bs_bs_banner

Journal of Diabetes 7 (2015) 523–530

O R I G I N A L A RT I C L E

Prevalence and characteristics of non-obese diabetes in

Japanese men and women: the Yuport Medical Checkup
Center Study
Saori KASHIMA,1 Kazuo INOUE,2 Masatoshi MATSUMOTO3 and Kimihiko AKIMOTO4
1
Department of Public Health and Health Policy, and 3Department of Community-Based Medical System, Institute of Biomedical and Health
Sciences, Hiroshima University, Hiroshima, 2Department of Community Medicine, Chiba Medical Center, Teikyo University School of
Medicine, Chiba, and 4Akimoto Occupational Health Consultant Office, Tokyo, Japan

Correspondence
Saori Kashima, Department of Public
Health and Health Policy, Institute of
Biomedical and Health Sciences,
Hiroshima University, 1-2-3 Kasumi,
Minami-ku, Hiroshima 734-0037, Japan.
Tel: +81 82 257 5167
Fax: +81 82 257 5169
E-mail: saori.ksm@gmail.com
Received 19 January 2014; revised 19
August 2014; accepted 29 August 2014.
doi: 10.1111/1753-0407.12213
Abstract
Background: This study examined the prevalence and characteristics of type
2 diabetes in non-obese subjects to compare their cardiometabolic markers to
those without diabetes.
Methods: Data were used from 17 098 men and 17 199 women from a vol-
untary health checkup program between 1998 and 2006 conducted in Japan.
The prevalence of diabetes (fasting plasma glucose ≥ 7.0 mmol/L, hemoglobin
A1c ≥ 6.5%, or known diabetes) was calculated in each subgroup of body
mass index (group 1, <22; group 2, 22–25; group3, 25–27.5; and group 4,
≥27.5). The effect of diabetes and obesity on risk of an abnormal level of
cardiometabolic marker was evaluated with a logistic regression model.
Results: In men, the prevalence of diabetes was 9.5% in total, 6.4%, 9.4%,
11.3% and 16.2% in group1 through 4, respectively. In women, it accounted
for 4.3% in total, 2.4%, 4.5%, 8.7% and 12.3% per group, respectively. Non-
obese diabetic subjects (in group 1 and 2) accounted for 60.8% and 62.0% of
all the diabetic subjects in men and women, respectively. Non-obese popula-
tion accounted for 71.2% and 83.6% of all men and women, respectively.
Levels of cardiometabolic markers were higher in diabetic subjects than in
non-diabetic subjects in each subgroup. Diabetes was associated with each
abnormal level of cardiometabolic marker independent of obesity.
Conclusion: Over 60% of the diabetic subjects in this Japanese population
were not obese. Non-obese diabetes is not widely addressed and should be
considered for increased public attention. The elevated levels of cardiometa-
bolic markers may contribute to an increased risk of cardiovascular disease in
non-obese diabetes.
Keywords: body mass index, metabolic disease, non-obesity, prevalence, type
2 diabetes mellitus.

Significant findings of the study: Over 60% of diabetic subjects were not obese (BMI < 25) in this Japanese
population. Levels of cardiometabolic markers were higher in diabetic subjects than in non-diabetic subjects in
both obese and non-obese subjects.
What this study adds: Non-obese individuals comprise more than half of all persons with diabetes, and thus, need
greater attention. The levels of cardiometabolic markers may contribute to an increased risk of cardiovascular
disease in non-obese persons with diabetes.

