Is the Diagnosis of Metabolic Syndrome Useful for Predicting Cardiovascular Disease in Asian Diabetic Patients?: Analysis from the Japan Diabetes Complications Study

The Japan Diabetes Complications Study (JDCS) is a nationwide, multicenter, prospective study of type 2 diabetic patients (38). In 1996, 2,205 patients aged 40–70 years with previously diagnosed type 2 diabetes and HbA_1c levels >6.5% were recruited and registered. The eligibility criteria for participating patients has been previously described (38). The duration of the study was 8 years. Of the 2,205 patients, the present study focused on 1,424 patients (771 men and 653 women) who had a complete set of data, including those parameters necessary to satisfy the WHO (2) and NCEP (3) criteria for the definition of MetS at baseline. The JDCS protocol, which is in accordance with the Declaration of Helsinki, received ethical approval from the institutional review boards of all of the participating institutes and was undertaken in accordance with the Ethical Guidelines for Clinical Studies of the Japanese Ministry of Health, Labor, and Welfare. All of the study participants gave written informed consent.

Both the WHO (2) and the NCEP (3) definitions were used to diagnose MetS in this study. However, because the original cut-off for abdominal obesity in the NCEP definition (waist circumference ≥102 cm for men and ≥89 cm for women) has previously been shown to be inappropriate for Asian populations (35,37) and the number of subjects who met these criteria was extremely low, the cut-off limit was adjusted according to the criteria proposed by the Japan Society for the Study of Obesity (≥85 cm for men or ≥90 cm for women), which were based on the risk of obesity-related disorders in a Japanese population (39). The WHO criteria for obesity were adopted because the waist-to-hip ratio (WHR) was used rather than waist circumference. The criteria used for analysis in this study are shown in Table 3. Because all of the study subjects were diabetic, those who fulfilled two or more of criteria 1a, 2a, 5, or 6 were classified as having WHO-MetS and those who fulfilled two or more of criteria 1b, 2b, 3, or 4 were diagnosed as having NCEP-MetS, using a modified NCEP definition (Table 3). For comparisons with other traditional risk factors for CVD, we also evaluated high LDL cholesterol levels, cigarette smoking, and excessive alcohol intake (40). Medication use, including agents for hypertension and hyperlipidemia, were not considered when diagnosing MetS in this study.

Waist and hip circumferences were measured at the umbilicus and trochanter level, respectively. A baseline dietary survey, comprised of food records and a food frequency questionnaire that included alcohol consumption, was undertaken. Information regarding cigarette smoking was collected using a standardized questionnaire. All laboratory tests were undertaken using the standard methods of each of the participating institutes, apart from the HbA_1c assays, which used a common standard, with 5.8% as the upper normal limit. Plasma LDL cholesterol was calculated using Friedewald’s equation, except for triglyceride levels >400 mg/dl, in which case the LDL cholesterol data were treated as “missing.” To estimate insulin resistance, the homeostasis model assessment of insulin resistance (HOMA-IR) was used (41). Plasma insulin levels and the HOMA-IR were not evaluated in patients treated with insulin.

Patients were assessed for CHD and stroke at baseline and yearly thereafter. In all subjects, a 12-lead electrocardiogram (ECG) was recorded at each assessment. Fatal and nonfatal CHD and stroke events identified during follow-up were certified by at least two members of the experts’ committee who were masked as to risk factor status and the other member’s diagnosis. With regard to CHD, myocardial infarction was defined according to the WHO Monitoring of Trends and Determinants in Cardiovascular Disease criteria (42) and angina pectoris was defined as typical effort-dependent chest pain or oppression relieved at rest or by using nitroglycerine, as validated by exercise-positive ECG and/or angiography. Stroke events were defined as a constellation of focal or global neurological deficits of sudden or rapid onset and for which there was no apparent cause other than a vascular accident, as determined by a detailed history, a neurological examination, and ancillary diagnostic procedures such as computed tomography, magnetic resonance imaging, cerebral angiography, and lumbar puncture. Stroke events were classified as cerebral infarction (including embolus), intracranial hemorrhage (including subarachnoid hemorrhage), transient ischemic attack, or stroke of undetermined type in accordance with WHO criteria (43). No cases of asymptomatic lesions detected by brain imaging (i.e., silent infarction) were included. Only “first-ever” CHD or stroke events during the study period were counted for the analysis; if a patient had both CHD and stroke events, each event was counted separately.

