- Email: [email protected]
Prediabetes and cardiovascular disease risk: A nested case-control study
Accepted Manuscript Prediabetes and cardiovascular disease risk: A nested case-control study Huanhuan Hu, Tetsuya Mizoue, Naoko Sasaki, Takayuki Ogasawara, Kentaro Tomita, Satsue Nagahama, Ai Hori, Akiko Nishihara, Teppei Imai, Makoto Yamamoto, Masafumi Eguchi, Takeshi Kochi, Toshiaki Miyamoto, Toru Honda, Tohru Nakagawa, Shuichiro Yamamoto, Hiroko Okazaki, Akihiko Uehara, Makiko Shimizu, Taizo Murakami, Keisuke Kuwahara, Akiko Nanri, Maki Konishi, Isamu Kabe, Seitaro Dohi
PII:
S0021-9150(18)31365-0
DOI:
10.1016/j.atherosclerosis.2018.09.004
Reference:
ATH 15702
To appear in:
Atherosclerosis
Received Date: 7 June 2018 Revised Date:
15 August 2018
Accepted Date: 7 September 2018
Please cite this article as: Hu H, Mizoue T, Sasaki N, Ogasawara T, Tomita K, Nagahama S, Hori A, Nishihara A, Imai T, Yamamoto M, Eguchi M, Kochi T, Miyamoto T, Honda T, Nakagawa T, Yamamoto S, Okazaki H, Uehara A, Shimizu M, Murakami T, Kuwahara K, Nanri A, Konishi M, Kabe I, Dohi S, for the Japan Epidemiology Collaboration on Occupational Health Study Group, Prediabetes and cardiovascular disease risk: A nested case-control study, Atherosclerosis (2018), doi: 10.1016/ j.atherosclerosis.2018.09.004. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Prediabetes and cardiovascular disease risk: A nested case-control study
2
Huanhuan Hu a,*, Tetsuya Mizouea, Naoko Sasakib, Takayuki Ogasawarab, Kentaro
3
Tomitac, Satsue Nagahamad, Ai Horie, Akiko Nishiharaf, Teppei Imaif, Makoto
4
Yamamotog, Masafumi Eguchih, Takeshi Kochih, Toshiaki Miyamotoi, Toru Hondaj,
5
Tohru Nakagawaj, Shuichiro Yamamotoj, Hiroko Okazakik, Akihiko Ueharal, Makiko
6
Shimizum, Taizo Murakamim, Keisuke Kuwaharaa,n, Akiko Nanria,o, Maki Konishia,
7
Isamu Kabeh, Seitaro Dohik, for the Japan Epidemiology Collaboration on Occupational
8
Health Study Group
9
a Department of Epidemiology and Prevention, National Center for Global Health and
M AN U
SC
RI PT
1
Medicine, Tokyo, Japan
11
b Mitsubishi Fuso Truck and Bus Corporation, Kanagawa, Japan
12
c Mitsubishi Plastics, Inc., Tokyo, Japan
13
d All Japan Labour Welfare Foundation, Tokyo, Japan
14
e Department of Global Public Health, University of Tsukuba, Ibaraki, Japan
15
f Azbil Corporation, Tokyo, Japan
16
g YAMAHA CORPORATION, Shizuoka, Japan
17
h Furukawa Electric Co., Ltd., Tokyo, Japan
18
i Nippon Steel & Sumitomo Metal Corporation Kimitsu Works, Chiba, Japan
AC C
EP
TE D
10
1
ACCEPTED MANUSCRIPT
j Hitachi, Ltd., Ibaraki, Japan
20
k Mitsui Chemicals, Inc., Tokyo, Japan
21
l Seijinkai Shizunai Hospital, Hokkaidō, Japan
22
m Mizue Medical Clinic, Keihin Occupational Health Center, Kanagawa, Japan
23
n Teikyo University Graduate School of Public Health, Tokyo, Japan
24
o Department of Food and Health Sciences, International College of Arts and Sciences,
25
Fukuoka Women’s University, Fukuoka, Japan
26
*Address correspondence to Huanhuan Hu, Department of Epidemiology and
27
Prevention, National Center for Global Health and Medicine, Toyama 1-21-1,
28
Shinjuku-ku, Tokyo 162-8655, Japan.
29
Phone: +81-3-3202-7181, Fax: +81-3-3202-7364
30
E-mail: [email protected]
EP
TE D
M AN U
SC
RI PT
19
AC C
31 32
Abstract
33
Background and aims: We aimed to examine the risk of cardiovascular disease (CVD)
34
with persistent prediabetes during the last four years prior to a CVD event in a large
35
occupational cohort in Japan.
36
Methods: We performed a nested case-control study using data from the Japan
2
ACCEPTED MANUSCRIPT
Epidemiology Collaboration on Occupational Health Study. A total of 197 registered
38
cases of CVD were identified and matched individually with 985 controls according to
39
age, sex, and worksite. Prediabetes was defined as fasting plasma glucose 100-125
40
mg/dL and/or HbA1c 5.7-6.4%. Persistent prediabetes was defined as having
41
prediabetes at years one and four prior to the onset/index date; persistent
42
normoglycemia was similarly defined. Associations between prediabetes and CVD risk
43
were assessed using conditional logistic regression models.
44
Results: Compared with people with persistent normoglycemia over the four years prior
45
to the onset/index date, the unadjusted odds ratio (95% confidence interval) for CVD
46
was 2.88 (1.56, 5.32) for people with persistent prediabetes. After adjusting for BMI,
47
smoking, hypertension, and dyslipidemia assessed four years before the onset/index date,
48
the association was slightly attenuated to an OR (95% confidence interval) of 2.62 (1.31,
49
5.25). Prediabetes assessed at single time points was also associated with an elevated
50
risk of CVD, with multivariable-adjusted odds ratio (95% confidence interval) of 1.72
51
(1.12, 2.64) and 2.13 (1.32, 3.43) for prediabetes at one and four years prior to the
52
onset/index date, respectively.
53
Conclusions: Prediabetes is associated with an increased risk of CVD. Identification
54
and management of prediabetes are important for the prevention of CVD.
AC C
EP
TE D
M AN U
SC
RI PT
37
3
ACCEPTED MANUSCRIPT
Keywords
56
Prediabetes; Diabetes; Cardiovascular disease; Nested case-control study
57
Introduction
58
Prediabetes represents a high-risk state for type 2 diabetes and affects various
59
populations around the world [1]. Globally, 352 million adults are living with
60
prediabetes (impaired glucose tolerance), and this number is projected to increase to 587
61
million in 2040 [2]. Japan is one of the top ten countries with the highest number of
62
adults with prediabetes [2]. Meta-analyses of cohort studies, in which only baseline
63
glycemia levels were measured, show that prediabetes is associated with an increased
64
risk of cardiovascular disease (CVD) [3,4]. However, given that 5% -10% of individuals
65
with prediabetes develop diabetes annually [1], it remains unclear whether this elevated
66
risk of CVD is due to prediabetes or the transition from prediabetes to diabetes during
67
follow-up.
AC C
EP
TE D
M AN U
SC
RI PT
55
To the best of our knowledge, only two prospective studies have examined the
68
69
prediabetes-CVD association while excluding people who developed diabetes during
70
follow-up [5,6]. A U.K. study reported an increased risk of CVD associated with
71
prediabetes in Europeans but not South Asians [5], and a Finnish study showed
72
increased morbidity and mortality of CVD associated with prediabetes [6]. One
4
ACCEPTED MANUSCRIPT
important limitation of these studies is that people who had prediabetes at baseline but
74
returned to normoglycemia during follow-up were not identified and therefore were not
75
excluded from the analyses, leading to exposure misclassification. To better assess the
76
association between prediabetes and CVD, studies using repeated measurements of
77
blood glucose are needed.
SC
RI PT
73
Using data from the Japan Epidemiology Collaboration of Occupational Health
79
(J-ECOH) Study, in which annual health checkup data were available throughout the
80
follow-up period [7,8], we performed a nested case-control study to examine the risk of
81
CVD posed by persistent prediabetes during the four years immediately prior to a CVD
82
event in comparison to cases of persistent normoglycemia. We also investigated the
83
risks of CVD associated with prediabetes one year or four years before the CVD event,
84
respectively.
85
Materials and methods
86
Study design
87
The J-ECOH Study is an ongoing multicenter, health checkup-based cohort study
88
among workers from several companies in Japan, and has been described elsewhere
89
[7,8]. In Japan, workers are obliged to undergo a health checkup at least once a year
90
under the Industrial Safety and Health Act; nearly all workers attend their health
AC C
EP
TE D
M AN U
78
5
ACCEPTED MANUSCRIPT
checkup each year. Briefly, participants in the J-ECOH Study underwent routine
92
physical and laboratory examinations (blood glucose, blood lipids, etc.) each year.