© 2014 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and Wiley Publishing Asia Pty Ltd 523
Characteristics of non-obese diabetes
Introduction
Obesity, an excessive accumulation of body fat, is a
known risk for progression to type 2 diabetes.1–4 Many
mechanisms have been reported regarding the associa-
tion of obesity with diabetes, including a relative insuf-
ficiency in insulin action, insulin resistance, and
prohibition of insulin action.5,6 Accordingly, weight loss
by obese individuals is effective in controlling diabetes or
even for normal blood glucose levels.7 Recently, diabetes
in non-obese individuals has gathered attention.8 Indeed,
according to five cohort studies in the US, those of
normal weight (body mass index [BMI] < 25]) with dia-
betes was 12% (overall),9 and another review reported
that 20% of patients with diabetes, at least in northern
European countries, are non-obese.8 The key defect
leading to the development of hyperglycemia in type 2
diabetes in non-obese individuals is impaired pancreatic
insulin secretion and lowered insulin resistance, as com-
pared with obese individuals.10–13
Of note is that non-obese diabetic patients have a
similar increased risk of cardiovascular disease as obese
diabetic patients.14 Furthermore, one meta-analysis indi-
cated that adults who were normal weight at the time of
incident diabetes had higher cardiovascular and non-
cardiovascular mortality than adults who are overweight
or obese.9 Therefore, non-obese diabetes may have
underlying pathophysiological changes that may lead to
worse prognosis than obese diabetes. However, a pos-
sible mechanism that may explain this phenomenon has
yet to be elucidated.
In Asian countries where a major proportion of people
are non-obese, it is possible that “non-obese” diabetes
may be a major concern for public health policy.15–17
However, the prevalence of non-obese diabetes remains
to be reliably reported in Asian countries, including
Japan. The prevalence of non-obese diabetes in all
persons suspected of having diabetes may range from
60–70% in men and 50–60% in women based on two
references.3,18 These references, however, did not provide
exact figures. For example, in Japan, the Ministry of
Health, Welfare and Labor published a report of a survey
on the current status of diabetes in 2010, with rather small
subject numbers and the number of non-obese diabetic
persons among the total number of strongly suspected
patients was not mentioned.19 Of additional note, whether
non-obese individuals with diabetes have higher risks of
cardiovascular disease that matches obese individuals
with diabetes remains to be explored.
Therefore, we conducted a cross-sectional study using
data from a large dataset of the Japanese population to
address the prevalence of non-obese diabetes in the
Japanese population and to compare cardiometabolic
524 © 2014 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and Wiley Publishing Asia Pty Ltd
S. KASHIMA et al.
markers among diabetic and non-diabetic subjects accor-
ding to non-obese and obese status. Finally, we evaluated
the interaction effect between obesity and diabetes on risk
for an abnormal level of cardiometabolic marker.
Methods
Study subjects
Study subjects were voluntary participants who attended
a health checkup program between April 1998 and
March 2006 at the Yuport Medical Checkup Center in
Tokyo. The details have been described in our previous
studies.20,21 Through the study period of 8 years, 34 303
persons undertook a total of 97 365 checkups. For repeat
participants who attended the screening twice or more,
data from the first checkup was used. We excluded five
subjects for whom fasting plasma glucose (FPG), hemo-
globin A1c (HbA1c), or BMI were not obtained. Finally,
the dataset contained 34 297 persons (17 098 men and
17 199 women) for the analysis.
During the study period, all the checkups were per-
formed in the same manner, including blood measure-
ments used for this study.
Diagnosis of type 2 diabetes
Based on the Japan Diabetes Society (JDS)22 and Ameri-
can Diabetes Association (ADA) criteria,23 diabetes was
diagnosed through meeting one of the following three
criteria: (i) a FPG of ≥7.0 mmol/L; (ii) HbA1c ≥ 6.5%; or
(iii) known diabetes. Known diabetes was identified
when the participants reported the presence of diabetes
at a medical interview irrespective of their FPG levels.
Obesity classification according to BMI
BMI was calculated as weight as kilograms divided by
height as meters squared (kg/m2). The Japan Society for
the Study of Obesity (JASSO) defined BMI = 22 as an
optimum24–26 and ≥25 as obesity class 1 in Japan.25,27 The
committee also classified BMI > 30 as obesity class 2, but
the number of the participants belonging to the obesity 2
category was scarce in this study (2%). According to the
recommendations for BMI for an Asian population by
the World Health Organization, cut-off points of 27.5
were to be addressed as points for public health action.28
Thus, we classified the subjects into four groups accord-
ing to their BMI (group 1, <22; group 2, 22–25; group 3,
25–27.5; and group 4, ≥27.5).
Laboratory tests
A blood sample was obtained after overnight fasting and
measured at the Center’s laboratory. FPG and HbA1c
