An informed consent was obtained from
each study participant before data collection. This study was reviewed
and approved by the Ethic Committee of Fudan University.
Data. The study participants (cases and controls) were face-to-face
interviewed by trained study coordinators using a standard question-
naire. To minimize recall bias, neither cases and controls nor inter-
*Correspondence to: Liming Bao or Hua Fu, Cincinnati Children’s Hospital
Medical Center, 3333 Burnet Avenue, Cincinnati, OH 45229, USA or School
of Public Health, Fudan University, 130 Dong An Road, Shanghai 200032,
China. E-mail: liming.bao@cchmc.org or hfu@shmu.edu.cn
Conflict of interest: Nothing to report.
1Departments of Occupational Medicine and Clinical Epidemiology, Huashan
Hospital, Fudan University, Shanghai, China; 2Departments of Hematology
and Clinical Epidemiology, Huashan Hospital, Fudan University, Shanghai,
China; 3Department of Occupational Health, School of Public Health, Fudan
University, Shanghai, China; 4Shanghai Children’s Hospital, Shanghai
Jiaotong University Affiliated Children’s Hospital, Shanghai, China; 5Center
for Clinical Molecular Medicine, Children’s Hospital of Chongqing Medical
University, Chongqing, China and 6Division of Human Genetics, Department
of Pediatrics, Cincinnati Children’s Hospital Medical Center and College of
Medicine, University of Cincinnati, Cincinnati, Ohio, USA
Ling Lv and Guowei Lin contributed equally to this work.
Contract grant sponsors: Shanghai Shen Kang Hospital Development
Center; (SHDC12007201); Shanghai Bureau of Public Health; (200702);
Chongqing Science and Technology Commission; (CSTC2009AB5217).
Received for publication 15 October 2010; Revised 10 November 2010;
Accepted 15 November 2010
Am. J. Hematol. 86:163–169, 2011.
Published online 29 November 2010 in Wiley Online Library (wileyonlinelibrary.
com).
DOI: 10.1002/ajh.21941
Research Article
VVC 2010 Wiley-Liss, Inc.
American Journal of Hematology 163 http://wileyonlinelibrary.com/cgi-bin/jhome/35105
viewers were informed of the study hypothesis. The interviewers were
not informed of study participant’s case/control status. Medication his-
tory (at least one month) in the past 5 years prior to at the time of
MDS diagnosis in cases or the time of interview in controls was sought.
The surveyed medications included anti-TB drugs, N-(4-methylben-
zene-sulfonyl)-N-butyl urea (D860) for diabetes, chloramphenicol, sulfo-
namides, diazepam, phenobarbital, colchicine, cyclophosphamide
(CTX), antihyperthyroidism drugs, antimalarial drugs, and traditional
Chinese medicines (angelica root, bezoar of ox, arsenic trioxide, and
tripterygium wilfordii hook). Information on hair dye use (frequency and
accumulative uses) was collected. Alcohol consumption was classified
as never and ever (former and current) drinkers. The amount of alcohol
intake was measured as 10 g of alcohol equivalent to either one bottle
of beer (5802600 ml), 25 ml wine, or 150 ml rice wine. The quantity of
alcohol consumption per week and estimated drinking duration (number
of years) were documented. For tobacco smoking, the study partici-
pants were classified as ever (current and former) smokers and never
smokers. Additional information on number of cigarettes smoked per
day, number of years smoked, and accumulative quantity of tobacco
consumption over lifetime estimated in pack-years (number of cigarette
packs per day 3 number of years) were collected from the smokers.
Other personal information collected included years of education, living
within 100 m of high-voltage power lines in the past year, living in new
buildings and/or newly renovated home/workplaces in the past year,
and a family history of malignancies.
Regarding occupational exposure, the hazards surveyed were ben-
zene, heavy metals, pesticides, herbicides, gasoline, diesel, glue/adhe-
sives, dyes, ionizing radiation, eproxy resin, and formaldehyde. The
work history queried jobs were held over each study participant’s entire
career. Assessment of occupational hazard exposure was performed
by a panel of experts who were familiar with workplace exposure in
Shanghai and its sounding regions using at least one, but often several
of the following sources: information recorded in the database at the
Shanghai Institute of Public Health (SIPH), which had historical expo-
sure measurements taken at factories since 1950s; records for surro-
gate factories, industries or jobs in the SIPH database; and Chinese
literature for relevant industries. All occupational hazard exposures
except benzene were recorded in a dichotomous classification of ‘‘ever’’
or ‘‘never.’’ Additional data sources used in benzene exposure assess-
ment included exposure measurements records in the SIPH database
for the jobs and locations where study participants worked; measure-
ment records and facility histories maintained at the factories where
study participants worked; onsite investigations including ad hoc walk-
through surveys and measurements conducted by a exposure assess-
ment team; records for surrogate factories industries or jobs in the
SIPH database; and reports on technology changes over time in key
industrial sectors. The committee members were blinded to case/con-
trol status of study participants. Exposure to benzene was categorized
into five exposure groups (EG): EG0 (nonexposure), EG1 (<1 mg/m3),
EG2 (1–10 mg/m3), EG3 (10–100 mg/m3), and EG4 (>100 mg/m3).