93
Additionally, a questionnaire that covered medical history, health-related lifestyle, and
94
work environment was completed. Annual health checkup data between January 2008
95
and December 2016 or between April 2008 and March 2017 were collected from over
96
100,000 employees in 11 participating companies. The study protocol, including the
97
consent procedure, was approved by the Ethics Committee of the National Center for
98
Global Health and Medicine, Japan. Using the J-ECOH Study data, we performed a
99
nested case-control study to investigate the association between prediabetes and incident
M AN U
SC
RI PT
91
CVD.
101
Health checkup
102
Body height and weight were measured using a scale with participants wore light
103
clothes and no shoes. Body mass index (BMI) was calculated as weight in kilograms
104
divided by the square of height in meters. Blood pressure was measured using an
105
automatic blood pressure monitor. Hypertension was defined as either systolic blood
106
pressure of at least 140 mmHg, diastolic blood pressure of at least 90 mmHg, or use of
107
treatment for hypertension [9]. Blood glucose was measured using either the enzymatic
108
or glucose oxidase peroxidative electrode method based on each company’s protocol.
AC C
EP
TE D
100
6
ACCEPTED MANUSCRIPT
HbA1c was measured using either a latex agglutination immunoassay,
110
high-performance liquid chromatography, or the enzymatic method. Triglyceride,
111
low-density lipoprotein-cholesterol, and high-density lipoprotein-cholesterol levels were
112
measured using the enzymatic method. Dyslipidemia was defined as either a
113
low-density lipoprotein-cholesterol level of at least 140 mg/dL, high-density
114
lipoprotein-cholesterol level of less than 40 mg/dL, triglyceride level of at least 150
115
mg/dL, or use of medications for dyslipidemia [10]. All of the laboratories involved in
116
the health checkups of the participating companies received satisfactory scores (rank A
117
or a score >95 out of 100) from external quality-control agencies.
118
Ascertainment of CVD cases and control selection
119
CVD cases were identified through a disease registry, which was set up in April 2012
120
within the J-ECOH study to collect data on CVD events from the occupational health
121
physicians of the participating companies [8]. For most nonfatal cases, occupational
122
physicians confirmed the diagnosis of each CVD event based on medical certificates,
123
which were written by a treating physician and submitted to the company through the
124
worker. Because the submission of a medical certificate is required when taking
125
long-term sick leave, this registry primarily covers relatively severe cases. For fatal
126
cases, occupational physicians judged the cause of death based on available information,
AC C
EP
TE D
M AN U
SC
RI PT
109
7
ACCEPTED MANUSCRIPT
including death certificates and information obtained from the bereaved family or
128
colleagues. In the CVD registry, approximately 80% cases were confirmed through
129
medical certificates, 10% were self-reported or from family members or colleagues, and
130
10% were from other sources. Each case was coded according to the 10th revision of the
131
International Classification of Diseases.
SC
RI PT
127
In the present study, cases were patients with a first event of myocardial infarction
133
or stroke between April 2012 and March 2017. A total of 235 incident CVD cases were
134
identified. After removing patients (n=38) who did not attend health checkup one year
135
before the onset date, 197 CVD cases (137 strokes and 60 myocardial infarctions; 36
136
fatal cases) remained in the present study (Figure 1).
TE D
M AN U
132
Controls were selected from study participants who did not self-report stroke or
138
heart disease at J-ECOH Study entry and did not develop CVD during the follow-up
139
period. Those who self-reported a history of CVD at annual health checkups during the
140
study period were also excluded. For each case, we created a pool of controls who were
141
matched by worksite, sex, and date of birth (±2 years). Controls were assigned an index
142
date that corresponded to the onset date of their matched case. We excluded people who
143
did not attend health checkup at one year prior to the index date. Finally, for a given
144
case, we randomly selected up to five controls from the pool of eligible controls. Once a
AC C
EP
137
8
ACCEPTED MANUSCRIPT
145
control was sampled, he/she was not again chosen as a control for other cases. A total of
146
985 matched controls were included in the present study. Using data from the 197 cases and 985 matched controls, we examined the risk of
RI PT
147
CVD associated with prediabetes at one year prior to the CVD event. After removing
149
people who did not attend a health checkup four years before the onset/index date, we
150
investigated the risk of CVD associated with having prediabetes four years before the
151
CVD event in 152 cases and 675 controls. We further excluded people who had
152
prediabetes at four years before but not one year before the onset/index date and those
153
who had normoglycemia at four years before but not one year before the onset/index
154
date, leaving 115 cases and 381 controls for examining the risk of CVD with persistent
155
prediabetes over the last four years before the CVD event.
156
Exposures
157
We retrieved data on diabetes treatment status, fasting plasma glucose (FPG), and
158
HbA1c collected at one and four years before the onset/index date. Diabetes was
159
defined as either FPG ≥126 mg/dL, HbA1c ≥6.5%, or receiving medical treatment for
160
diabetes according to the American Diabetes Association criteria [11]. Prediabetes was
161
defined as FPG 100-125 mg/dL and/or HbA1c 5.7-6.4% [11]. Normoglycemia was
162
defined as FPG <100 mg/dL and HbA1c <5.7%. Persistent prediabetes was defined for
AC C
EP
TE D
M AN U
SC
148
9
ACCEPTED MANUSCRIPT
those who had prediabetes at both one and four years prior to the onset/index date.
164
Persistent normoglycemia was defined for those who had normoglycemia at both one
165
and four years prior to the onset/index date. We also defined prediabetes using either
166
FPG (100-125 mg/dL) or HbA1c (5.7-6.4%) to examine whether the risk of CVD
167
associated with prediabetes differs between the two definitions.
168
Statistical analysis
169
Characteristics of participants were expressed as percentages and means for categorical
170
and continuous variables, respectively. Chi-square tests for categorical variables and
171
t-tests for continuous variables were used to examine differences in characteristics
172
between cases and controls.
TE D
M AN U
SC
RI PT
163
Conditional logistic regression was used to estimate odds ratios (ORs) and 95%
173
confidence intervals (CIs) for the development of CVD associated with persistent
175
prediabetes. In model 1, the analysis was conditioned on the matching variables (sex
176
and age). In model 2, we adjusted for a priori-specified potential confounders, including
177
BMI, smoking (current smoker, non-current smoker), hypertension, and dyslipidemia
178
assessed at four years before the onset/index date. We also examined the associations
179
between CVD risk and prediabetes assessed at different time points: one and four years
180
before the onset/index date, respectively.
AC C
EP
174
10
ACCEPTED MANUSCRIPT
The present study had 73% power to detect a relative risk of 2.0 with statistical
181
significance (two-sided alpha level of 0.05), based on the given condition of the
183
proportion of people with persistent prediabetes among controls (50%), and the number
184
of cases (47 persistent prediabetes and 23 persistent normoglycemia) and number of
185
matched controls (350). All statistical analyses were performed using SAS version 9.4
186
(SAS Institute, Cary, NC, USA). A two-sided p < 0.05 was considered statistically
187
significant.
188
Results
189
Table 1 shows the characteristics of cases and controls at one year prior to the
190
onset/index date. Cases had higher means for BMI, triglyceride, low-density lipoprotein
191
cholesterol, blood pressure, FPG, and HbA1c than controls. The prevalence rates of
192
smoking, hypertension, dyslipidemia, and diabetes were higher among cases compared
193
with controls. There was no material difference in the proportion of patients who were
194
receiving medication for hypertension or dyslipidemia between cases and controls.
195
Table 2 shows the association between prediabetes and CVD. Compared with
AC C
EP
TE D
M AN U
SC
RI PT
182
196
people with persistent normoglycemia over the past four years before the onset/index
197
date, the unadjusted OR (95% CI) for CVD was 2.88 (1.56, 5.32) for people with
198
persistent prediabetes. After adjusting for BMI, smoking, hypertension, and
11
ACCEPTED MANUSCRIPT
dyslipidemia, the association was slightly attenuated to an OR (95% CI) of 2.62 (1.31,
200
5.25). For people with prediabetes one year before, the OR (95% CI) was 1.72 (1.12,
201
2.64) compared with those with normoglycemia, after adjusting for BMI, smoking,
202
hypertension, and dyslipidemia assessed one year before the onset/index date. For
203
people with prediabetes four years before, the OR (95% CI) was 2.13 (1.32, 3.43), with
204
adjustment for BMI, smoking, hypertension, and dyslipidemia assessed four years
205
before the onset/index date.
M AN U
SC
RI PT
199
Table 3 shows the risk of CVD associated with prediabetes as defined using either
206
FPG or HbA1c. When prediabetes was defined as FPG 100-125 mg/dL, people with
208
prediabetes tended to have a higher risk of CVD compared with those without
209
prediabetes. Multivariable-adjusted ORs (95% CI) were 1.76 (0.91, 3.38), 1.33 (0.88,
210
2.01), and 1.77 (1.10, 2.86) for people with persistent prediabetes, prediabetes one year
211
before, and prediabetes four years before, respectively. When prediabetes was defined
212
as HbA1c 5.7-6.4%, the corresponding ORs (95% CI) were 2.07 (1.09, 3.97) 1.28 (0.82,
213
2.02), and 1.93 (1.21, 3.08), respectively.