S. KASHIMA et al.
levels were measured using a Toshiba TBA-40FR Autoa-
nalyzer. FPG levels were measured by the hexokinase-
G6PD method (Denka Seiken, Niigata, Japan) with an
inter-assay coefficient of covariation of 3.0% or less.
HbA1c levels were measured by the latex immuno-
agglutinin method assigned by the JDS, with an inter-
assay coefficient of covariation of 1.7–2.1%. The JDS
value of HbA1c was converted into a National Glycohe-
moglobin Standardization Program (NGSP) assigned
value by adding 0.4% in this study.22 Other serum markers
of cardiometabolic risk used in this study were serum
levels of liver enzymes and lipids, and measurement of
white blood cell count. Liver enzymes were aspartate
aminotransferase, alanine aminotransferase and gamma-
glutamyltranspeptidase. Serum lipids were triglycerides,
and total and high-density lipoprotein cholesterol. Low-
density lipoprotein (LDL) cholesterol was estimated by
Friedewald’s equation.29
Anthropometric and non-blood indices
Non-blood data were the following: age, and systolic
and diastolic blood pressure. Blood pressures were
measured by a trained nurse using a sphygmomanom-
eter according to guidelines for health screening by the
Japanese Association for Cerebro-cardiovascular
Disease Control. Basically, blood pressure was mea-
sured once, and a second measurement was added when
high blood pressure was observed. Measuring arm
(right or left) was not specified.
Statistical analysis
The total number and the prevalence of diabetes and
basic characteristics of the 34 297 study subjects classi-
fied by BMI level were described. Men and women
subjects were analyzed separately due to the gender-
difference in the prevalence of diabetes and anthropo-
metric characteristics. Next, we classified these subjects
into non-diabetes and diabetes groups, and described
their basic characteristics. Analysis of covariance
(ANCOVA) tests adjusted for age were used to compare
the mean or median of each value between the non-
diabetes and diabetes groups in each BMI group.
Finally, we calculated a proportion of subjects with
abnormal levels of cardiovascular and metabolic
markers in the non-diabetes and diabetes groups. We
calculated the prevalence of an abnormal level of the
selected cardiometabolic markers from the four catego-
ries (systolic blood pressure from blood pressure, LDL
cholesterol from lipids, alanine aminotransferase from
liver enzymes, and white blood cell count).30,31 In terms of
blood pressure, for example, systolic blood pressure is
more robust than diastolic blood pressure as a cardio-
© 2014 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and Wiley Publishing Asia Pty Ltd
Characteristics of non-obese diabetes
vascular disease risk factor,32 and as such we chose sys-
tolic blood pressure and the other factors in the same
way. According to the reference range set by the Yuport
Medical Checkup Center, we adopted an abnormal level
for the cardiometabolic markers as follow: ≥140 mmHg
for systolic blood pressure, >4.1 mmol/L for LDL cho-
lesterol, and >8.5 × 109/L for white blood cell count.
Since the laboratory normal range for alanine amino-
transferase was 6 to 50 IU/L for men and 6 to 30 IU/L
for women, in this study we adopted as an abnormal
range of alanine aminotransferase as >50 IU/L for men,
and >30 IU/L for women.
In order to evaluate the effect of diabetes and obesity
on a risk of each or at least one abnormal level of car-
diometabolic marker, we first entered diabetes and
obesity status into the same model and calculated each
odds ratio (OR) and their confidence interval (CI)
through a logistic regression model. In this mode, we
used non-diabetic and non-obese (BMI < 25) as the ref-
erence category, and adjusted age as continuous vari-
ables, and disease history, which reflects the status of
person with a present or past history of cardiovascular
disease (hypertension, hyperlipidemia, cerebral infarc-
tion, myocardial infarction and angina pectoris)
(adjusted model). Next, in order to evaluate whether an
obese status modifies the effect of diabetes on a risk of an
abnormal level of cardiometabolic marker, we evaluated
the statistical interaction between diabetes and obesity
status by entering the interaction term into the same
adjusted model. Finally, if there was significant interac-
tion between diabetic and obese status, we calculated an
adjusted OR of diabetes for an abnormal level of cardio-
metabolic marker stratified by obesity status.
SPSS software (IBM, version 20.0J) was used for the
descriptive and logistic analyses. A P-value less than 0.05
was considered as statistically significant.
Ethics
For anonymous participation in epidemiological
research, informed consent was obtained at every
checkup. The review board of the Center approved this
study.
Results
The mean (standard deviation) of age of the study sub-
jects was 51.2 (13.1) in 17 098 men and 52.3 (13.0) years
in 17 199 women subjects. Tables 1 and 2 show the basic
characteristics of the men and women classified by BMI
into a non-diabetic and diabetic subgroup. In men, 9.5 %
(1619/17 098) were diabetic in total and the prevalence of
diabetes was 6.4% (327/5144), 9.4% (658/7025), 11.3 %
525
Characteristics of non-obese diabetes S. KASHIMA et al.