Data are presented as means ± SD or as a proportion, unless otherwise specified. To compare the distributions of baseline characteristics between groups, Wilcoxon’s rank-sum test or Fisher’s exact test was used. Incidence rates in the two groups were assessed by a score test under the Poisson assumption. Cox regression analysis was used to calculate the adjusted hazard ratio (HR) and 95% confidence interval (CI) of MetS risk factors with CHD, stroke, or both. Statistical analyses were performed separately by sex. The SAS software package (Version 8.0, Cary, NC) was used for all analyses. P < 0.05 was considered to be significant.

The baseline characteristics of the study subjects are shown in Table 1. In all, 51% of male and 53% of female subjects met WHO criteria for MetS, whereas 45% of male and 38% of female subjects met NCEP criteria for MetS. Plasma insulin levels and HOMA-IR were significantly higher in patients with MetS (both definitions) than in those without MetS; however, there were no significant differences in HbA_1c or the frequency of oral hypoglycemic agent use. Insulin usage was significantly lower in women with MetS by either definition and in men with NCEP-MetS. Blood pressure and serum triglycerides were significantly higher and HDL cholesterol was significantly lower in MetS patients, despite the fact that the use of medications for both hypertension and hyperlipidemia was much more common than in patients without MetS. Daily energy intake did not differ between patients with and without MetS (data not shown).

During the 8-year study period, the total number of CVD events was 117, comprised of 62 CHD and 59 stroke events. The combined incidence (per 1,000 patient-years) of CHD and/or stroke was significantly greater in patients with MetS (except in female patients with NCEP-MetS) than in those without MetS (Table 2).

HRs were calculated to determine which definition of MetS was the better predictor of CVD and which of the individual MetS components (or other classic risk factors) could most efficiently predict CVD events in our subjects (Table 3). In male patients, WHO-MetS was not significantly associated with an increased risk for either CHD or stroke separately, but was associated with the combination of both (HR = 1.6). Triglyceride, LDL cholesterol (both for CHD), and blood pressure (≥140/90 mmHg) levels (for stroke) showed higher HRs. NCEP-MetS was a significant predictor of CHD in male patients, although its HR (1.9) was lower than that for triglycerides (2.9) or LDL cholesterol (2.1). Thus, neither definition of MetS was a substantially better predictor of CVD than the component parts in male patients. In contrast, in the female patients, WHO-MetS was a significant and strong predictor of CHD (HR = 2.8), stroke (HR = 3.7), and both CHD and stroke (HR = 3.2). In female patients, none of the individual elements nor the other classic risk factors showed significant increases in HRs, with the exception of hypertension (≥140/90 mmHg) for stroke, although its HR (2.4) was still lower than that for WHO-MetS. NCEP-MetS was not a significant risk factor for CHD or stroke in female patients (Table 3).

To examine the clustering effects of the individual components of MetS, the association between CVD risk and the number of MetS components fulfilled (other than diabetes) was analyzed (Table 3). Increasing the cut-off component number for the diagnosis of NCEP-MetS from ≥2 to ≥3 in male subjects did not dramatically improve the HR but did greatly reduce the number of patients diagnosed as having MetS, from 45 to 14.5% (Table 3). In female patients, changing the diagnostic cut-off component numbers was not particularly beneficial in improving the prognostic value of WHO-MetS (Table 3).