214
Discussion
215
In this nested case-control study, we found that prediabetes (defined as FPG 100-125
216
mg/dL and/or HbA1c 5.7-6.4%) one or four years prior to the onset/index date was
AC C
EP
TE D
207
12
ACCEPTED MANUSCRIPT
associated with significantly increased risk of CVD. More importantly, a much higher
218
risk for CVD was observed among people with persistent prediabetes. To our
219
knowledge, this is one of the few studies to examine whether prediabetes is associated
220
with the development of CVD.
RI PT
217
The present findings are in line with those of the existing yet limited studies [5,6].
SC
221
We found that prediabetes one and four years prior to the onset/index date was
223
associated with a 1.7- and 2.1-fold increased risk of CVD, respectively. However, when
224
the analyses were limited to those who had prediabetes, both one and four years before
225
the onset/index date, people with persistent prediabetes were 2.6 times more likely to
226
develop CVD. This difference in the strength of associations implies that the association
227
between prediabetes and CVD risk may not be accurately captured if prediabetes was
228
assessed at a single time point due to changes in glycemia levels over time. In our study,
229
out of 291 controls with prediabetes four years before the index date, 22 developed
230
diabetes within three years, and 68 returned to normoglycemia. Of 303 controls with
231
normoglycemia four years prior, 92 developed prediabetes three years later. Using
232
multiple measures of glycemia to minimize the misclassification of exposure, our data
233
confirm that prediabetes is a risk factor for CVD. Proactive identification and
234
management of prediabetes are warranted to reduce future risk of CVD.
AC C
EP
TE D
M AN U
222
13
ACCEPTED MANUSCRIPT
We observed a more pronounced association with HbA1c-based prediabetes than
235
FPG-based prediabetes, consistent with findings from the Whitehall II Study [12]. A
237
meta-analysis showed no significant difference in the risk of CVD between baseline
238
prediabetes defined using FPG (100-125 mg/dL) and that defined using HbA1c
239
(5.7-6.4%) [3]. Of 16 studies included in that meta-analysis, only three made a
240
head-to-head comparison between the two definitions; two reported a significant
241
association with HbA1c-based prediabetes but not with FPG-based prediabetes, and
242
another did not detect any association with these definitions [3]. Taken together,
243
prediabetes defined by HbA1c appears to be more strongly associated with CVD than
244
prediabetes defined by FPG.
TE D
M AN U
SC
RI PT
236
Non-glucose-mediated and glucose-mediated mechanisms have been proposed as
245
explanations for the association between prediabetes and CVD. One of the
247
non-glucose-mediated mechanisms involves insulin resistance, which is an underlying
248
condition of prediabetes [13]. It promotes atherogenesis and atherosclerotic plaque
249
formation via down-regulation of insulin signaling at the level of the intimal cells that
250
participate in atherosclerosis, including macrophages, vascular smooth muscle cells, and
251
endothelial cells [14,15]. The other non-glucose-mediated mechanisms involve obesity
252
and inflammation (adipocytokines and oxidative stress), dyslipidemia (atherogenic
AC C
EP
246
14
ACCEPTED MANUSCRIPT
lipoproteins and increased circulating free fatty acids), and prothrombosis state
254
(increased levels of PAI-1 and platelet activation), which are commonly present in
255
people with prediabetes [14,15]. As for glucose-mediated mechanisms, hyperglycemia
256
could promote atherosclerosis through increased oxidative stress, activation of
257
thrombosis, and epigenetic changes (increased expression of inflammatory genes via
258
NF-κB), which can be induced even by transient or episodic hyperglycemia [14,15].
259
Strengths and limitations
M AN U
SC
RI PT
253
A major advantage of the present study was the periodic assessment of diabetic
260
status using HbA1c and FPG until one year before the onset of CVD, which enabled us
262
to maximally reduce the degree of misclassification of prediabetes. In addition, potential
263
confounders such as smoking, dyslipidemia, and hypertension were also assessed
264
annually. Nevertheless, this study has some limitations. First, the present study had a
265
power less than 80% to detect an association with a relative risk of 2.0. Using repeated
266
assessment of diabetic status, however, we found a stronger association (relative risk of
267
2.5 or more) between prediabetes and CVD risk, increasing the power of our study to
268
90%. Second, we only have recent data on diabetes status (within the past four years
269
prior to the onset of CVD). The lack of data for earlier periods precludes us from
270
analyzing the effect of long-term exposure to prediabetes on CVD development. Third,
AC C
EP
TE D
261
15
ACCEPTED MANUSCRIPT
only workers who took long-term sick leave were required to submit a medical
272
certificate to the company; thus, milder forms of CVD were not well-covered by this
273
registration system. Fourth, we defined prediabetes using FPG and HbA1c but not
274
2-hour oral glucose tolerance test, which has not been used in regular health checkups.
275
As such, people with impaired glucose tolerance may be misclassified as
276
normoglycemia, resulting in an underestimation of the association between prediabetes
277
and CVD. Lastly, due to lack of data on socioeconomic status, family history of CVD,
278
and lifestyle characteristics other than smoking (e.g., diet, physical activity), we were
279
unable to control for the potential effects of these factors.
M AN U
SC
RI PT
271
In conclusion, the present nested case-control study based on repeated
TE D
280
measurements of fasting glucose and HbA1c indicates that prediabetes is associated
282
with an increased risk of CVD. Identification and management of prediabetes are
283
important for the prevention of CVD. More research is required to confirm or refute
284
whether pre-diabetic patients who are free of conventional risk factors such as
285
hypertension and hypercholesterolemia are at increased risk of CVD.
AC C
EP
281
286
287
Conflict of interest
288
The authors declared they do not have anything to disclose regarding conflict of interest
16
ACCEPTED MANUSCRIPT
with respect to this manuscript.
290
Financial support
291
This study was supported by the Industrial Health Foundation, Industrial Disease
292
Clinical Research Grants (grant numbers 140202-01, 150903-01, 170301-01), Japan
293
Society for the Promotion of Science KAKENHI (grant number 16H05251), and Grant
294
of National Center for Global Health and Medicine (grant number 28-Shi-1206).
295
Author contributions
296
S.D. and T.Mizoue were involved in the design of the study as the principal
297
investigators; N.S., T.O., K.T., S.N., A.H., A.Nishihara, T.I., M.Y., M.E., T.K.,
298
T.Miyamoto, T.H., T.N., S.Y., H.O., A.U., M.S., T.Murakami, I.Kabe., and S.D.
299
collected health check-up data; M.K. and A.Nanri cleaned CVD data; T.Mizoue, M.K.,
300
and K.K. created database; H.H. and T.Mizoue drafted the plan for the data analysis;
301
H.H. conducted data analysis; H.H. drafted manuscript; HH and T Mizoue had primary
302
responsibility for final content; and all authors were involved in interpretation of the
303
results and revision of the manuscript and approved the final version of the manuscripts,
304
had full access to all of the data in the study, and take responsibility for the integrity of
305
the data and the accuracy of the data analysis. HH and T Mizoue are guarantors.
AC C
EP
TE D
M AN U
SC
RI PT
289
306
17
ACCEPTED MANUSCRIPT
Acknowledgements
308
We thank Dr. Toshiteru Okubo (Chairperson of Industrial Health Foundation) for
309
scientific advice on the conduct of J-ECOH Study; Eri Yamada (National Center for
310
Global Health and Medicine) for data management; and Rika Osawa (National Center
311
for Global Health and Medicine) for administrative support.
312
References
313
1 Tabák AG, Herder C, Rathmann W, et al. Prediabetes: a high-risk state for diabetes
314
development. Lancet 2012; 379:2279-2290.
315
2 International Diabetes Federation. IDF Diabetes Atlas. 8th ed. Available from
316
www.diabetesatlas.org. Accessed December 2017
317
3 Huang Y, Cai X, Mai W, et al. Association between prediabetes and risk of
318
cardiovascular disease and all-cause mortality: systematic review and meta-analysis.
319
BMJ 2016;355: i5953.
320
4 Ford ES, Zhao G, Li C. Pre-diabetes and the risk for cardiovascular disease: a
321
systematic review of the evidence. J Am Coll Cardiol 2010; 55:1310-1317.
322
5 Eastwood SV, Tillin T, Sattar N, et al. Associations between prediabetes, by three
323
different diagnostic criteria, and incident CVD differ in South Asians and Europeans.
324
Diabetes Care 2015; 38:2325-2332.