Table 1 Basic characteristics of the 17 098 study subjects of men classified by body mass index in diabetic and non-diabetic subgroups

Body mass index

<22 22–25 25–27.5 ≥ 27.5

Non Diabetes
n 4817 6367 2979 1316
Age (years) 49.7 ± 14.1 51.4 ± 12.9 50.9 ± 12.4 47.4 ± 12.6
History of cardiovascular disease [n (%)] 103 (2) 257 (4) 172 (6) 50 (4)
Fasting plasma glucose (mmol/L) 5.26 ± 0.47 5.42 ± 0.48 5.54 ± 0.48 5.61 ± 0.51
Hemoglobin A1c (%) 5.2 ± 0.4 5.3 ± 0.4 5.4 ± 0.4 5.4 ± 0.4
Systolic blood pressure (mmHg) 119.8 ± 16.3 126.4 ± 16.8 131.0 ± 16.5 134.6 ± 16.2
Diastolic blood pressure (mmHg) 73.1 ± 10.1 77.4 ± 10.6 80.3 ± 10.6 82.6 ± 10.5
Triglycerides (mmol/L) 0.95 (0.69, 1.33) 1.23 (0.89, 1.76) 1.50(1.08, 2.16) 1.72(1.17, 2.45)
HDL cholesterol (mmol/L) 1.50 ± 0.37 1.34 ± 0.32 1.25 ± 0.29 1.19 ± 0.26
LDL cholesterol (mmol/L) 3.16 ± 0.81 3.50 ± 0.79 3.66 ± 0.81 3.76 ± 0.84
Asparate aminotransferase (U/L) 20 (17, 24) 21 (18, 25) 23(19, 28) 26(22, 35)
Alanine aminotransferase (U/L) 17 (14, 23) 21 (16, 29) 27(20, 38) 36(24, 55)
Gamma-glutamyltranspeptidase (U/L) 21 (14, 34) 27 (17, 46) 35(22, 59) 44(26, 70)
White blood cell count (109/L) 5.8 ± 1.6 6.0 ± 1.6 6.2 ± 1.7 6.5 ± 1.5
Diabetes†
n 327 658 380 254
Age (years) 60.9 ± 9.1*** 59.8 ± 9.2*** 57.3 ± 10.0*** 52.9 ± 11.7***
History of cardiovascular disease [n (%)] 13 (4) 64 (10)*** 35 (9) 17 (7)
Fasting plasma glucose (mmol/L) 8.35 ± 2.81*** 8.16 ± 2.22*** 8.11 ± 2.06*** 8.45 ± 2.29***
Hemoglobin A1c (%) 7.5 ± 2.0*** 7.2 ± 1.5*** 7.1 ± 1.4*** 7.4 ± 1.7***
Systolic blood pressure (mmHg) 125.9 ± 18.3* 133.2 ± 17.5*** 135.9 ± 18.1** 140.3 ± 17.3***
Diastolic blood pressure (mmHg) 76.5 ± 11.0* 80.4 ± 10.8** 82.5 ± 10.6 85.9 ± 10.9***
Triglycerides (mmol/L) 1.15 (0.80, 1.69)*** 1.47 (1.05, 2.10)*** 1.66(1.13, 2.54)*** 1.91(1.31, 2.85)***
HDL cholesterol (mmol/L) 1.45 ± 0.38*** 1.29 ± 0.31*** 1.22 ± 0.28* 1.16 ± 0.27*
LDL cholesterol (mmol/L) 3.37 ± 0.86 3.55 ± 0.83 3.73 ± 0.85 3.94 ± 0.92***
Asparate aminotransferase (U/L) 21 (18, 26) 22 (18, 28)*** 25(21, 33)*** 29(22, 42)***
Alanine aminotransferase (U/L) 20 (15, 26)** 23 (17, 34)*** 30(21, 45)*** 40(27, 61)***
Gamma-glutamyltranspeptidase (U/L) 27 (17, 52)*** 34 (20, 59)*** 44(25, 73)*** 54(30, 94)***
White blood cell count (109/L) 6.1 ± 1.9*** 6.3 ± 1.6*** 6.4 ± 1.6** 6.8 ± 1.9***

Data are expressed as mean ± standard deviation, median (25 percentile, 75 percentile) or number (%). HDL, high-density lipoprotein; LDL,
low-density lipoprotein.
†
For the comparisons of mean of age and age-adjusted mean of each value between non-diabetes and diabetes by analysis of covariance, a
P-value of 0.05 was used to determine statistical significance (*P < 0.05, **P < 0.01, and ***P < 0.001).

(380/3359) and 16.2% (254/1570) in BMI group 1
through 4, respectively. In women, 4.3% (735/17 199)
were diabetic in total and the prevalence of diabetes was
2.4% (219/9085), 4.5% (237/5301), 8.7% (165/1888) and
12.3% (114/925) in BMI group 1 through 4, respectively.
In total, non-obese diabetic subjects (BMI < 25)
accounted for 60.8% (985/1619) and 62.0% (456/735) of
all the diabetic subjects in men and women, who
belonged to the non-obese population that occupied
71.2% (12 169/17 098) and 83.6% (14 386/17 199) of all
the men and women subjects. The median of cardiovas-
cular and metabolic markers such as systolic blood pres-
sure, LDL cholesterol, alanine aminotransferase, and
white blood cell count in diabetic subjects increased
throughout from group 1 to 4, both in men and women.
In general, these markers were higher in diabetic subjects
than in non-diabetic subjects in each group in both men
and women throughout group 1 to 4.
Figure 1 shows the proportion of men and women
with abnormal levels of cardiometabolic markers in the
groups for non-diabetes and diabetes. The total propor-
tion of subjects with one or more abnormal levels of
markers increased along with an increased BMI level and
the presence of diabetes among both men and women.
Among the non-obese group with diabetes (BMI < 25),
61% in men (49% in BMI group 1 and 67% in group 2)
and 65% of women (58% in BMI group 1 and 71% in
group 2) had at least one adverse cardiometabolic risk
factor. Those rates decreased to 44% among the non-
obese men without diabetes and 33% among non-obese

526 © 2014 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and Wiley Publishing Asia Pty Ltd
S. KASHIMA et al. Characteristics of non-obese diabetes

Table 2 Basic characteristics of the 17 199 study subjects of women classified by body mass index in diabetic and non-diabetic subgroups