The ranges of these EGs were intended as approximate estimates of
exposure and not meant to imply any numerical precision of the esti-
mates. EG scores (EGS) with values of 0–4 were used to correspond-
ing EG.
Statistics analysis. Descriptive analyses were conducted using X2
tests. Odds ratios (ORs) and 95% confidence intervals (CIs) were used
to assess relation of the parameters to risk of MDS. Parameters with P
� 0.2 in the univariate analyses (anti-TB drugs, D860, Chinese medi-
cines, exposed to high-voltage power lines, new building/renovation,
benzene, pesticides, herbicides, gasoline, gules, tobacco smoking,
alcohol intake, hair dye use (�2 times a year), and education) were
subjected to multivariate conditional logistic regression analysis to
adjust for potential confounding factors. For variables with more than
two categories, a trend test was performed. The statistical tests were
two-sided. Probability values equal or less than 0.05 were considered
as statistically significant. All analyses were performed using STATA
software version 7.0 (StataCorp, College Station, TX).
Results
The median ages of cases and controls were 59 and 61
years, respectively, and MDS patients were male domi-
nance (Table I). Basic characteristics of cases and controls
including age, sex, regions, and occupations were compa-
rable (Tables I and II). Of 403 MDS cases, 211 (52.36%)
were 60 or younger and 192 (47.64%) over 60. Significant
difference in education was noted between cases and con-
trols (P < 0.01). The subtype distribution was 263 RCMD
(65.26%), including 17 RCMD-ringed sideroblast (RCMD-
RS), 99 RAEB (24.57%) (40 RAEB-1 and 59 RAEB-2), 25
MDS-u (6.20%), 8 RA (1.99%), 5 RARS (1.24%), and 3
MDS 5q- (0.74%) (Table III). RCMD and RAEB subtypes
had large numbers, and they were subjected for subse-
quent analyses.
Anti-TB regimens usually contain isoniazid, rifampicin,
ethambutol, or pyrazinamide. Our analysis revealed that
study participants who used anti-TB drugs before MDS
were more likely to develop all MDS (OR 5 2.11, 95% CI
5 1.11–4.02) and RCMD (OR 5 3.43; 95% CI 5 1.35–
8.71) (Table IV). Elevated risk of RAEB (OR 5 1.33; 95%
CI 5 0.47–3.75) was also noted among those who took
anti-TB drugs although the difference was not of statistical
significance. RCMD patients were more likely to have a his-
tory of taking Chinese medicines (OR 5 2.17; 95% CI 5
1.15–4.07). The numbers of study participants who used
other medications were too small for statistics analysis. For
instance, the numbers of study participants who used chlor-
amphenicol were 3 and 0 in case and control groups,
respectively, sulfonamides: 0 and 2, respectively, and CTX:
0 and 2, respectively. None of study participants reported a
TABLE III. Distribution of WHO Subtypes
Subtypes No. of cases (%) Male (%) Female (%)
Median
age (years)
RA 8 (1.99) 3 (1.32) 5 (2.84) 62
RARS 5 (1.24) 1 (0.44)) 4 (2.27) 79
RCMD 263 (65.26) 148 (65.20) 115 (65.34) 57
RAEB 99 (24.57) 65 (28.63) 34 (19.32) 63
RAEB-1 40 (9.26) 25 (11.01) 16 (8.52) 63
RAEB-2 59 (14.64) 38 (16.74) 21 (11.93) 60
MDS-u 25 (6.20) 9 (3.96) 16 (9.09) 61
5q- syndrome 3 (0.74) 1 (0.44) 2 (1.14) 74
All-MDS (total) 403 227 176 59
TABLE I. Characteristics of Cases and Controls
Cases (%) Controls (%)
(n 5 403) (n 5 806) X2 P
Median age (years) 59 (20–86) 61 (24–88)
�60 211 (52.36) 388 (48.14)
>60 192 (47.64) 418 (51.86) 1.91 0.17
Sex
Female 176 (43.67) 352 (43.67)
Male 227 (56.33) 454 (56.33) 0.01 0.94
Places of residence
Metropolitan Shanghai 263 (65.26) 564 (69.98) 3.62 0.31
Suburb Shanghai 27 (6.70) 54 (6.70)
Cities in neighboring provinces 69 (17.12) 121 (15.01)
Rural areas in neighboring
provinces
44 (10.92) 67 (8.31)
Education (years)
None 48 (11.91) 44 (5.46)
�5 88 (21.84) 122 (15.14)
6–9 119 (29.53) 240 (29.78)
0–12 81 (20.10) 219 (27.17)
>12 59 (14.64) 167 (20.72) 32.4 <0.01
Missing data 8 (1.99) 14 (1.74)
TABLE II. Occupations of Cases and Controls
Cases
[n 5 403 (%)]
Controls
[n 5 806 (%)]
Clerks 46 (11.41) 122 (15.14)
Armed forces 20 (4.96) 50 (6.20)
Plant and mechine operators 73 (18.11) 170 (21.09)
Technicans 26 (6.45) 33 (4.09)
Service and sales workers 33 (8.19) 84 (10.42)
Elementary occupations 31 (7.69) 61 (7.57)
Skilled/agricultural 72 (17.87) 87 (10.79)
Craft workers 24 (5.96) 43 (5.33)
Managers 12 (2.98) 8 (0.90)
164 American Journal of Hematology
research article
history of taking diazepam, phenobarbital, colchicine, anti-
hyperthyroidism drugs, and antimalarial drugs.