AC C
EP
TE D
M AN U
SC
RI PT
307
18
ACCEPTED MANUSCRIPT
6 Qiao Q, Jousilahti P, Eriksson J, et al. Predictive properties of impaired glucose
326
tolerance for cardiovascular risk are not explained by the development of overt diabetes
327
during follow-up. Diabetes Care 2003; 26:2910-2914.
328
7 Hu H, Nagahama S, Nanri A, et al. Duration and degree of weight change and risk of
329
incident diabetes: Japan Epidemiology Collaboration on Occupational Health Study.
330
Prev Med 2017; 96:118-123.
331
8 Hu H, Nakagawa T, Honda T, et al. Metabolic syndrome over four years before the
332
onset of cardiovascular disease: nested case-control Study. Circ J 2018; 82:430-436.
333
9 Shimamoto K, Ando K, Fujita T, et al. The Japanese Society of Hypertension
334
Guidelines for the Management of Hypertension (JSH 2014). Hypertens Res 2014;
335
37:253–390.
336
10 Teramoto T, Sasaki J, Ishibashi S, et al. Executive Summary of the Japan
337
Atherosclerosis Society Guidelines for the Diagnosis and Prevention of Atherosclerotic
338
Cardiovascular Diseases in Japan -2012 version. J Atherosclerosis Thromb 2013;
339
20:517–523.
340
11 American Diabetes Association. American Diabetes Association Standards of
341
medical care in diabetes–2017. Diabetes Care 2017; 40(Suppl.1): S1-S135.
342
12 Vistisen D, Witte DR, Brunner E, et al. Risk of Cardiovascular Disease and Death in
AC C
EP
TE D
M AN U
SC
RI PT
325
19
ACCEPTED MANUSCRIPT
Individuals With Prediabetes Defined by Different Criteria: The Whitehall II Study.
344
Diabetes Care 2018; 41:899-906.
345
13 Abdul-Ghani MA, DeFronzo RA. Pathophysiology of prediabetes. Curr Diab Rep
346
2009; 9:193-199.
347
14 Milman S, Crandall JP. Mechanisms of vascular complications in prediabetes. Med
348
Clin North Am 2011; 95:309-325.
349
15 Buysschaert M, Medina JL, Bergman M, et al. Prediabetes and associated disorders.
350
Endocrine 2015; 48:371-393
M AN U
351
SC
RI PT
343
TE D
352
353
EP
Table 1 Characteristics of cases and controls at one year before the onset/index date Cases
Controls
197
985
52.9±7.3
52.8±7.3
91.9
91.9
25.0±4.0
23.9±3.6a
54.8
31.8a
TG (mg/dL)
149.9±96.9
125.0±84.5a
HDL-C (mg/dL)
53.4±15.1
58.8±15.8a
N
AC C
354
Age (years) Men, %
BMI (kg/m2) Current smoker, %
20
ACCEPTED MANUSCRIPT
129.2±37.9
123.6±29.4
Dyslipidemia, %
69.0
51.5a
Lipid-lowering treatment, %b
19.9
20.9
SBP (mmHg)
134.7±17.3
124.4±15.8a
DBP (mmHg)
85.2±10.6 65.5
Anti-hypertensive treatment, %c
52.7
FPG (mg/dL)
34.9a
Diabetes, %
57.6
120.3±50.5
102.9±18.2a
6.2±1.3
5.7±0.7a
M AN U
HbA1c (%)
78.9±11.2a
SC
Hypertension, %
RI PT
LDL-C (mg/dL)
33.5
13.7a
355
a
356
BMI: body mass index; CVD, cardiovascular disease; DBP: diastolic blood pressure;
357
FPG: fasting plasma glucose; HbA1c, glycated hemoglobin; HDL-C: high-density
358
lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; SBP, systolic blood
359
pressure; TG, triglyceride.
360
b
The denominator is the total number of people with dyslipidemia.
361
c
The denominator is the total number of people with hypertension.
362
363
364
365 21
AC C
EP
TE D
p < 0.05
ACCEPTED MANUSCRIPT
366
Table 2 Associations between prediabetes and risk of cardiovascular disease Case Control
Odds ratio (95% CI)
Over the four years prior to the onset/index date
Model 2
RI PT
Model 1
23
161
1
Persistent prediabetes
47
155
2.88 (1.56, 5.32)
Diabetes
45
65
6.50 (3.31, 12.8)
Normoglycemia
47
Prediabetes
84
Diabetes
66
M AN U
One year before the onset/index date
1
2.62 (1.31, 5.25)
SC
Persistent normoglycemia
4.11 (1.81, 9.30)
418
1
1
432
1.98 (1.33, 2.96)
1.72 (1.12, 2.64)
135
5.19 (3.28, 8.22)
3.41 (2.05, 5.69)
Four years before the onset/index date
Prediabetes
303
1
1
71
291
2.40 (1.53, 3.77)
2.13 (1.32, 3.43)
45
81
5.66 (3.29, 9.74)
3.43 (1.87, 6.32)
EP
Diabetes
36
TE D
Normoglycemia
Model 1, unadjusted.
368
Model 2, adjusted for BMI, current smoker (yes or no), hypertension, and dyslipidemia
AC C
367
369 370 371 372 373 374 375 376 377 22
ACCEPTED MANUSCRIPT
378
Table 3 Associations between prediabetes (defined as FPG 100-125 mg/dL or HbA1c
379
5.7-6.4%, respectively) and risk of cardiovascular disease Control
Odds ratio (95% CI) Model 1
FPG (mg/dL)
42
274
Persistent FPG 100-125
31
133
Diabetes
45
One year before the onset/index date 68
100-125
63
Diabetes
1
2.12 (1.16, 3.89)
1.76 (0.91, 3.38)
80
4.34 (2.40, 7.83)
2.36 (1.17, 4.75)
518
1
1
332
1.64 (1.11, 2.42)
1.33 (0.88, 2.01)
4.29 (2.82, 6.54)
2.73 (1.71, 4.36)
TE D
<100
1
M AN U
Persistent FPG <100
66
Model 2
SC
Over the four years prior to the onset/index date
RI PT
Case
135
Four years before the onset/index date
100-125
AC C
Diabetes
57
EP
<100
381
1
1
49
204
2.02 (1.28, 3.18)
1.77 (1.10, 2.86)
45
80
4.55 (2.74, 7.57)
2.77 (1.56, 4.91)
HbA1c (%)
Over the four years prior to the onset/index date Persistent HbA1c <5.7
46
376
1
1
Persistent HbA1c 5.7-6.4
26
111
2.38 (1.34, 4.22)
2.07 (1.09, 3.97)
Diabetes
45
83
4.95 (2.86, 8.59)
3.27 (1.71, 6.26)
One year before the onset/index date
23
ACCEPTED MANUSCRIPT
86
613
1
1
5.7-6.4
45
237
1.48 (0.98, 2.23)
1.28 (0.82, 2.02)
Diabetes
66
135
3.82 (2.58, 5.65)
2.57 (1.65, 4.01)
Four years before the onset/index date
RI PT
<5.7
57
421
1
1
5.7-6.4
53
196
2.19 (1.43, 3.37)
1.93 (1.21, 3.08)
Diabetes
45
83
4.45 (2.75, 7.20)
2.77 (1.60, 4.78)
SC
<5.7
Model 1, unadjusted.
381
Model 2, adjusted for BMI, current smoker (yes or no), hypertension, and dyslipidemia
M AN U
380
382 383 384 385
TE D
386 387 388 389
EP
390 391 392
AC C
393 394 395 396 397 398 399 400 401 402 403 24
ACCEPTED MANUSCRIPT
404
Figure 1 Flow diagram of study population, J-ECOH Study, Japan, 2008-2016
405
AC C
EP
TE D
M AN U
SC
RI PT
406
25
ACCEPTED MANUSCRIPT
143,499 participants who had health checkup
RI PT
Excluded (n=940): History of stroke or ischemic heart disease at J-ECOH study entry 142,324 participants
Excluded (n=19,791): Did not attend any follow-up examination
SC
Excluded (n=1,120): Self-reported ischemic heart disease and stroke during follow-up
M AN U
235 registered CVD
121,413 participants without CVD during follow-up
TE D
Excluded (n=38) Did not attend the health check-up at 1 year before the event date 197 registered CVD
985 controls
AC C
EP
Excluded 45 cases and 310 controls who did not attend health check-up at 4 years before the event/index date
Excluded 24 cases and 158 controls who had prediabetes at 4 years before but not 1 year before the event/index date Excluded 13 cases and 136 controls who had normoglycemia at 4 years before but not 1 year before the event date
To examine the risk of CVD with persistent prediabetes during the last four years prior to a CVD event
To examine the risk of CVD associated with prediabetes one year before the CVD event
To examine the risk of CVD associated with prediabetes four years before the CVD event
Figure 1 Flow diagram of study population, J-ECOH Study, Japan, 2008-2016
ACCEPTED MANUSCRIPT
Prediabetes and cardiovascular disease risk: A nested case-control study
•
RI PT
Highlights Prediabetes assessed at single time points was associated with an elevated risk of CVD.
