Diabetes Mellitus and Risk of Stroke and Its Subtypes Among Japanese
The Japan Public Health Center Study
Background and Purpose—Although diabetes mellitus has been recognized as a risk factor for ischemic stroke, evidence is limited for ischemic stroke subtypes among Asians.
Methods—A survey was conducted of a total of 13 129 men and 22 528 women aged 40 to 69 years at baseline. During the median 12.0-year follow-up, there were 526 incidences of ischemic strokes (259 lacunar infarctions, 91 large-artery occlusive, and 140 embolic infarctions).
Results—The risk of ischemic stroke for both sexes was approximately 2 to 4 times higher for diabetic subjects than for those with normal glucose levels. Significant excess risks of lacunar and embolic infarction for both sexes and of large-artery occlusive infarctions for women were also observed in diabetic subjects. Diabetes mellitus was not associated with risk of intraparenchymal or subarachnoid hemorrhage for either sex.
Conclusions—Diabetes mellitus is a significant risk factor for ischemic stroke and all subtypes of ischemic stroke for middle-aged Japanese.
Prospective studies have demonstrated that diabetes mellitus is a risk factor for ischemic stroke.1–4 Furthermore, the Atherosclerosis Risk in Communities Study has reported that diabetes mellitus was associated with increased risk of large-artery occlusive infarction.2 Although several Japanese studies have examined such associations for infarction subtypes such as lacunar and embolic infarction,3,4 none of them have examined the association of diabetes mellitus with risk of large-artery occlusive infarction due to the small number of cases among Japanese.4 The probable reason for this small number is that the prevalence of diabetes mellitus and the proportion of large-artery occlusive infarction among stroke cases in Asians is lower than among other ethnic groups2–5 so that the statistical power of the study population would be too low for detecting an association between diabetes mellitus and risk of stroke subtypes.
We therefore used the findings of the Japan Public Health Center-based prospective (JPHC) Study, 1 of the largest cohort studies for Asians to investigate the relationship between serum glucose levels and risk of stroke and its subtypes among Japanese.
The JPHC Study was initiated in 1990 (Cohort I) and 1993 (Cohort II) in 9 public health centers.6 Serum glucose data from the baseline survey were available for 10 560 men and 19 241 women aged 40 to 69 years and for another 3098 men and 4168 women from the 5-year follow-up survey of Cohort I in 1995 and of Cohort II in 1998. Excluded from analysis were persons who reported having cardiovascular disease (n=565) or cancer (n=845) at baseline. The subjects were followed up until the end of 2004. The study protocol was approved by the Institutional Review Board of the National Cancer Center, Tokyo, Japan.
Baseline Survey and Confirmation of Stroke Incidence
The study protocols and stroke surveillance have been described in detail elsewhere.6,7 Diabetes mellitus was defined as a fasting glucose level of ≥7.0 mmol/L or a nonfasting glucose level ≥11.0 mmol/L or use of medication for diabetes mellitus with corresponding values for normal glucose level of <6.1 mmol/L or <7.8 mmol/L and for borderline diabetes mellitus of 6.1 to 6.9 mmol/L or 7.8 to 11.0 mmol/L.
For both fatal and nonfatal, stroke was confirmed by medical records according to the criteria of the National Survey of Stroke, which requires a constellation of neurological deficits of sudden or rapid onset lasting at least 24 hours or until death. Stroke events were classified as intraparenchymal hemorrhage, subarachnoid hemorrhage, ischemic stroke (lacunar, large-artery occlusive, and embolic infarction) primarily based of CT and/or MRI (97%).7 Only confirmed strokes were used for the analyses.
The hazard ratios (HRs) and 95% CIs for strokes according to serum glucose categories were calculated by using the Cox proportional hazards model. The confounding variables comprised age (year), smoking status (never, ex-, and current smoking: 1 to 19, 20 to 29, and ≥30 cigarettes/day), alcohol intake (never, ex-, and current drinking: ethanol intake <150, 150 to 299, 300 to 449, ≥450 g/week), quartiles of body mass index (kg/m2), systolic blood pressure (a single reading, mm Hg), quartiles of serum total and high-density lipoprotein cholesterol levels and triglycerides (mmol/L), antihypertensive medication use (yes), fasting status (<8 and ≥8 hours), and residential areas. SAS Version 9.13 was used for the statistical analyses (2-tailed).
Table 1 shows baseline characteristics according to serum glucose categories. Higher glucose categories were positively associated with higher body mass index, blood pressure, antihypertensive medication use, total cholesterol, triglycerides, alcohol intake, and smoking, whereas they were inversely associated with high-density lipoprotein cholesterol levels for both men and women.
During the median 12.0-year follow-up, there were 904 incidences of strokes, including 378 hemorrhagic and 526 ischemic strokes (comprising 259 lacunar infarctions, 91 large-artery occlusive, and 140 embolic infarctions).