The prevalence of MetS in our diabetic patients who were free from CVD was not as high as that reported in previous studies that included patients with previous CVD (12–20) but was nevertheless relatively high (38–53%). Although we did not have age-matched nondiabetic control subjects, the prevalence of MetS was much higher than that reported in Japanese general population workers, namely 19.5% in men and 7.9% in women (33). Hypertension and dyslipidemia are much more common in diabetic patients than in nondiabetic subjects (21), and it has been speculated that the features of MetS more easily aggregate, even in the absence of current or previous CVD, leading to the observed increase in the prevalence of MetS. On the other hand, the prevalence of NCEP-MetS in the U.S. general population age 50 years and older is 44% (20), which is relatively close to that in our Japanese diabetic patients. However, even in the U.S. (excluding Asian Americans), the prevalence of MetS in those who have a BMI range equivalent to that of Japanese subjects is not >10% (44). This implies that in the U.S., obesity has a potent impact on the prevalence of MetS, as has also been shown in a recent study (45). This is in contrast to findings in Japan, where diabetes rather than obesity may have the greater influence on the prevalence of MetS, as Japanese diabetic patients are not obese by comparison with white diabetic patients or nondiabetic Japanese subjects (31,32).

The clinical importance of MetS is related to its putative impact on CVD morbidity and mortality. Among Italian patients with type 2 diabetes, the risk for CVD was 4.9 (CI 1.2–20.7) times higher in patients with WHO-MetS than in those without it (16), which was a higher rate than that seen in our male (1.6 [CI 1.0–2.6] times) and female (3.2 [CI 1.6–6.5] times) patients. These results suggest that the clinical impact of MetS on diabetic patients varies by ethnic group. Comparing cardiovascular risk factors in our Japanese patients to those in patients in the U.K. Prospective Diabetes Study (UKPDS) (46,47), hypertension is a common and potent risk factor for stroke (Table 3) (46). By contrast, HDL cholesterol levels, hypertension, and smoking, all of which were identified as significant risk factors for CHD in UKPDS patients (47), were not associated with a significant elevation of HRs in our Japanese patients (Table 3). Instead, triglyceride levels, which were not significant in UKPDS patients (47), were a strong predictor for CHD in male Japanese patients. These findings imply that the critical therapeutic targets among the components of MetS for preventing cardiovascular complications (28) may need to be modified according to a patient’s ethnic group.

Most of the previous studies evaluating the predictive power of MetS for CVD calculated the HRs by including sex as one of the independent variables for statistical adjustment, and very few studies have analyzed CVD risk separately by sex (24). Sex is reportedly an independent predictor for CVD, with an odds ratio of 2.6, which is larger than that of age, HbA_1c, and even of MetS itself in type 2 diabetic patients (16). Our results revealed drastic differences in the HRs between sexes. In our female patients, WHO-MetS presented an increased risk for CVD events to a greater degree than could be predicted by the sum of the individual components (Table 3), whereas, in contrast, in our male patients, WHO-MetS was not even a significant risk factor for CVD. At baseline, obvious sex differences were observable in the proportion of subjects who smoked or consumed excessive alcohol, both of which were much higher in male patients. Of particular interest, the proportion of male subjects with excessive alcohol intake was at least twice as high in male patients with MetS than in those without MetS, whereas the proportion of current smokers did not differ in patients with and without MetS (Table 1). It can be speculated that excessive alcohol intake could be closely associated with MetS in male Japanese diabetic patients. Moreover, moderate alcohol intake, which has been shown to be beneficial for preventing CHD in U.S. and European diabetic patients, is not beneficial for Japanese patients (40).