Much higher risk of CVD was observed among people with persistent prediabetes compared with persistent normoglycemia.
Prediabetes is associated with a higher risk of CVD.
AC C
EP
TE D
M AN U
•
SC
•
PII:
S0021-9150(18)31365-0
DOI:
10.1016/j.atherosclerosis.2018.09.004
Reference:
ATH 15702
To appear in:
Atherosclerosis
Received Date: 7 June 2018 Revised Date:
15 August 2018
Accepted Date: 7 September 2018
Please cite this article as: Hu H, Mizoue T, Sasaki N, Ogasawara T, Tomita K, Nagahama S, Hori A, Nishihara A, Imai T, Yamamoto M, Eguchi M, Kochi T, Miyamoto T, Honda T, Nakagawa T, Yamamoto S, Okazaki H, Uehara A, Shimizu M, Murakami T, Kuwahara K, Nanri A, Konishi M, Kabe I, Dohi S, for the Japan Epidemiology Collaboration on Occupational Health Study Group, Prediabetes and cardiovascular disease risk: A nested case-control study, Atherosclerosis (2018), doi: 10.1016/ j.atherosclerosis.2018.09.004. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Prediabetes and cardiovascular disease risk: A nested case-control study
2
Huanhuan Hu a,*, Tetsuya Mizouea, Naoko Sasakib, Takayuki Ogasawarab, Kentaro
3
Tomitac, Satsue Nagahamad, Ai Horie, Akiko Nishiharaf, Teppei Imaif, Makoto
4
Yamamotog, Masafumi Eguchih, Takeshi Kochih, Toshiaki Miyamotoi, Toru Hondaj,
5
Tohru Nakagawaj, Shuichiro Yamamotoj, Hiroko Okazakik, Akihiko Ueharal, Makiko
6
Shimizum, Taizo Murakamim, Keisuke Kuwaharaa,n, Akiko Nanria,o, Maki Konishia,
7
Isamu Kabeh, Seitaro Dohik, for the Japan Epidemiology Collaboration on Occupational
8
Health Study Group
9
a Department of Epidemiology and Prevention, National Center for Global Health and
M AN U
SC
RI PT
1
Medicine, Tokyo, Japan
11
b Mitsubishi Fuso Truck and Bus Corporation, Kanagawa, Japan
12
c Mitsubishi Plastics, Inc., Tokyo, Japan
13
d All Japan Labour Welfare Foundation, Tokyo, Japan
14
e Department of Global Public Health, University of Tsukuba, Ibaraki, Japan
15
f Azbil Corporation, Tokyo, Japan
16
g YAMAHA CORPORATION, Shizuoka, Japan
17
h Furukawa Electric Co., Ltd., Tokyo, Japan
18
i Nippon Steel & Sumitomo Metal Corporation Kimitsu Works, Chiba, Japan
AC C
EP
TE D
10
1
ACCEPTED MANUSCRIPT
j Hitachi, Ltd., Ibaraki, Japan
20
k Mitsui Chemicals, Inc., Tokyo, Japan
21
l Seijinkai Shizunai Hospital, Hokkaidō, Japan
22
m Mizue Medical Clinic, Keihin Occupational Health Center, Kanagawa, Japan
23
n Teikyo University Graduate School of Public Health, Tokyo, Japan
24
o Department of Food and Health Sciences, International College of Arts and Sciences,
25
Fukuoka Women’s University, Fukuoka, Japan
26
*Address correspondence to Huanhuan Hu, Department of Epidemiology and
27
Prevention, National Center for Global Health and Medicine, Toyama 1-21-1,
28
Shinjuku-ku, Tokyo 162-8655, Japan.
29
Phone: +81-3-3202-7181, Fax: +81-3-3202-7364
30
E-mail: [email protected]
EP
TE D
M AN U
SC
RI PT
19
AC C
31 32
Abstract
33
Background and aims: We aimed to examine the risk of cardiovascular disease (CVD)
34
with persistent prediabetes during the last four years prior to a CVD event in a large
35
occupational cohort in Japan.
36
Methods: We performed a nested case-control study using data from the Japan
2
ACCEPTED MANUSCRIPT
Epidemiology Collaboration on Occupational Health Study. A total of 197 registered
38
cases of CVD were identified and matched individually with 985 controls according to
39
age, sex, and worksite. Prediabetes was defined as fasting plasma glucose 100-125
40
mg/dL and/or HbA1c 5.7-6.4%. Persistent prediabetes was defined as having
41
prediabetes at years one and four prior to the onset/index date; persistent
42
normoglycemia was similarly defined. Associations between prediabetes and CVD risk
43
were assessed using conditional logistic regression models.
44
Results: Compared with people with persistent normoglycemia over the four years prior
45
to the onset/index date, the unadjusted odds ratio (95% confidence interval) for CVD
46
was 2.88 (1.56, 5.32) for people with persistent prediabetes. After adjusting for BMI,
47
smoking, hypertension, and dyslipidemia assessed four years before the onset/index date,
48
the association was slightly attenuated to an OR (95% confidence interval) of 2.62 (1.31,
49
5.25). Prediabetes assessed at single time points was also associated with an elevated
50
risk of CVD, with multivariable-adjusted odds ratio (95% confidence interval) of 1.72
51
(1.12, 2.64) and 2.13 (1.32, 3.43) for prediabetes at one and four years prior to the
52
onset/index date, respectively.
53
Conclusions: Prediabetes is associated with an increased risk of CVD. Identification
54
and management of prediabetes are important for the prevention of CVD.
AC C
EP
TE D
M AN U
SC
RI PT
37
3
ACCEPTED MANUSCRIPT
Keywords
56
Prediabetes; Diabetes; Cardiovascular disease; Nested case-control study
57
Introduction
58
Prediabetes represents a high-risk state for type 2 diabetes and affects various
59
populations around the world [1]. Globally, 352 million adults are living with
60
prediabetes (impaired glucose tolerance), and this number is projected to increase to 587
61
million in 2040 [2]. Japan is one of the top ten countries with the highest number of
62
adults with prediabetes [2]. Meta-analyses of cohort studies, in which only baseline
63
glycemia levels were measured, show that prediabetes is associated with an increased
64
risk of cardiovascular disease (CVD) [3,4]. However, given that 5% -10% of individuals
65
with prediabetes develop diabetes annually [1], it remains unclear whether this elevated
66
risk of CVD is due to prediabetes or the transition from prediabetes to diabetes during
67
follow-up.
AC C
EP
TE D
M AN U
SC
RI PT
55
To the best of our knowledge, only two prospective studies have examined the
68
69
prediabetes-CVD association while excluding people who developed diabetes during
70
follow-up [5,6]. A U.K. study reported an increased risk of CVD associated with
71
prediabetes in Europeans but not South Asians [5], and a Finnish study showed
72
increased morbidity and mortality of CVD associated with prediabetes [6]. One
4
ACCEPTED MANUSCRIPT
important limitation of these studies is that people who had prediabetes at baseline but
74
returned to normoglycemia during follow-up were not identified and therefore were not
75
excluded from the analyses, leading to exposure misclassification. To better assess the
76
association between prediabetes and CVD, studies using repeated measurements of
77
blood glucose are needed.
SC
RI PT
73
Using data from the Japan Epidemiology Collaboration of Occupational Health
79
(J-ECOH) Study, in which annual health checkup data were available throughout the
80
follow-up period [7,8], we performed a nested case-control study to examine the risk of
81
CVD posed by persistent prediabetes during the four years immediately prior to a CVD
82
event in comparison to cases of persistent normoglycemia. We also investigated the
83
risks of CVD associated with prediabetes one year or four years before the CVD event,
84
respectively.
85
Materials and methods
86
Study design
87
The J-ECOH Study is an ongoing multicenter, health checkup-based cohort study
88
among workers from several companies in Japan, and has been described elsewhere
89
[7,8]. In Japan, workers are obliged to undergo a health checkup at least once a year
90
under the Industrial Safety and Health Act; nearly all workers attend their health
AC C
EP
TE D
M AN U
78
5
ACCEPTED MANUSCRIPT
checkup each year. Briefly, participants in the J-ECOH Study underwent routine
92
physical and laboratory examinations (blood glucose, blood lipids, etc.) each year.
93
Additionally, a questionnaire that covered medical history, health-related lifestyle, and
94
work environment was completed. Annual health checkup data between January 2008
95
and December 2016 or between April 2008 and March 2017 were collected from over
96
100,000 employees in 11 participating companies. The study protocol, including the
97
consent procedure, was approved by the Ethics Committee of the National Center for
98
Global Health and Medicine, Japan. Using the J-ECOH Study data, we performed a
99
nested case-control study to investigate the association between prediabetes and incident
M AN U
SC
RI PT
91
CVD.