Hair dye users at frequency of twice a year or more showed
an elevated risk of MDS (OR 5 1.46; 95% CI 5 1.03–2.07)
(Table IV). The trend test for number of hair dye users per
year was statistically significant (Ptrend 5 0.02). But the trend
test for lifetime hair dye users did not attain statistically signifi-
cant. The data also revealed significantly increased risk of all
MDS (OR 5 1.69, 95% CI 5 1.11–2.00) and marginal ele-
vated risks of RAEB (OR 5 1.65, 95% CI 5 0.67–4.07) and
RCMD (OR 5 1.60, 95% CI 5 0.96–2.68) among those who
lived in new buildings and/or newly renovated homes/work-
places in the past year. Finally, a family history of malignan-
cies did not appear to have effects on MDS risk (Table IV).
Positive association was observed between smoking and
development of all MDS (OR 5 1.44; 95% CI 5 1.04–2.01)
and RAEB (OR 5 2.32; 95% CI 5 1.12–4.77) (Table V).
The trend for accumulative smoking and risk of RAEB was
significant (Ptrend < 0.001), suggesting dose-dependent
relationship. With regard to alcohol intake, although
increases in risk of MDS were not statistically significant,
the P values of the trend tests for weekly alcohol consump-
tion and years of drinking attained statistically significant,
indicating a dose-response pattern between alcohol intake
and RARB risk (Table V). We did not obtain information on
types of liquor from our study participants and therefore,
possible effects of different liquors such as beer, wine, and
hard liquor on MDS susceptibility were not ascertained.
Table VI shows that study participants with none or ele-
mentary school education (�5 years) were more suscepti-
ble to all MDS (OR 5 20.67; 95% CI 5 2.74–156.1 and
OR 5 1.65; 95% CI 5 1.00–2.72, respectively). Moreover,
TABLE V. Relationship Between Smoking Tobacco and Alcohol Consumption and Susceptible to MDS
Variables
All MDS RAEB RCMD
OR (95% CI) Cases/controls OR (95% CI) Cases/controls OR (95% CI) Cases/controls
Smoking
Never 1 262/563 1 62/144 1 174/355
Ever (former and current) 1.44 (1.04–2.01) 141/243 2.32 (1.12–4.77) 37/54 1.10 (0.73–1.65) 89/171
Current 1.37 (0.89–2.12) 71/118 2.33 (0.95–5.73) 17/23 1.01 (0.59–1.72) 44/89
Former 1.48 (0.94–2.33) 70/125 2.28 (0.83–6.27) 20/31 1.20 (0.69–2.01) 45/82
Duration (years)
1–19 1.32 (0.67–2.61) 22/37 3.29 (0.82–13.15) 8/14 1.39 (0.58–3.32) 13/28
�20 1.22 (0.91–1.63) 108/96 2.35 (0.97–5.68) 27/36 0.96 (0.61–1.53) 68/134
Trend test P-trend 5 0.44 P-trend 5 0.02 P-trend 5 0.85
No. of cigarettes per day
1–19 1.21 (0.82–1.78) 62/99 3.95 (1.41–11.05) 18/22 1.05 (0.61–1.80) 34/70
�20 1.06 (0.75–1.50) 73/138 1.08 (0.42–276) 19/31 1.09 (0.64–1.88) 50/94
Trend test P-trend 5 0.52 P-trend 5 0.07 P-trend 5 0.53
Pack-years
1–19 1.11 (0.76–1.61) 68/119 1.43 (0.45–4.61) 7/27 0.96 (0.57–1.63) 39/85
�20 1.06 (0.73–1.55) 61/116 2.66 (1.06–6.69) 27/22 1.08 (0.60–1.92) 46/80