Body mass index

<22 22–25 25–27.5 ≥ 27.5

Non Diabetes
n 8866 5064 1723 811
Age (years) 48.9 ± 13.4 55.4 ± 11.4 56.5 ± 11.3 53.9 ± 12.2
History of cardiovascular disease [n (%)] 145 (2) 175 (3) 107 (6) 42 (5)
Fasting plasma glucose (mmol/L) 5.01 ± 0.43 5.20 ± 0.47 5.32 ± 0.48 5.48 ± 0.52
Hemoglobin A1c (%) 5.2 ± 0.4 5.4 ± 0.4 5.5 ± 0.4 5.5 ± 0.4
Systolic blood pressure (mmHg) 114.4 ± 16.5 124.1 ± 17.5 129.7 ± 17.6 135.6 ± 17.8
Diastolic blood pressure (mmHg) 69.3 ± 10.1 74.6 ± 10.6 77.7 ± 10.5 81.2 ± 11.0
Triglycerides (mmol/L) 0.75 (0.56, 1.00) 0.98 (0.73, 1.37) 1.10(0.82, 1.54) 1.21(0.87, 1.68)
HDL cholesterol (mmol/L) 1.72 ± 0.37 1.56 ± 0.35 1.47 ± 0.32 1.41 ± 0.31
LDL cholesterol (mmol/L) 3.28 ± 0.86 3.69 ± 0.86 3.84 ± 0.85 3.91 ± 0.87
Asparate aminotransferase (U/L) 19 (16, 22) 20 (17, 23) 21(18, 25) 21(18, 26)
Alanine aminotransferase (U/L) 14 (11, 18) 16 (13, 21) 18(14, 25) 21(16, 30)
Gamma-glutamyltranspeptidase (U/L) 13 (10, 19) 15 (10, 22) 17(11, 27) 20(13, 33)
White blood cell count (109/L) 5.2 ± 1.8 5.5 ± 1.4 5.7 ± 1.3 6.0 ± 1.4
Diabetes†
n 219 237 165 114
Age (years) 61.3 ± 9.4*** 64.4 ± 7.9*** 62.3 ± 7.9*** 60.2 ± 11.6***
History of cardiovascular disease [n (%)] 19 (9)*** 20 (8)* 24 (15)** 12 (11)
Fasting plasma glucose (mmol/L) 7.69 ± 2.48*** 7.90 ± 2.40*** 8.22 ± 2.49*** 7.94 ± 2.07***
Hemoglobin A1c (%) 7.2 ± 1.5*** 7.5 ± 1.6*** 7.4 ± 1.6*** 7.3 ± 1.2***
Systolic blood pressure (mmHg) 128.1 ± 18.7*** 133.3 ± 19.4*** 137.3 ± 17.8** 143.8 ± 18.2**
Diastolic blood pressure (mmHg) 75.2 ± 10.9*** 77.7 ± 10.5 80.4 ± 11.0 83.7 ± 10.5
Triglycerides (mmol/L) 1.02 (0.75, 1.51)*** 1.32 (0.97, 1.76)*** 1.35(1.07, 1.99)*** 1.45(1.04, 1.92)***
HDL cholesterol (mmol/L) 1.60 ± 0.39*** 1.43 ± 0.31*** 1.39 ± 0.28** 1.42 ± 0.32
LDL cholesterol (mmol/L) 3.78 ± 0.91* 4.00 ± 0.83 4.14 ± 0.90** 4.15 ± 0.89*
Asparate aminotransferase (U/L) 20 (18, 24)* 21 (17, 27)* 22(19, 28)** 24(20, 35)***
Alanine aminotransferase (U/L) 18 (15, 24)** 20 (15, 30)*** 23(18, 35)*** 28(19, 48)***
Gamma-glutamyltranspeptidase (U/L) 16 (11, 29)*** 20 (13, 31)*** 23(15, 42)*** 29(15, 50)**
White blood cell count (109/L) 5.7 ± 1.4*** 6.0 ± 1.4*** 5.8 ± 1.3 6.4 ± 1.7***

Data are expressed as mean ± standard deviation, median (25 percentile, 75 percentile) or number (%). HDL, high-density lipoprotein; LDL,
low-density lipoprotein.
†
For the comparisons of mean of age and age-adjusted mean of each value between non-diabetes and diabetes by analysis of covariance, a
P-value of 0.05 was used to determine statistical significance (*P < 0.05, **P < 0.01, and ***P < 0.001).

women without diabetes. On the other hand, within the
obese group (BMI ≥ 25), a relatively high proportion
(73% of men and 62% of women) of subjects with abnor-
mal levels of cardiometabolic markers were also
observed among subjects without diabetes.
Table 3 shows the ORs for an abnormal level of car-
diometabolic marker associated with diabetes among
men and women that takes into account obesity
(BMI ≥ 25) as a possible confounder, and the interaction
between diabetes and obesity. In comparison with non-
diabetic subjects, the significant increased odds both of
diabetes and obesity for all abnormal levels of cardio-
metabolic markers were observed among both men and
women with diabetes. Among the cardiometabolic
markers in both men and women, the adjusted ORs for
the risk of diabetes to elevated alanine aminotransferase
were the highest. Namely, subjects with diabetes were 2.0
and 2.7 times more likely to have an abnormal level of
alanine aminotransferase than those without diabetes
among both men and women, respectively. Along with
diabetes, the highest adjusted ORs among those who
were obese was observed for risk of an abnormal level of
alanine aminotransferase (OR: 4.0 in men and 3.5 in
women).
The significant interaction effect between diabetes and
being obese was observed for only the risk of an abnor-
mal level of alanine aminotransferase in women
(P = 0.03). Thus, we performed a subgroup analysis
stratified by obesity status. The adjusted ORs for the risk
of diabetes to elevated alanine aminotransferase in
women were 3.0 (95% CI: 2.3, 3.8) in the non-obese
group and 2.4 (95% CI: 1.8, 3.2) in the obese group.
Thus, this indicated that the OR in the non-obese group
was higher than that in the obese group.