101
Health checkup
102
Body height and weight were measured using a scale with participants wore light
103
clothes and no shoes. Body mass index (BMI) was calculated as weight in kilograms
104
divided by the square of height in meters. Blood pressure was measured using an
105
automatic blood pressure monitor. Hypertension was defined as either systolic blood
106
pressure of at least 140 mmHg, diastolic blood pressure of at least 90 mmHg, or use of
107
treatment for hypertension [9]. Blood glucose was measured using either the enzymatic
108
or glucose oxidase peroxidative electrode method based on each company’s protocol.
AC C
EP
TE D
100
6
ACCEPTED MANUSCRIPT
HbA1c was measured using either a latex agglutination immunoassay,
110
high-performance liquid chromatography, or the enzymatic method. Triglyceride,
111
low-density lipoprotein-cholesterol, and high-density lipoprotein-cholesterol levels were
112
measured using the enzymatic method. Dyslipidemia was defined as either a
113
low-density lipoprotein-cholesterol level of at least 140 mg/dL, high-density
114
lipoprotein-cholesterol level of less than 40 mg/dL, triglyceride level of at least 150
115
mg/dL, or use of medications for dyslipidemia [10]. All of the laboratories involved in
116
the health checkups of the participating companies received satisfactory scores (rank A
117
or a score >95 out of 100) from external quality-control agencies.
118
Ascertainment of CVD cases and control selection
119
CVD cases were identified through a disease registry, which was set up in April 2012
120
within the J-ECOH study to collect data on CVD events from the occupational health
121
physicians of the participating companies [8]. For most nonfatal cases, occupational
122
physicians confirmed the diagnosis of each CVD event based on medical certificates,
123
which were written by a treating physician and submitted to the company through the
124
worker. Because the submission of a medical certificate is required when taking
125
long-term sick leave, this registry primarily covers relatively severe cases. For fatal
126
cases, occupational physicians judged the cause of death based on available information,
AC C
EP
TE D
M AN U
SC
RI PT
109
7
ACCEPTED MANUSCRIPT
including death certificates and information obtained from the bereaved family or
128
colleagues. In the CVD registry, approximately 80% cases were confirmed through
129
medical certificates, 10% were self-reported or from family members or colleagues, and
130
10% were from other sources. Each case was coded according to the 10th revision of the
131
International Classification of Diseases.
SC
RI PT
127
In the present study, cases were patients with a first event of myocardial infarction
133
or stroke between April 2012 and March 2017. A total of 235 incident CVD cases were
134
identified. After removing patients (n=38) who did not attend health checkup one year
135
before the onset date, 197 CVD cases (137 strokes and 60 myocardial infarctions; 36
136
fatal cases) remained in the present study (Figure 1).
TE D
M AN U
132
Controls were selected from study participants who did not self-report stroke or
138
heart disease at J-ECOH Study entry and did not develop CVD during the follow-up
139
period. Those who self-reported a history of CVD at annual health checkups during the
140
study period were also excluded. For each case, we created a pool of controls who were
141
matched by worksite, sex, and date of birth (±2 years). Controls were assigned an index
142
date that corresponded to the onset date of their matched case. We excluded people who
143
did not attend health checkup at one year prior to the index date. Finally, for a given
144
case, we randomly selected up to five controls from the pool of eligible controls. Once a
AC C
EP
137
8
ACCEPTED MANUSCRIPT
145
control was sampled, he/she was not again chosen as a control for other cases. A total of
146
985 matched controls were included in the present study. Using data from the 197 cases and 985 matched controls, we examined the risk of
RI PT
147
CVD associated with prediabetes at one year prior to the CVD event. After removing
149
people who did not attend a health checkup four years before the onset/index date, we
150
investigated the risk of CVD associated with having prediabetes four years before the
151
CVD event in 152 cases and 675 controls. We further excluded people who had
152
prediabetes at four years before but not one year before the onset/index date and those
153
who had normoglycemia at four years before but not one year before the onset/index
154
date, leaving 115 cases and 381 controls for examining the risk of CVD with persistent
155
prediabetes over the last four years before the CVD event.
156
Exposures
157
We retrieved data on diabetes treatment status, fasting plasma glucose (FPG), and
158
HbA1c collected at one and four years before the onset/index date. Diabetes was
159
defined as either FPG ≥126 mg/dL, HbA1c ≥6.5%, or receiving medical treatment for
160
diabetes according to the American Diabetes Association criteria [11]. Prediabetes was
161
defined as FPG 100-125 mg/dL and/or HbA1c 5.7-6.4% [11]. Normoglycemia was
162
defined as FPG <100 mg/dL and HbA1c <5.7%. Persistent prediabetes was defined for
AC C
EP
TE D
M AN U
SC
148
9
ACCEPTED MANUSCRIPT
those who had prediabetes at both one and four years prior to the onset/index date.
164
Persistent normoglycemia was defined for those who had normoglycemia at both one
165
and four years prior to the onset/index date. We also defined prediabetes using either
166
FPG (100-125 mg/dL) or HbA1c (5.7-6.4%) to examine whether the risk of CVD
167
associated with prediabetes differs between the two definitions.
168
Statistical analysis
169
Characteristics of participants were expressed as percentages and means for categorical
170
and continuous variables, respectively. Chi-square tests for categorical variables and
171
t-tests for continuous variables were used to examine differences in characteristics
172
between cases and controls.
TE D
M AN U
SC
RI PT
163
Conditional logistic regression was used to estimate odds ratios (ORs) and 95%
173
confidence intervals (CIs) for the development of CVD associated with persistent
175
prediabetes. In model 1, the analysis was conditioned on the matching variables (sex
176
and age). In model 2, we adjusted for a priori-specified potential confounders, including
177
BMI, smoking (current smoker, non-current smoker), hypertension, and dyslipidemia
178
assessed at four years before the onset/index date. We also examined the associations
179
between CVD risk and prediabetes assessed at different time points: one and four years
180
before the onset/index date, respectively.
AC C
EP
174
10
ACCEPTED MANUSCRIPT
The present study had 73% power to detect a relative risk of 2.0 with statistical
181
significance (two-sided alpha level of 0.05), based on the given condition of the
183
proportion of people with persistent prediabetes among controls (50%), and the number
184
of cases (47 persistent prediabetes and 23 persistent normoglycemia) and number of
185
matched controls (350). All statistical analyses were performed using SAS version 9.4
186
(SAS Institute, Cary, NC, USA). A two-sided p < 0.05 was considered statistically
187
significant.
188
Results
189
Table 1 shows the characteristics of cases and controls at one year prior to the
190
onset/index date. Cases had higher means for BMI, triglyceride, low-density lipoprotein
191
cholesterol, blood pressure, FPG, and HbA1c than controls. The prevalence rates of
192
smoking, hypertension, dyslipidemia, and diabetes were higher among cases compared
193
with controls. There was no material difference in the proportion of patients who were
194
receiving medication for hypertension or dyslipidemia between cases and controls.
195
Table 2 shows the association between prediabetes and CVD. Compared with
AC C
EP
TE D
M AN U
SC
RI PT
182
196
people with persistent normoglycemia over the past four years before the onset/index
197
date, the unadjusted OR (95% CI) for CVD was 2.88 (1.56, 5.32) for people with
198
persistent prediabetes. After adjusting for BMI, smoking, hypertension, and
11
ACCEPTED MANUSCRIPT
dyslipidemia, the association was slightly attenuated to an OR (95% CI) of 2.62 (1.31,
200
5.25). For people with prediabetes one year before, the OR (95% CI) was 1.72 (1.12,
201
2.64) compared with those with normoglycemia, after adjusting for BMI, smoking,
202
hypertension, and dyslipidemia assessed one year before the onset/index date. For
203
people with prediabetes four years before, the OR (95% CI) was 2.13 (1.32, 3.43), with
204
adjustment for BMI, smoking, hypertension, and dyslipidemia assessed four years
205
before the onset/index date.
M AN U
SC
RI PT
199
Table 3 shows the risk of CVD associated with prediabetes as defined using either
206
FPG or HbA1c. When prediabetes was defined as FPG 100-125 mg/dL, people with
208
prediabetes tended to have a higher risk of CVD compared with those without
209
prediabetes. Multivariable-adjusted ORs (95% CI) were 1.76 (0.91, 3.38), 1.33 (0.88,
210
2.01), and 1.77 (1.10, 2.86) for people with persistent prediabetes, prediabetes one year
211
before, and prediabetes four years before, respectively. When prediabetes was defined
212
as HbA1c 5.7-6.4%, the corresponding ORs (95% CI) were 2.07 (1.09, 3.97) 1.28 (0.82,
213
2.02), and 1.93 (1.21, 3.08), respectively.