Few studies have applied both the WHO and NCEP definitions of MetS to the same subjects to compare the prevalence of MetS or its predictive value for CVD. It has been reported that the prevalence of WHO-MetS is generally higher than that of NCEP-MetS in both sexes (7,12). This was confirmed in our Japanese diabetic subjects, although the difference in prevalence was not great. Regarding the predictive value of MetS, in subjects without diabetes or other cardiovascular risks, Hunt et al. (27) reported that the NCEP-MetS tended to be more predictive for cardiovascular mortality than the WHO-MetS, whereas Lakka et al. (9) reported a contrary result. In our diabetic patients, the NCEP guidelines, even modified for optimal use by Japanese subjects, were not more predictive than the WHO guidelines in female patients nor did they show excellent clinical usefulness in male patients. One possible explanation for this difference in our patients could be the hypertension cut-off used, with 140/90 mmHg in the WHO definition being a significant predictor for stroke, whereas 130/85 mmHg in the NCEP definition is not.

The strengths of our study were that 1) it is the first prospective study to determine the predictive value of MetS on CVD in Asian subjects, 2) the two most widely used definitions of MetS were applied to the same cohort for the evaluation of their clinical usefulness, and 3) the follow-up was mainly carried out in university or large general hospitals, which facilitated the reliable assessment of follow-up data and event diagnosis/records. Nevertheless, we acknowledge that the study had certain limitations: 1) Our study subjects were hospital-based patients with diabetes of a relatively long duration; therefore, we cannot make inferences beyond a similar group. 2) We analyzed both intervention (lifestyle modification through diabetes self-management care) and control (continuance of conventional care) groups of the JDCS together, although mild intervention produced only limited differences in glycemic control (0.1–0.2% in HbA_1c) as well as a lack of significant differences in known classical cardiovascular risk factors, as previously reported (38). 3) We did not consider medication use in the diagnosis of MetS in this study. 4) Mortality was not analyzed because we did not have sufficient occurrences at this stage of the study.

In conclusion, we found a high prevalence of MetS among diabetic patients with no history of CVD. For Japanese female patients with type 2 diabetes, WHO-MetS but not NCEP-MetS was predictive for CVD. In male patients, although both WHO-MetS and NCEP-MetS were somewhat predictive for CVD, hyperlipidemia or hypertension had equivalent or higher HRs for CVD and seemed to be sufficient for the prediction of CVD. We suggest that the commonly used definitions of MetS, at least in their present forms, have limited clinical usefulness for Asian diabetic patients and may need some ethnic group−specific modifications for global use.

Nobuhio Yamada (University of Tsukuba)

Yasuo Akanuma (Institute for Adult Diseases Asahi Life Foundation)

