Metabolic syndrome components and diabetes incidence according to the presence or absence of impaired fasting glucose: The Japan Epidemiology Collaboration on Occupational Health Study

Background We prospectively examined the association of diabetes risk with the number of metabolic abnormalities, as well as their combinations, according to the presence or absence of impaired fasting glucose (IFG) in a large-scale Japanese working population. Methods Participants included 55,271 workers at 11 companies who received periodic health check-ups between 2008 and 2013. The metabolic syndrome (MetS) components were defined using the 2009 Joint Interim Statement. IFG was defined as fasting plasma glucose 5.6–6.9 mmol/L. Diabetes newly diagnosed after the baseline examination was defined according to the American Diabetes Association criteria. We calculated the hazard ratios (HRs) for diabetes incidence using the Cox proportional hazards model. Results During the follow-up period (median 4.95 years), 3183 subjects developed diabetes. In individuals with normal fasting glucose levels, the risk of diabetes increased steadily with the increasing number of MetS components; the multivariable-adjusted HRs for incident diabetes for the number of MetS components were 2.0, 4.3, 7.0, and 10.0 for one, two, three, or four MetS components, respectively, compared with the absence of components. A similar association was observed among individuals with IFG; the corresponding HRs were 17.6, 23.8, 33.9, and 40.7. The combinations that included central obesity appeared to be more strongly associated with diabetes risk than other combinations with the same number of MetS components within the same glucose status. Conclusions Our findings indicate that risk stratification of individuals by the presence or absence of IFG and the number of MetS components can detect individuals with a high risk of diabetes.


a b s t r a c t
Background: We prospectively examined the association of diabetes risk with the number of metabolic abnormalities, as well as their combinations, according to the presence or absence of impaired fasting glucose (IFG) in a large-scale Japanese working population. Methods: Participants included 55,271 workers at 11 companies who received periodic health check-ups between 2008 and 2013. The metabolic syndrome (MetS) components were defined using the 2009 Joint Interim Statement. IFG was defined as fasting plasma glucose 5.6e6.9 mmol/L. Diabetes newly diagnosed after the baseline examination was defined according to the American Diabetes Association criteria. We calculated the hazard ratios (HRs) for diabetes incidence using the Cox proportional hazards model. Results: During the follow-up period (median 4.95 years), 3183 subjects developed diabetes. In individuals with normal fasting glucose levels, the risk of diabetes increased steadily with the increasing number of MetS components; the multivariable-adjusted HRs for incident diabetes for the number of MetS components were 2.0, 4.3, 7.0, and 10.0 for one, two, three, or four MetS components, respectively, compared with the absence of components. A similar association was observed among individuals with IFG; the corresponding HRs were 17.6, 23.8, 33.9, and 40.7. The combinations that included central

Introduction
Metabolic syndrome (MetS) is a cluster of risk factors, including raised blood pressure, dyslipidemia, impaired fasting glucose (IFG), and central obesity. 1,2 Diabetes risk has been shown to increase with the number of MetS components. 3e6 Given that IFG is an early stage of developing diabetes 7 and a strong predictor of diabetes, 8 it would be preferable to treat the glucose component separately from other components in stratifying diabetes risk. Few studies, however, have adopted this analytic strategy in assessing diabetes risk in relation to the number of metabolic abnormalities. 9 In addition to the number of MetS components, it remains unclear whether specific combinations of the components confer higher risks of diabetes. Wilson et al reported that combinations that included central obesity and IFG were more strongly associated with diabetes risk than others with the same number of components. 5,10 However, that study was relatively small in size (n z 2500), so a larger study is required to provide a more stable risk estimate for each combination of metabolic components. Here, we prospectively examined the association of diabetes risk with the number of MetS components, as well as their combinations, according to the presence or absence of IFG in a large-scale multicenter cohort of Japanese male and female workers.