214
Discussion
215
In this nested case-control study, we found that prediabetes (defined as FPG 100-125
216
mg/dL and/or HbA1c 5.7-6.4%) one or four years prior to the onset/index date was
AC C
EP
TE D
207
12
ACCEPTED MANUSCRIPT
associated with significantly increased risk of CVD. More importantly, a much higher
218
risk for CVD was observed among people with persistent prediabetes. To our
219
knowledge, this is one of the few studies to examine whether prediabetes is associated
220
with the development of CVD.
RI PT
217
The present findings are in line with those of the existing yet limited studies [5,6].
SC
221
We found that prediabetes one and four years prior to the onset/index date was
223
associated with a 1.7- and 2.1-fold increased risk of CVD, respectively. However, when
224
the analyses were limited to those who had prediabetes, both one and four years before
225
the onset/index date, people with persistent prediabetes were 2.6 times more likely to
226
develop CVD. This difference in the strength of associations implies that the association
227
between prediabetes and CVD risk may not be accurately captured if prediabetes was
228
assessed at a single time point due to changes in glycemia levels over time. In our study,
229
out of 291 controls with prediabetes four years before the index date, 22 developed
230
diabetes within three years, and 68 returned to normoglycemia. Of 303 controls with
231
normoglycemia four years prior, 92 developed prediabetes three years later. Using
232
multiple measures of glycemia to minimize the misclassification of exposure, our data
233
confirm that prediabetes is a risk factor for CVD. Proactive identification and
234
management of prediabetes are warranted to reduce future risk of CVD.
AC C
EP
TE D
M AN U
222
13
ACCEPTED MANUSCRIPT
We observed a more pronounced association with HbA1c-based prediabetes than
235
FPG-based prediabetes, consistent with findings from the Whitehall II Study [12]. A
237
meta-analysis showed no significant difference in the risk of CVD between baseline
238
prediabetes defined using FPG (100-125 mg/dL) and that defined using HbA1c
239
(5.7-6.4%) [3]. Of 16 studies included in that meta-analysis, only three made a
240
head-to-head comparison between the two definitions; two reported a significant
241
association with HbA1c-based prediabetes but not with FPG-based prediabetes, and
242
another did not detect any association with these definitions [3]. Taken together,
243
prediabetes defined by HbA1c appears to be more strongly associated with CVD than
244
prediabetes defined by FPG.
TE D
M AN U
SC
RI PT
236
Non-glucose-mediated and glucose-mediated mechanisms have been proposed as
245
explanations for the association between prediabetes and CVD. One of the
247
non-glucose-mediated mechanisms involves insulin resistance, which is an underlying
248
condition of prediabetes [13]. It promotes atherogenesis and atherosclerotic plaque
249
formation via down-regulation of insulin signaling at the level of the intimal cells that
250
participate in atherosclerosis, including macrophages, vascular smooth muscle cells, and
251
endothelial cells [14,15]. The other non-glucose-mediated mechanisms involve obesity
252
and inflammation (adipocytokines and oxidative stress), dyslipidemia (atherogenic
AC C
EP
246
14
ACCEPTED MANUSCRIPT
lipoproteins and increased circulating free fatty acids), and prothrombosis state
254
(increased levels of PAI-1 and platelet activation), which are commonly present in
255
people with prediabetes [14,15]. As for glucose-mediated mechanisms, hyperglycemia
256
could promote atherosclerosis through increased oxidative stress, activation of
257
thrombosis, and epigenetic changes (increased expression of inflammatory genes via
258
NF-κB), which can be induced even by transient or episodic hyperglycemia [14,15].
259
Strengths and limitations
M AN U
SC
RI PT
253
A major advantage of the present study was the periodic assessment of diabetic
260
status using HbA1c and FPG until one year before the onset of CVD, which enabled us
262
to maximally reduce the degree of misclassification of prediabetes. In addition, potential
263
confounders such as smoking, dyslipidemia, and hypertension were also assessed
264
annually. Nevertheless, this study has some limitations. First, the present study had a
265
power less than 80% to detect an association with a relative risk of 2.0. Using repeated
266
assessment of diabetic status, however, we found a stronger association (relative risk of
267
2.5 or more) between prediabetes and CVD risk, increasing the power of our study to
268
90%. Second, we only have recent data on diabetes status (within the past four years
269
prior to the onset of CVD). The lack of data for earlier periods precludes us from
270
analyzing the effect of long-term exposure to prediabetes on CVD development. Third,
AC C
EP
TE D
261
15
ACCEPTED MANUSCRIPT
only workers who took long-term sick leave were required to submit a medical
272
certificate to the company; thus, milder forms of CVD were not well-covered by this
273
registration system. Fourth, we defined prediabetes using FPG and HbA1c but not
274
2-hour oral glucose tolerance test, which has not been used in regular health checkups.
275
As such, people with impaired glucose tolerance may be misclassified as
276
normoglycemia, resulting in an underestimation of the association between prediabetes
277
and CVD. Lastly, due to lack of data on socioeconomic status, family history of CVD,
278
and lifestyle characteristics other than smoking (e.g., diet, physical activity), we were
279
unable to control for the potential effects of these factors.
M AN U
SC
RI PT
271
In conclusion, the present nested case-control study based on repeated
TE D
280
measurements of fasting glucose and HbA1c indicates that prediabetes is associated
282
with an increased risk of CVD. Identification and management of prediabetes are
283
important for the prevention of CVD. More research is required to confirm or refute
284
whether pre-diabetic patients who are free of conventional risk factors such as
285
hypertension and hypercholesterolemia are at increased risk of CVD.
AC C
EP
281
286
287
Conflict of interest
288
The authors declared they do not have anything to disclose regarding conflict of interest
16
ACCEPTED MANUSCRIPT
with respect to this manuscript.
290
Financial support
291
This study was supported by the Industrial Health Foundation, Industrial Disease
292
Clinical Research Grants (grant numbers 140202-01, 150903-01, 170301-01), Japan
293
Society for the Promotion of Science KAKENHI (grant number 16H05251), and Grant
294
of National Center for Global Health and Medicine (grant number 28-Shi-1206).
295
Author contributions
296
S.D. and T.Mizoue were involved in the design of the study as the principal
297
investigators; N.S., T.O., K.T., S.N., A.H., A.Nishihara, T.I., M.Y., M.E., T.K.,
298
T.Miyamoto, T.H., T.N., S.Y., H.O., A.U., M.S., T.Murakami, I.Kabe., and S.D.
299
collected health check-up data; M.K. and A.Nanri cleaned CVD data; T.Mizoue, M.K.,
300
and K.K. created database; H.H. and T.Mizoue drafted the plan for the data analysis;
301
H.H. conducted data analysis; H.H. drafted manuscript; HH and T Mizoue had primary
302
responsibility for final content; and all authors were involved in interpretation of the
303
results and revision of the manuscript and approved the final version of the manuscripts,
304
had full access to all of the data in the study, and take responsibility for the integrity of
305
the data and the accuracy of the data analysis. HH and T Mizoue are guarantors.
AC C
EP
TE D
M AN U
SC
RI PT
289
306
17
ACCEPTED MANUSCRIPT
Acknowledgements
308
We thank Dr. Toshiteru Okubo (Chairperson of Industrial Health Foundation) for
309
scientific advice on the conduct of J-ECOH Study; Eri Yamada (National Center for
310
Global Health and Medicine) for data management; and Rika Osawa (National Center
311
for Global Health and Medicine) for administrative support.
312
References
313
1 Tabák AG, Herder C, Rathmann W, et al. Prediabetes: a high-risk state for diabetes
314
development. Lancet 2012; 379:2279-2290.
315
2 International Diabetes Federation. IDF Diabetes Atlas. 8th ed. Available from
316
www.diabetesatlas.org. Accessed December 2017
317
3 Huang Y, Cai X, Mai W, et al. Association between prediabetes and risk of
318
cardiovascular disease and all-cause mortality: systematic review and meta-analysis.
319
BMJ 2016;355: i5953.
320
4 Ford ES, Zhao G, Li C. Pre-diabetes and the risk for cardiovascular disease: a
321
systematic review of the evidence. J Am Coll Cardiol 2010; 55:1310-1317.
322
5 Eastwood SV, Tillin T, Sattar N, et al. Associations between prediabetes, by three
323
different diagnostic criteria, and incident CVD differ in South Asians and Europeans.
324
Diabetes Care 2015; 38:2325-2332.
AC C
EP
TE D
M AN U
SC
RI PT
307
18
ACCEPTED MANUSCRIPT
6 Qiao Q, Jousilahti P, Eriksson J, et al. Predictive properties of impaired glucose
326
tolerance for cardiovascular risk are not explained by the development of overt diabetes
327
during follow-up. Diabetes Care 2003; 26:2910-2914.
328
7 Hu H, Nagahama S, Nanri A, et al. Duration and degree of weight change and risk of
329
incident diabetes: Japan Epidemiology Collaboration on Occupational Health Study.