Keita Ato, Masaaki Eto, Hiroshi Ito (Asahikawa Medical College); Azuma Kanatsuka, Naotake Hashimoto, Yasushi Saito, Kazuo Takahashi, Kazuo Yagi (Chiba University); Tadami Takekoshi, Takanobu Wakasugi (Fukui Prefectural Hospital); Shigetake Toyooka (Fukui Red Cross Hospital); Yukihiro Bando (Fukui Saiseikai Hospital); Tsugihiko Nakai, Koji Oida, Jinya Suzuki (Fukui University); Yasuaki Fukumoto, Seiichi Sumi (Garatia Hostiptal); Genshi Egusa, Rumi Fujikawa, Masamichi Okubo, Kiminori Yamane (Hiroshima University); Takao Koike, Narihito Yoshioka (Hokkaido University); Motonobu Anai, Ritsuko Honda, Masatoshi Kikuchi (Institute for Adult Diseases Asahi Life Foundation); Shun Ishibashi (Jichi Medical School); Masanobu Kawakami, Kazuyuki Namai (Jichi Medical School Omiya Medical Center); Takashi Sasaki, Masami Nemoto (Jikei University); Ryuzo Kawamori, Yasushi Tanaka (Juntendo University); Toshihiko Ishida (Kagawa University); Izumi Takei (Keio University); Yoshikuni Fujita, Keiji Tanaka, Yoshihiro Yajima (Kitazato University); Hideki Kishikawa, Tetsushi Toyonaga (Kumamoto University); Shingo Komichi, Zenji Makita, Kyohei Nonaka, Kentaro Yamada (Kurume University); Naoto Nakamura, Koji Nakano (Kyoto Prefectural University of Medicine); Toyoshi Iguchi, Hajime Nawata (Kyushu University); Yasuhisa Matsushima (Matsudo City Hospital); Hideo Takahashi (Minami Akatsuka Clinic); Hiroyuki Toyoshima (Minoh City Hospital); Shoichi Akazawa, Eiji Kawasaki, Shigenobu Nagataki (Nagasaki University); Nigishi Hotta, Jiro Nakamura (Nagoya University); Kentaro Doi, Yu Harano, Yasunao Yoshimasa (National Cardiovascular Center); Yoichi Hayashi (Nihon University); Shinichi Oikawa (Nippon Medical School); Ryuzo Abe, Hiroaki Seino, Daishiro Yamada (Ohta-Nishinouchi Hospital); Mitsuru Hoshi, Takao Watarai (Osaka Koseinenkin Hospital); Masatoshi Imaizumi, Ryohei Todo (Osaka National Hospital); Keisuke Kosugi, Yasuhisa Shimizu, Yutaka Umayahara (Osaka Police Hospital); Junichiro Miyagawa, Mitsuyoshi Namba, Kaoru Takemura, Yoshimitsu Yamasaki (Osaka University); Kazuhiro Hosokawa, Kempei Matsuoka (Saiseikai Central Hospital); Junko Nakano, Hirotaka Umezu (Saiseikai Fukushima General Hospital); Akihiko Hoshino, Toshihiko Nishiyama, Tetsushi Nogami (Saisekai Kumamoto Hospital); Hideo Nunome (Saiseikai Mito Hospital); Shigehiro Katayama, Atsuhito Togashi (Saitama Medical College); Kenichi Yamada (Sakura National Hospital); Atsunori Kashiwagi, Yoshihiko Nishio (Shiga University of Medical Science); Yukio Yoshimura (Shikoku University); Tatsuhide Inoue (Shizuoka General Hospital); Masafumi Kitaoka (Showa General Hospital); Toshio Kitada, Akio Shirai, Ryoichiro Watanabe (Takeda General Hospital); Takaichi Miyagawa (Tama Minami Clinic); Yoshikazu Sakamoto, Osamu Mokuta, Ryo Okazaki (Teikyo Universiy Ichihara Hospital); Kazuma Takahashi (Tohoku University); Koji Shirai, Hiroshi Miyashita (Toho University Sakura Hospital); Akira Tanaka (Tokyo Medical and Dental University); Yoshiaki Fujita (Tokyo Metropolitan Institute of Gerontology); Hideki Ito (Tama-Hokubu Medical Center) Reiko Kawahara, Yasue Omori, Asako Sato (Tokyo Women’s Medical University); Toshio Murase, Mitsuhiko Noda, Masato Odawara (Toranomon Hospital); Masashi Kobayashi, Masaharu Urakaze (Toyama Medical and Pharmaceutical University); Hitomi Fujii, Satoshi Iimuro, Takashi Kadowaki, Sachiko Mizuno, Yasuo Ohashi, Junichi Osuga, Yasuyoshi Ouchi, Akane Takahashi (University of Tokyo); Hirohito Sone, Kamejiro Yamashita (University of Tsukuba); Ryo Kawasaki, Hidetoshi Yamashita (Yamagata University); Hisahiko Sekihara, Yasumichi Mori (Yokohama City University); Tetsuo Nishikawa (Yokohama Rosai Hospital); Hiroto Furuta, Kishio Nanjo (Wakayama Medical University).

This study was financially supported by the Ministry of Health, Labor, and Welfare of Japan, the Japan Arteriosclerosis Prevention Fund, and the Japan Heart Foundation.

We gratefully acknowledge all the patients, physicians, and staff taking part in the JDCS.