Study procedure
The Japan Epidemiology Collaboration on Occupational Health (J-ECOH) Study is an ongoing multicenter epidemiologic study among workers of 12 companies in Japan. 11,12 In Japan, employees are obliged to undergo a general health examination at least once a year under the Industrial Safety and Health Act. As of March 2015, 11 of 12 participating companies provided data from health check-ups that were conducted between January 2008 and December 2013 or between April 2008 and March 2014. The date of the earliest examination (mostly in 2008) was regarded as the baseline, but if the 2008 dataset contained a large amount of missing data, the data of the 2009 or 2010 examination was used as the baseline (for two companies). Subjects were followed from the baseline until the date of the most recent examination (maximally December 2013 or March 2014). The study protocol was approved by the Ethics Committee of the National Center for Global Health and Medicine, Japan. The requirement for written informed consent was waived.

Subjects
Of the 95,040 subjects with baseline data, we excluded 1532 subjects aged under 20 years, 15,660 subjects with missing data regarding the diagnosis of diabetes, and 5339 subjects with diabetes, which was defined as a fasting plasma glucose level of !7.0 mmol/L (126 mg/dL), casual plasma glucose level of !11.1 mmol/L (200 mg/dL), hemoglobin A1c (HbA1c) of !48 mmol/ mol (6.5%), or current use of an anti-diabetic drug, in accordance with the American Diabetes Association criteria. 13 After further exclusion of 13,695 subjects with missing data on waist circumference, fasting triglyceride, high density lipoprotein (HDL)cholesterol, blood pressure, current use of cholesterol-lowering drugs and anti-hypertensive drugs, or smoking status, 58,814 subjects remained. We further excluded 3543 subjects who did not attend any subsequent health check-ups, leaving 55,271 subjects (47,160 men and 8111 women) for the analysis. Compared with those who were included in the present study, those who were excluded (n ¼ 3543) were older and had higher levels of blood pressure and triglyceride and higher proportions of antihypertensive and hypolipidemic drug use.

Laboratory measurements
In the participating companies, HbA1c was measured using a latex agglutination immunoassay, the HPLC method, or the enzymatic method. Plasma glucose was measured using the enzymatic method or the glucose oxidase peroxidative electrode method, and HDL-cholesterol and triglycerides were measured using the enzymatic method. All laboratories involved in the health checkup at the participating companies have received satisfactory results (rank A or score >95 out of 100) from external quality control surveillance.

Statistical analysis
We calculated the person-time from the date of the baseline examination to the first date when diabetes was confirmed at a follow-up examination or to the date of the most recent examination, whichever came first. We also calculated the hazard ratios (HRs) and 95% confidence intervals (CIs) for the incidence of diabetes using Cox proportional-hazards regression models by individual MetS component and by the number of MetS components according to the presence or absence of IFG. In the multivariableadjusted model, we adjusted for age (years, continuous), sex, company, and smoking status (non-smoker or current smoker). We also calculated the HRs for each combination of MetS components. In that analysis, we created a dyslipidemia component if the subject had high triglyceride and/or low HDL-cholesterol level, in accordance with the definition of dyslipidemia in the 2009 Joint Interim Statement. 1 A two sided P < 0.05 was considered statistically significant. The SAS software package (version 9.3; SAS Institute, Cary, NC, USA) was used to perform all statistical analyses.
As shown in Fig. 1, in individuals with normal fasting plasma glucose levels, the risk of diabetes increased steadily with the increasing number of MetS components; the multivariableadjusted HRs for incident diabetes of the number of MetS components from 1 to 4 were 2.0 (95% CI, 1.6e2.6), 4.3 (95% CI, 3.4e5.4), 7.0 (95% CI, 5.3e9.2), and 10.0 (95% CI, 6.4e15.8), respectively, compared with the absence of components. A similar association was observed among individuals with IFG; the multivariableadjusted HRs for diabetes of the number of metabolic components from 1 to 4 were 17.6 (95% CI, 14.5e21.2), 23.8 (95% CI, 19.6e28.8), 33.9 (95% CI, 27.8e41.5), and 40.7 (95% CI, 30.6e54.1), respectively, compared with the absence of the component. Individuals with IFG alone (HR 12.7; 95% CI, 2.28e2.77) had a higher risk of diabetes than those with all other MetS components. Table 2 shows the HRs of diabetes for each combination of MetS components, including IFG, central obesity, high blood pressure, and dyslipidemia. Among individuals without IFG, the presence of one of the other MetS components (central obesity, high blood pressure, or dyslipidemia) was associated with a 2-fold higher risk of diabetes compared with the absence of the components. Among normalglucose individuals who had two MetS components, the combinations that included central obesity had a 68e74% higher risk of diabetes than those without central obesity (i.e., those with high blood pressure and dyslipidemia only). Among individuals with two metabolic abnormalities in addition to IFG, the combinations that included central obesity were associated with a 23e25% higher risk of diabetes than those without central obesity (i.e., the combination that included both high blood pressure and dyslipidemia). The combinations that included central obesity appear to be more strongly associated with diabetes risk than others with the same number of MetS components within the same glucose status.