© 2014 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and Wiley Publishing Asia Pty Ltd 527
Characteristics of non-obese diabetes S. KASHIMA et al.

Figure 1 Proportion of subjects with abnormal levels of cardiometabolic markers in non-diabetes (non-DM) and diabetes (DM) classified by
body mass index (BMI). Abnormal levels of cardiometabolic markers (systolic blood pressure from blood pressure, low density lipoprotein
cholesterol from lipids, alanine aminotransferase from liver enzymes, and white blood cell count) are indicated in Table 3. The subjects were
classified according to the number of abnormal levels of markers (0, 1, 2, ≥3).

Table 3 Adjusted odd ratios† of diabetes and obesity (BMI ≥ 25) for each abnormal level of a cardiometabolic marker, and their interaction
between diabetes and obesity in men and women.

abnormal level Diabetes Obesity P for interaction of

of marker N (%) OR (95% CI) OR (95% CI) diabetes and obesity

Men
Systolic blood pressure (≥140 mmHg) 3780 (22.1) 1.4 (1.3, 1.6) 2.4 (2.2, 2.6) 0.63
LDL cholesterol (≥4.14 mmol/L) 3440 (20.1) 1.3 (1.1, 1.4) 2.0 (1.8, 2.2) 0.95
Alanine aminotransferase (>50 U/L ) 4563 (26.7) 2.0 (1.8, 2.3) 4.0 (3.7, 4.3) 0.25
White blood cell count (>8.5 109/L) 1220 (7.1) 1.7 (1.4, 2.0) 1.2 (1.1, 1.4) 0.94
Cardiometabolic marker‡ 9237 (54.0) 1.8 (1.6, 2.1) 3.4 (3.2, 3.7) 0.13
Women
Systolic blood pressure (≥140 mmHg) 2690 (15.6) 1.8 (1.5, 2.1) 2.8 (2.5, 3.0) 0.07
LDL cholesterol (≥4.14 mmol/L) 4053 (23.6) 1.3 (1.1, 1.5) 1.7 (1.6, 1.9) 0.94
Alanine aminotransferase (>30 U/L ) 1365 (7.9) 2.7 (2.2, 3.2) 3.5 (3.1, 3.9) 0.03
White blood cell count (>8.5 109/L) 466 (2.7) 2.1 (1.5, 3.1) 1.7 (1.3, 2.1) 0.37
Cardiometabolic marker‡ 6722 (39.1) 1.9 (1.6, 2.3) 2.9 (2.7, 3.2) 0.46

CI, confidence interval; LDL, low-density lipoprotein; OR, odds ratio.

†
Diabetes (reference: non diabetes) and obesity (reference: non-obese) was entered into same model adjusted for, age (continuous), and
history of cardiovascular disease (yes or no).
‡
Cardiometabolic marker indicates a subject with at least one abnormal level of the selected cardiometabolic markers.

diabetes. Among the non-obese subjects with diabetes,

Discussion
This study confirmed that more than half of individuals
with diabetes in Japan are not obese (BMI < 25) at the
current diagnostic criteria using both fasting plasma
glucose and hemoglobin A1c, in addition to known
more than 61% of men and 65% of women had at least
one abnormal level of a cardiometabolic marker such
as for systolic blood pressure, low-density lipoprotein
cholesterol, alanine aminotransferase, and white blood
cell count. These proportions were higher than among