330
Prev Med 2017; 96:118-123.
331
8 Hu H, Nakagawa T, Honda T, et al. Metabolic syndrome over four years before the
332
onset of cardiovascular disease: nested case-control Study. Circ J 2018; 82:430-436.
333
9 Shimamoto K, Ando K, Fujita T, et al. The Japanese Society of Hypertension
334
Guidelines for the Management of Hypertension (JSH 2014). Hypertens Res 2014;
335
37:253–390.
336
10 Teramoto T, Sasaki J, Ishibashi S, et al. Executive Summary of the Japan
337
Atherosclerosis Society Guidelines for the Diagnosis and Prevention of Atherosclerotic
338
Cardiovascular Diseases in Japan -2012 version. J Atherosclerosis Thromb 2013;
339
20:517–523.
340
11 American Diabetes Association. American Diabetes Association Standards of
341
medical care in diabetes–2017. Diabetes Care 2017; 40(Suppl.1): S1-S135.
342
12 Vistisen D, Witte DR, Brunner E, et al. Risk of Cardiovascular Disease and Death in
AC C
EP
TE D
M AN U
SC
RI PT
325
19
ACCEPTED MANUSCRIPT
Individuals With Prediabetes Defined by Different Criteria: The Whitehall II Study.
344
Diabetes Care 2018; 41:899-906.
345
13 Abdul-Ghani MA, DeFronzo RA. Pathophysiology of prediabetes. Curr Diab Rep
346
2009; 9:193-199.
347
14 Milman S, Crandall JP. Mechanisms of vascular complications in prediabetes. Med
348
Clin North Am 2011; 95:309-325.
349
15 Buysschaert M, Medina JL, Bergman M, et al. Prediabetes and associated disorders.
350
Endocrine 2015; 48:371-393
M AN U
351
SC
RI PT
343
TE D
352
353
EP
Table 1 Characteristics of cases and controls at one year before the onset/index date Cases
Controls
197
985
52.9±7.3
52.8±7.3
91.9
91.9
25.0±4.0
23.9±3.6a
54.8
31.8a
TG (mg/dL)
149.9±96.9
125.0±84.5a
HDL-C (mg/dL)
53.4±15.1
58.8±15.8a
N
AC C
354
Age (years) Men, %
BMI (kg/m2) Current smoker, %
20
ACCEPTED MANUSCRIPT
129.2±37.9
123.6±29.4
Dyslipidemia, %
69.0
51.5a
Lipid-lowering treatment, %b
19.9
20.9
SBP (mmHg)
134.7±17.3
124.4±15.8a
DBP (mmHg)
85.2±10.6 65.5
Anti-hypertensive treatment, %c
52.7
FPG (mg/dL)
34.9a
Diabetes, %
57.6
120.3±50.5
102.9±18.2a
6.2±1.3
5.7±0.7a
M AN U
HbA1c (%)
78.9±11.2a
SC
Hypertension, %
RI PT
LDL-C (mg/dL)
33.5
13.7a
355
a
356
BMI: body mass index; CVD, cardiovascular disease; DBP: diastolic blood pressure;
357
FPG: fasting plasma glucose; HbA1c, glycated hemoglobin; HDL-C: high-density
358
lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; SBP, systolic blood
359
pressure; TG, triglyceride.
360
b
The denominator is the total number of people with dyslipidemia.
361
c
The denominator is the total number of people with hypertension.
362
363
364
365 21
AC C
EP
TE D
p < 0.05
ACCEPTED MANUSCRIPT
366
Table 2 Associations between prediabetes and risk of cardiovascular disease Case Control
Odds ratio (95% CI)
Over the four years prior to the onset/index date
Model 2
RI PT
Model 1
23
161
1
Persistent prediabetes
47
155
2.88 (1.56, 5.32)
Diabetes
45
65
6.50 (3.31, 12.8)
Normoglycemia
47
Prediabetes
84
Diabetes
66
M AN U
One year before the onset/index date
1
2.62 (1.31, 5.25)
SC
Persistent normoglycemia
4.11 (1.81, 9.30)
418
1
1
432
1.98 (1.33, 2.96)
1.72 (1.12, 2.64)
135
5.19 (3.28, 8.22)
3.41 (2.05, 5.69)
Four years before the onset/index date
Prediabetes
303
1
1
71
291
2.40 (1.53, 3.77)
2.13 (1.32, 3.43)
45
81
5.66 (3.29, 9.74)
3.43 (1.87, 6.32)
EP
Diabetes
36
TE D
Normoglycemia
Model 1, unadjusted.
368
Model 2, adjusted for BMI, current smoker (yes or no), hypertension, and dyslipidemia
AC C
367
369 370 371 372 373 374 375 376 377 22
ACCEPTED MANUSCRIPT
378
Table 3 Associations between prediabetes (defined as FPG 100-125 mg/dL or HbA1c
379
5.7-6.4%, respectively) and risk of cardiovascular disease Control
Odds ratio (95% CI) Model 1
FPG (mg/dL)
42
274
Persistent FPG 100-125
31
133
Diabetes
45
One year before the onset/index date 68
100-125
63
Diabetes
1
2.12 (1.16, 3.89)
1.76 (0.91, 3.38)
80
4.34 (2.40, 7.83)
2.36 (1.17, 4.75)
518
1
1
332
1.64 (1.11, 2.42)
1.33 (0.88, 2.01)
4.29 (2.82, 6.54)
2.73 (1.71, 4.36)
TE D
<100
1
M AN U
Persistent FPG <100
66
Model 2
SC
Over the four years prior to the onset/index date
RI PT
Case
135
Four years before the onset/index date
100-125
AC C
Diabetes
57
EP
<100
381
1
1
49
204
2.02 (1.28, 3.18)
1.77 (1.10, 2.86)
45
80
4.55 (2.74, 7.57)
2.77 (1.56, 4.91)
HbA1c (%)
Over the four years prior to the onset/index date Persistent HbA1c <5.7
46
376
1
1
Persistent HbA1c 5.7-6.4
26
111
2.38 (1.34, 4.22)
2.07 (1.09, 3.97)
Diabetes
45
83
4.95 (2.86, 8.59)
3.27 (1.71, 6.26)
One year before the onset/index date
23
ACCEPTED MANUSCRIPT
86
613
1
1
5.7-6.4
45
237
1.48 (0.98, 2.23)
1.28 (0.82, 2.02)
Diabetes
66
135
3.82 (2.58, 5.65)
2.57 (1.65, 4.01)
Four years before the onset/index date
RI PT
<5.7
57
421
1
1
5.7-6.4
53
196
2.19 (1.43, 3.37)
1.93 (1.21, 3.08)
Diabetes
45
83
4.45 (2.75, 7.20)
2.77 (1.60, 4.78)
SC
<5.7
Model 1, unadjusted.
381
Model 2, adjusted for BMI, current smoker (yes or no), hypertension, and dyslipidemia
M AN U
380
382 383 384 385
TE D
386 387 388 389
EP
390 391 392
AC C
393 394 395 396 397 398 399 400 401 402 403 24
ACCEPTED MANUSCRIPT
404
Figure 1 Flow diagram of study population, J-ECOH Study, Japan, 2008-2016
405
AC C
EP
TE D
M AN U
SC
RI PT
406
25
ACCEPTED MANUSCRIPT
143,499 participants who had health checkup
RI PT
Excluded (n=940): History of stroke or ischemic heart disease at J-ECOH study entry 142,324 participants
Excluded (n=19,791): Did not attend any follow-up examination
SC
Excluded (n=1,120): Self-reported ischemic heart disease and stroke during follow-up
M AN U
235 registered CVD
121,413 participants without CVD during follow-up
TE D
Excluded (n=38) Did not attend the health check-up at 1 year before the event date 197 registered CVD
985 controls
AC C
EP
Excluded 45 cases and 310 controls who did not attend health check-up at 4 years before the event/index date
Excluded 24 cases and 158 controls who had prediabetes at 4 years before but not 1 year before the event/index date Excluded 13 cases and 136 controls who had normoglycemia at 4 years before but not 1 year before the event date
To examine the risk of CVD with persistent prediabetes during the last four years prior to a CVD event
To examine the risk of CVD associated with prediabetes one year before the CVD event
To examine the risk of CVD associated with prediabetes four years before the CVD event
Figure 1 Flow diagram of study population, J-ECOH Study, Japan, 2008-2016
ACCEPTED MANUSCRIPT
Prediabetes and cardiovascular disease risk: A nested case-control study
•
RI PT
Highlights Prediabetes assessed at single time points was associated with an elevated risk of CVD.
Much higher risk of CVD was observed among people with persistent prediabetes compared with persistent normoglycemia.
Prediabetes is associated with a higher risk of CVD.
AC C
EP
TE D
M AN U
•
SC
•