Discussion
In a large-scale cohort of the working population in Japan, we found that the risk of diabetes incidence sharply increased with the increasing number of MetS components and that the combinations that included the central obesity component were more strongly associated with diabetes risk than others with the same number of metabolic abnormalities with the same glucose status. This study is among few prospective investigations of diabetic risk in relation to specific combinations of MetS components.
Our findings of a steadily increasing trend of diabetes risk with the number of MetS components for both IFG and non-IFG are in line with those of Japanese population-based studies. 9,10 As expected, those with IFG had a much higher risk of diabetes than those without IFG for a given number of MetS components; the diabetes risk was 6.3-, 4.1-, 3.4-, and 3.4-fold higher in those with IFG than that in those without IFG for individuals with 1e4 metabolic abnormalities, respectively. In the Framingham Offspring Study, Wilson et al showed that the combinations of MetS components that included IFG were associated with an approximately 2-fold higher risk of diabetes than other combinations with the same number of MetS components. 5 Although it is not feasible to directly compare the relative risk between that study and ours due to the different definition of the comparison group, the results of both studies support the importance of considering the glucose level separately from other MetS components in stratifying diabetes risk.
In the present study, the combinations that included central obesity were associated with a 23e74% higher risk of diabetes than others with the same number of MetS components within the same glucose status, which is consistent with the results of Wilson's study, 5 in which the corresponding value was 17e61%. Similarly, the Niigata Wellness study in Japan showed that the combinations that included central obesity were associated with a 41e134% higher risk of diabetes than others with the same number of MetS components. 10 These findings suggest that there might be a combined effect of central obesity and other metabolic abnormalities on the development of diabetes among those with the same number of MetS components within the same glucose status. Taken together with our overall results, as shown in Fig. 1 and Table 2, diabetes risk was largely determined via IFG and the number of MetS components, and the difference in diabetes risk was small among groups that had the same fasting glucose status and the same number of MetS components.
The major strengths of the present study included the enrollment of a large number of participants from several Japanese companies. The limitations of the present study should also be mentioned. First, the follow-up period was relatively short. The incidence rate and relative risk in the present study, however, are comparable to those in a long-term prospective study in Japan. 9,10 Second, the definition of MetS components was based on a single measurement at baseline. Because specific health checkups and specific counseling focusing on MetS have been conducted since 2008 in Japan, 17 individuals with MetS at baseline might have been instructed to modify their lifestyle. If so, risk estimates would be attenuated towards null. Third, potential confounding variables, including dietary intake and physical activity, were not assessed in this study. Finally, most subjects in the present study were middleaged employees of large companies, so our findings may not be generalized to employees of small-to medium-sized companies, the self-employed, and the unemployed. Give that age-specific prevalence of diabetes in the present study population was similar to that in a nationally-representative sample, 12,18 however, we speculate that the association observed in the present study may be replicated in the general population.
In summary, the present study indicates that considering the presence or absence of IFG and the number of MetS components can detect individuals with a high risk of future diabetes. Among individuals with the same number of MetS components within the same glucose status, the combinations that included central obesity might identify a high-risk population of diabetes.

Conflicts of interest
None declared.