528 © 2014 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and Wiley Publishing Asia Pty Ltd
S. KASHIMA et al.
non-obese subjects without diabetes. In addition, the
presence of diabetes contributed to an abnormal level
of cardiometabolic marker, even in the non-obese
subjects.
This study has some features worth mentioning.
First, a direct observation of this figure had been
lacking in previous studies. For example, we calculated
that among persons who are strongly suspected as
having diabetes, 57% (272/477) were non-obese
(BMI < 25) from the previous survey,18 which was
similar to the findings of this study. The previous
survey, however, did not indicate the prevalence of
non-obese diabetes as an original finding. Second, this
study examined a large population of a single Asian
ethnicity (Japanese) using the same laboratory tests.
This is favorable in preventing biases due to multiple
ethnicities and a mixture of multiple assays.
Except for alanine aminotransferase in women, there
was no significant interaction between diabetes and
obesity in the association with abnormal cardiometa-
bolic risk markers. This finding suggests that individuals
with diabetes may have a higher risk of abnormal levels
of cardiometabolic markers compared with non-diabetes
subjects, regardless of their BMI levels. Alanine amino-
transferase is a known marker for liver fatty changes that
are one manifestation of visceral obesity. Although only
observed in women, the increased odds of elevated
alanine aminotransferase in non-obese diabetes may
indicate the significance of hidden visceral obesity in dia-
betes without overall obesity.
There are some limitations in the present study. First,
since the study subjects voluntarily participated in the
health checkup, they might be healthier than the general
population, which was causing a selection bias such as
more favorable values of cardiometabolic risks. This
may underestimate the real prevalence of diabetes.
However, the prevalence of diabetes is similar to a 2004
report from the Ministry of Health, Labour and Welfare
of Japan (prevalence of strongly suspected diabetes:
12.8% for men and 6.5% for women).18 Second, we
defined obesity classification by BMI according to the
classification by the Japan Society for the Study of
Obesity25,27 and recommendations for BMI for an Asian
population by the World Health Organization.28 BMI
does not directly measure body fat, and other methods of
estimating body fat and body fat distribution such as
measurements of skinfold thickness and waist circumfer-
ence, and waist-to-hip circumference ratios might be
required to define a classification for obesity. BMI,
however, strongly correlates with fat percentage33 and
one of the acceptable indicators for population assess-
ment of overweight and obesity.34 Thus, our classifica-
tion of obesity may be adequate for this study analysis.
© 2014 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and Wiley Publishing Asia Pty Ltd
Characteristics of non-obese diabetes
Third, data on smoking status and alcohol intake was
not available from this dataset. These two factors may
have an impact on the relation between BMI and
cardiometabolic markers such as blood pressure and
alanine aminotransferase. Finally, data on concomitant
medications that could affect cardiovascular and
metabolic markers was not available. Although the
disease history was adjusted for treating this issue,
residual confounders due to medications might
affect the findings of this study. However, persons
who received such medications might be more pro-
nounced in the obesity group. Indeed, percentage
of disease history was larger in the obese group
with diabetes than those in non-obesity with the
non-diabetes group (Tables 1 and 2).
In summary, over 60% of diabetic subjects were not
obese in this Japanese population, and non-obese diabe-
tes should be considered as a public health issue. The
elevated levels of metabolic markers may, at least partly,
account for the increased risk of cardiovascular disease
in non-obese diabetes.
Acknowledgement
No funding received.
Disclosure
None declared.
References
1. Chou P, Li CL, Wu GS, Tsai ST. Progression to type 2
diabetes among high-risk groups in Kin-Chen, Kinmen.
Exploring the natural history of type 2 diabetes. Diabetes
Care. 1998; 21: 1183–7.
2. Edelstein SL, Knowler WC, Bain RP et al. Predictors of
progression from impaired glucose tolerance to NIDDM:
An analysis of six prospective studies. Diabetes. 1997; 46:
701–10.
3. Boffetta P, McLerran D, Chen Y et al. Body mass index
and diabetes in Asia: A cross-sectional pooled analysis of
900,000 individuals in the Asia cohort consortium. PLoS
ONE. 2011; 6: e19930.
4. Kodama S, Horikawa C, Fujihara K et al. Comparisons
of the strength of associations with future type 2 diabetes
risk among anthropometric obesity indicators, including
waist-to-height ratio: A meta-analysis. Am J Epidemiol.
2012; 176: 959–69.
5. DeFronzo RA, Bonadonna RC, Ferrannini E. Pathogen-
esis of NIDDM. A balanced overview. Diabetes Care.
1992; 15: 318–68.
529
Characteristics of non-obese diabetes
6. Arner P. Free fatty acids – Do they play a central role in
type 2 diabetes? Diabetes Obes Metab. 2001; 3 (Suppl 1):
11–9.