Trend test P-trend 5 0.20 P-trend 5 <0.001 P-trend 5 0.55
Alcohol consumption
Never 1 331/684 1 80/175 1 216/435
Ever 1.26 (0.89–1.78) 72/122 1.86 (0.94–3.69) 19/23 1.05 (0.69–1.60) 47/91
Current 1.40 (0.82–2.38) 30/51 2.39 (0.88–6.49) 9 /8 0.80 (0.40–1.60) 16/42
Former 1.26 (0.80–1.99) 42/64 1.33 (0.54–3.34) 10/15 1.36 (0.79–2.35) 31/44
Alcohol drinking per week
<70 g 1.00 (0.67–1.49) 49/98 1.34 (0.59–3.02) 12/18 0.99 (0.64–1.54) 41/82
�70 g 1.68 (0.87–3.24) 20/24 3.31 (0.96–11.35) 7/4 1.00 (0.16–6.14) 3/6
Trend test P-trend 5 0.28 P-trend 5 0.03 P-trend 5 0.95
Duration (years)
1–19 1.60 (0.90–2.84) 25/38 1.55 (0.43–5.57) 5/9 1.51 (0.72–3.17) 15/26
�20 1.07 (0.70–1.64) 46/83 2.09 (093–4.69) 14/14 0.90 (0.53–1.52) 31/63
Trend test P-trend 5 0.34 P-trend 5 0.05 P-trend 5 1.00
10 g alcohol 5 a can of beer (580/620 ml) or 25 ml of red wine or 150 ml of rice wine.
Values of statistical significance are indicated in bold.
TABLE IV. Relationship Between Nonoccupational Exposures and Risk of MDS
Variables
All MDS RAEB RCMD
OR (95% CI) Cases Controls OR (95% CI) Cases Controls OR (95% CI) Cases Controls
Anti-TB drugs 2.11 (1.11–4.02) 19 18 1.33 (0.47–3.75) 6 9 3.43 (1.35–8.71) 12 7
D860 0.44 (0.15–1.31) 4 18 1.33 (0.22–7.98) 2 3 0.18 (0.02–1.41) 1 11
Traditional Chinese medicines 1.54 (0.93–2.55) 32 45 0.56 (0.20–1.54) 6 19 2.17 (1.15–4.07) 18 22
Hair dyes 1.24 (0.96–1.62) 170 303 1.00 (0.59–1.70) 37 74 1.19 (0.86–1.65) 113 207
<2 time per year 1.05 (0.68–1.63) 49 106 1.00 (0.39–2.53) 10 18 1.17 (0.69–1.99) 36 79
�2 times per year 1.46 (1.03–2.07) 97 146 0.92 (0.47–1.82) 20 43 1.20 (0.77–1.87) 61 95
Trend test P-trend 5 0.02 P-trend 5 0.93 P-trend 5 0.15
Total lifetime uses
<70 1.28 (0.97–1.68) 160 283 1.00 (0.56–1.78) 31 64 1.27 (0.91–1.77) 111 197
�70 0.92 (0.34–2.51) 7 19 1.00 (0.24–4.25) 3 10 0.26 (0.03–2.22) 2 9
Trend test P-trend 5 0.22 P-trend 5 0.68 P-trend 5 0.47
Living within 100 m of high-voltage power lines 1.30 (0.97–1.75) 96 159 1.34 (0.70–2.56) 19 31 1.25 (0.87–1.79) 64 107
New building/renovation 1.69 (1.11–2.00) 47 60 1.65 (0.67–4.07) 10 13 1.60 (0.96–2.68) 30 40
Family history of malignancy 1.00 (0.74–1.36) 91 185 0.81 (0.45–1.47) 21 51 0.89 (0.61–1.32) 56 121
Values of statistical significance are indicated in bold.
American Journal of Hematology 165
research article
an elevated risk of RCMD was noted among those with
either none (OR 5 2.30; 95% CI 5 1.20–4.44) or primary
education (OR 5 1.62; 95% CI 5 0.88–3.
本文档为【染发与癌症】,请使用软件OFFICE或WPS软件打开。作品中的文字与图均可以修改和编辑,
图片更改请在作品中右键图片并更换,文字修改请直接点击文字进行修改,也可以新增和删除文档中的内容。
该文档来自用户分享,如有侵权行为请发邮件ishare@vip.sina.com联系网站客服,我们会及时删除。
[版权声明] 本站所有资料为用户分享产生,若发现您的权利被侵害,请联系客服邮件isharekefu@iask.cn,我们尽快处理。
本作品所展示的图片、画像、字体、音乐的版权可能需版权方额外授权,请谨慎使用。
网站提供的党政主题相关内容(国旗、国徽、党徽..)目的在于配合国家政策宣传,仅限个人学习分享使用,禁止用于任何广告和商用目的。