Patients of both sexes with diabetes mellitus had an approximately 2 to 4 times higher risk of ischemic stroke than did normal subjects of both sexes. Diabetes mellitus was also associated with higher risks of lacunar and embolic infarctions for both men and women and with higher risk of large-artery occlusive infarction for women (Table 2).
We combined the data of men and women because the associations between diabetes mellitus and risk of stroke and its subtypes did not vary by sex (P for interaction >0.13). The multivariable HRs (95% CI) associated with diabetes mellitus were 2.65 (2.04 to 3.44) for ischemic stroke, 2.65 (1.82 to 3.87) for lacunar infarction, 2.58 (1.41 to 4.72) for large-artery occlusive infarction, and 3.32 (2.02 to 5.14) for embolic infarction (not shown in Table 2). These HRs (95% CI) did not vary by fasting status. The multivariable HRs based on nonfasting and fasting blood samples, respectively, were 2.32 (1.69 to 3.19) and 1.49 (1.06 to 2.08) for total stroke, 0.98 (0.46 to 2.10) and 0.49 (0.21 to 1.12) for hemorrhagic stroke, 3.08 (2.15 to 4.41) and 2.30 (1.57 to 3.37) for ischemic stroke, 3.16 (1.91 to 5.23) and 2.59 (1.45 to 4.63) for lacunar infarction, 3.28 (1.43 to 7.53) and 1.94 (0.80 to 4.71) for large-artery occlusive infarction, and 2.98 (1.44 to 6.17) and 3.32 (1.77 to 6.24) for embolic infarction. Furthermore, the multivariable HRs (95% CI) associated with a 1-SD increment of fasting glucose levels was 1.19 (1.09 to 1.30) for total stroke, 0.93 (0.75 to 1.16) for hemorrhagic stroke, 1.30 (1.19 to 1.42) for ischemic stroke, 1.38 (1.24 to 1.54) for lacunar infarction, 1.20 (0.92 to 1.57) for large-artery occlusive infarction, and 1.32 (1.12 to 1.56) for embolic infarction.
We confirmed that subjects of both sexes with diabetes mellitus had an approximately 2- to 3- fold higher risk of ischemic stroke than did those with normal glucose levels,1–4 and extended the evidence that diabetes mellitus was associated with risk of large-artery occlusive infarction among Asians.
The excess risk of embolic infarction for diabetic subjects in our study was consistent with that reported by the Atherosclerosis Risk in Communities Study,2 partly due to atrial fibrillation. Higher plasma glucose levels were associated with an increased risk of atrial fibrillation,8 and atrial fibrillation is a major risk factor for embolic infarction.3
The association between diabetes mellitus and risk of ischemic stroke tended to be stronger in women than in men, although the sex interaction did not reach statistical significance. This finding was consistent with a previous study conducted in Japan, which reported a multivariable HR (95% CI) of ischemic stroke for diabetes mellitus of 1.8 (1.0 to 3.2) for men and 2.2 (1.2 to 4.0) for women,4 and a recent large meta-analysis of 102 prospective studies, which found a multivariable HR of 2.2 (1.8 to 2.5) for men and 2.8 (2.4 to 3.4) for women.1 However, these studies could not offer a satisfactory explanation for this sex difference.
The strengths of this study are the large population size and standardized measurements of cardiovascular risk factors in addition to confirmation of stroke incidence based on imaging studies.7
The primary limitation is a dilution bias due to the single glucose measurement and the use of an epidemiological cut point for diabetes mellitus based on casual blood glucose levels than oral glucose tolerance testing,9,10 both of which may have led to the underestimation of diabetes–stroke associations. We examined the association between continuous fasting glucose levels and risk of ischemic stroke and still found a significant positive association. Furthermore, the prediction of ischemic stroke by the presence of diabetes mellitus did not differ whether nonfasting or fasting glucose levels are used.
In conclusion, our findings provide evidence that diabetes mellitus is associated with increased risk of ischemic stroke and all subtypes of ischemic stroke for middle-aged Japanese.
Sources of Funding
This study was supported by grants-in-aid for Cancer Research and for the Third Term Comprehensive Ten-Year Strategy for Cancer Control from the Ministry of Health, Labor and Welfare of Japan.
We thank all staff members in each study area and in the central office for their painstaking efforts to conduct the baseline survey and follow-up.
- Received January 27, 2011.
- Revision received March 8, 2011.
- Accepted March 11, 2011.
- © 2011 American Heart Association, Inc.
- Ohira T,
- Shahar E,
- Chambless LE,
- Rosamond WD,
- Mosley TH Jr.,
- Folsom AR
- Tanizaki Y,
- Kiyohara Y,
- Kato I,
- Iwamoto H,
- Nakayama K,
- Shinohara N,
- et al
- McNeely MJ,
- Boyko EJ
- Psaty BM,
- Manolio TA,
- Kuller LH,
- Kronmal RA,
- Cushman M,
- Fried LP,
- et al
- Modan M,
- Harris MI
- Hyvärinen M,
- Tuomilehto J,
- Mähönen M,
- Stehouwer CDA,
- Pyörälä K,
- Zethelius B,
- et al