7. Hamman RF, Wing RR, Edelstein SL et al. Effect of
weight loss with lifestyle intervention on risk of diabetes.
Diabetes Care. 2006; 29: 2102–7.
8. Vaag A, Lund SS. Non-obese patients with type 2 diabe-
tes and prediabetic subjects: Distinct phenotypes requir-
ing special diabetes treatment and (or) prevention? Appl
Physiol Nutr Metab. 2007; 32: 912–20.
9. Carnethon MR, De Chavez PJ, Biggs ML et al. Associa-
tion of weight status with mortality in adults with inci-
dent diabetes. JAMA. 2012; 308: 581–90.
10. Holman RR. Long-term efficacy of sulfonylureas: A
United Kingdom Prospective Diabetes Study perspective.
Metabolism. 2006; 55: S2–5.
11. UK Prospective Diabetes Study Group. UK prospective
diabetes study 16. Overview of 6 years’ therapy of type II
diabetes: A progressive disease. UK Prospective Diabetes
Study Group. Diabetes. 1995; 44: 1249–58.
12. Vaag A. On the pathophysiology of late onset non-insulin
dependent diabetes mellitus. Current controversies and
new insights. Danish Med Bull. 1999; 46: 197–234.
13. Racette SB, Weiss EP, Hickner RC, Holloszy JO. Modest
weight loss improves insulin action in obese African
Americans. Metabolism. 2005; 54: 960–5.
14. Manson JE, Colditz GA, Stampfer MJ et al. A prospec-
tive study of maturity-onset diabetes mellitus and risk of
coronary heart disease and stroke in women. Arch Intern
Med. 1991; 151: 1141–7.
15. Yoon KH, Lee JH, Kim JW et al. Epidemic obesity and
type 2 diabetes in Asia. Lancet. 2006; 368: 1681–8.
16. Nakagami T, Qiao Q, Carstensen B et al. Age, body mass
index and Type 2 diabetes-associations modified by eth-
nicity. Diabetologia. 2003; 46: 1063–70.
17. Ma RC, Chan JC. Type 2 diabetes in East Asians:
Similarities and differences with populations in Europe
and the United States. Ann N Y Acad Sci. 2013; 1281:
64–91.
18. Ministry of Health, Labour and Welfare, Japan. [Report
of Survey on the Current Status of Diabetes in 2002].
Ministry of Health, Labour and Welfare, Tokyo, 2004.
Available from: http://www.mhlw.go.jp/toukei/kouhyo/
data-kou4/data14/to01.pdf (accessed 2 September 2013).
19. Ministry of Health, Labour and Welfare, Japan. [The
National Health and Nutrition Survey in Japan, 2010].
Ministry of Health, Labour and Welfare, Tokyo, 2012.
Available from: http://www.mhlw.go.jp/bunya/kenkou/
eiyou/dl/h22-houkoku-01.pdf (accessed 19 August 2014).
20. Inoue K, Matsumoto M, Akimoto K. The threshold for
definition of impaired fasting glucose in a Japanese popu-
lation. Diabet Med. 2009; 26: 1175–8.
21. Inoue K, Inoue M, Matsumoto M, Akimoto K. Persis-
tent fasting hyperglycaemia more predictive of Type 2
530 © 2014 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and Wiley Publishing Asia Pty Ltd
S. KASHIMA et al.
diabetes than is transient hyperglycaemia. Diabet Med.
2012; 29: 1327–34.
22. Committee of the Japan Diabetes Society on the
Diagnostic Criteria of Diabetes Mellitus. Seino Y, Nanjo
K, Tajima N et al. Report of the committee on the clas-
sification and diagnostic criteria of diabetes mellitus.
J Diabetes Invest. 2010; 1: 212–28.
23. International Expert Committee. International Expert
Committee report on the role of the A1C assay in the
diagnosis of diabetes. Diabetes Care. 2009; 32: 1327–34.
24. Matsuzawa Y, Tokunaga K, Kotani K, Keno Y,
Kobayashi T, Tarui S. Simple estimation of ideal body
weight from body mass index with the lowest morbidity.
Diab Res Clin Pract. 1990; 10 (Suppl 1): S159–64.
25. Obesity Diagnosis Guideline Preparation Committee of
the Japan Society for the Study of Obesity, Saito Y,
Shirai K, Nakamura T et al. [New criteria for “obesity
disease” for Japanese in 2011] (Japanese). Journal of
Japan Society for the Study of Obesity. 2011; 17 (Extra
edition): 1–78.
26. Kanazawa M, Yoshiike N, Osaka T, Numba Y, Zimmet
P, Inoue S. Criteria and classification of obesity in Japan
and Asia-Oceania. World Review of Nutrition and Dietet-
ics. 2005; 94: 1–12.
27. Kanazawa M, Yoshiike N, Osaka T, Numba Y, Zimmet
P, Inoue S. Criteria and classification of obesity in Japan
and Asia-Oceania. Asia Pac J Clin Nutr. 2002; 11 (Suppl
8): S732–7.
28. WHO Expert Consultation. Appropriate body-mass
index for Asian populations and its implications for
policy and intervention strategies. Lancet. 2004; 363: 157–
63.
29. Friedewald WT, Levy RI, Fredrickson DS. Estimation of
the concentration of low-density lipoprotein cholesterol
in plasma, without use of the preparative ultracentrifuge.
Clin Chem. 1972; 18: 499–502.
30. Reaven GM. Banting lecture 1988. Role of insulin resis-
tance in human disease. Diabetes. 1988; 37: 1595–607.
31. Liese AD, Mayer-Davis EJ, Haffner SM. Development
of the multiple metabolic syndrome: An epidemiologic
perspective. Epidemiol Rev. 1998; 20: 157–72.
32. Kannel WB, Gordon T, Schwartz MJ. Systolic versus
diastolic blood pressure and risk of coronary heart
disease. The Framingham study. Am J Cardiol. 1971; 27:
335–46.
33. Ranasinghe C, Gamage P, Katulanda P, Andraweera N,
Thilakarathne S, Tharanga P. Relationship between
Body Mass Index (BMI) and body fat percentage, esti-
mated by bioelectrical impedance, in a group of Sri
Lankan adults: A cross sectional study. BMC Public
Health. 2013; 13: 797.
34. WHO. Obesity: Preventing and managing the global epi-
demic – Introduction. Who Tech Rep Ser. 2000; 894:
1–253.