Serum Total Homocysteine Concentrations and Risk of Stroke and Its Subtypes in Japanese

Elevated levels of homocysteine, either fasting or nonfasting, have been found to be more prevalent among patients with stroke than among control subjects.^1,2 However, this finding could reflect an effect rather than the cause, because homocysteine concentrations may increase after the onset of acute stroke.³ Thus, to reduce this bias, data from prospective studies are desirable. However, the findings from prospective studies have been inconsistent. Several prospective studies of white subjects showed a significant association between homocysteine and the risk of total^4–8 and ischemic strokes,⁸ whereas others showed no association.^9,10 There have been no prospective studies of the effect of homocysteine on the risk of stroke in Asian populations.¹¹ Furthermore, only one prospective study among whites has reported the association between homocysteine levels and the age- and sex-adjusted risk of stroke subtypes; that study suggested that hyperhomocysteinemia was more predictive of risk for lacunar infarction than other stroke subtypes.⁸ Because moderately elevated homocysteine concentrations are common in general populations of Japan as well as in Western countries,^12,13 it is important to examine the relation of homocysteine levels and the risk of total stroke and stroke subtypes among Japanese men and women.

We hypothesized a priori that hyperhomocysteinemia increases the risk of ischemic stroke, more specifically lacunar infarction, but not the risk of hemorrhagic stroke among Japanese men and women. To test this hypothesis, we conducted a prospective, nested, case-control study of men and women in three Japanese communities by using stored serum samples.

Methods

Surveyed Populations

The surveyed populations comprised 11 846 men and women 40 to 85 years of age who participated in cardiovascular risk surveys between 1984 and 1995 in a central rural community (Kyowa; the participants and the census population for ages 40 to 85, n=5952 and n=8037, respectively) and between 1989 and 1995 in a northeast rural community (Ikawa; n=2502 and 3295, respectively) and a southwest rural community (Noichi; n=3392 and 7083, respectively). The participation rate in cardiovascular risk surveys among men and women 40 to 85 years of age was 74% in Kyowa, 76% in Ikawa, and 48% in Noichi and 64% for the total population. A 1.0- to 2.0-mL serum sample obtained from each participant was stored at −80°C for 1 to 17 years (average, 9.0 years). Participants with a history of stroke or coronary heart disease (n=232) were excluded from the analyses. The Ethics Committee of the University of Tsukuba approved this study.

Surveillance of Stroke and Classification of Stroke Subtypes

The participants were followed up to determine incident strokes occurring by the end of 1999. The follow-up was conducted by annual cardiovascular risk surveys to obtain histories of incident strokes; for nonparticipants, the confirmation of stroke was achieved by mailing a questionnaire to the nonparticipants and by the use of death certificates. From death certificates, cases with stroke as the underlying cause of death (ICD 9 classification: 430–438) were selected. We also used national insurance claims, ambulance records, reports by local physicians, and reports by public health nurses and health volunteers for possible stroke identifiaction.¹⁴ To confirm the diagnosis, all living patients were visited or invited to take part in risk factor surveys to obtain medical history, and/or if cases were still alive, neurological examinations by study physicians, and their medical records were reviewed. For deaths, histories were obtained from families, and medical records were reviewed.

Stroke was defined as a focal neurological disorder with rapid onset, which persisted at least 24 hours or until death and was confirmed by CT and/or MRI.¹⁵ A diagnosis of embolic infarction was made when evidence of an embolic source was present in the medical records and if imaging studies and a neurology consultation supported the diagnosis. Classification of other stroke subtypes (large-artery occlusive infarction, lacunar infarction, subarachnoid hemorrhage, and intraparenchymal hemorrhage) was based on imaging studies. Strokes with negative findings on imaging studies and unclassified strokes were not included in the present study. For each new case of stroke, 3 control subjects were selected randomly from the participants with no incident stroke, matched for sex, age (±2 years), community, year of serum storage, and fasting status at serum collection (<8 and ≥8 hours).

Determination of Serum Total Homocysteine

Nonfasting venous blood was collected in a 7- to 10-mL plain tube and allowed to stand for <30 minutes and then centrifuged at 4°C at 1500g for 10 minutes for serum separation. The serum samples were aliquoted immediately and placed on dry ice at survey sites and then stored at −80°C until analysis. These procedures prevent an artificial rise in serum homocysteine concentrations caused by the escape of homocysteine from erythrocytes.¹⁶ Total homocysteine was determined according to a previously reported method,¹⁶ with the following modifications.¹³ Separation was performed in an ERC-ODS-1171 (6×200 mm) column with the use of a high-performance liquid chromatography system (HITACHI L-7600). Different compounds were eluted with isocratic elution buffer, consisting of 0.1 mol/L potassium dihydrogen phosphate and 2.0% acetonitrile (pH 6.5), at a flow rate of 1 mL/min. The standard for homocysteine was prepared in 2-mmol EDTA solution. The interassay coefficient of variation was 2.9% (n=10), and the intra-assay coefficient of variation was 4.1% (n=10).

Determination of Confounding Variables

An interview was conducted to ascertain history of cigarette smoking, alcohol intake, medication use for high blood pressure, and high serum glucose levels. Height in stocking feet and weight in light clothing were measured. Body mass index (BMI) was calculated as weight (kg)/height (m²).

Systolic and diastolic blood pressures were measured by trained observers using a standard mercury sphygmomanometer on the right arm of seated participants after a 5-minute rest. Hypertension was defined as systolic blood pressure ≥160 mm Hg and/or diastolic blood pressure ≥95 mm Hg and/or taking antihypertensive medication; normotension was defined as systolic blood pressure <140 mm Hg and diastolic blood pressure <90 mm Hg and not taking antihypertensive medication. All others were classified as having borderline hypertension.

Serum total cholesterol, triglycerides and glucose were measured by enzymatic methods (SMAC, Technicon Instrument Corp). The measurement of serum lipids was standardized by the Lipid Standardization Program, Center for Disease Control, Atlanta, Ga.¹⁷ Serum high-sensitivity C-reactive protein was measured by the immunonephelometric assay on a BN ProcSpec analyzer (Dade Behring), and serum creatinine was measured by Jaffe reaction method. Serum glucose was measured by the hexokinase method. Impaired glucose tolerance was defined as a fasting glucose of 6.1 to 6.9 mmol/L and/or a nonfasting glucose level of 7.8 to 11.0 mmol/L, without medication use for diabetes. Diabetes was defined as a fasting glucose level of ≥7.0 mmol/L and/or a nonfasting glucose level of ≥11.1 mmol/L and/or use of medication for diabetes.

Statistical Analysis

The paired Student’s t test was used to compare the mean values of baseline cardiovascular risk factors and log-transformed serum total homocysteine levels between incident cases and control subjects. The χ² test was used to compare proportions between cases and control subjects. The multiple linear regression analysis was used to examine associations between total homocysteine levels and potential confounding factors listed below. The odds ratios of total stroke and stroke subtypes were estimated according to quartiles of serum total homocysteine with conditional logistic regression models. Adjustments for hypertension status (normal, borderline, and hypertension), BMI (kg/m²), current alcohol intake (g/d), cigarette smoking status (never, ex-smoker, and current), serum total cholesterol levels (mmol/L), log-transformed triglyceride levels (mmol/L), quartiles of high-sensitivity C-reactive protein levels, and serum glucose category (normal, impaired glucose tolerance, and diabetes) were also conducted. The significance of the interaction of homocysteine with age, sex, smoking, and hypertension status for total and ischemic stokes was tested by using interaction terms of 3 categorical variables by dichotomous homocysteine variables in the multivariate models. All probability values for statistical significance were 2 tailed, and all confidence intervals were estimated at the 95% level.

Results

During the follow-up period, we identified 150 incident strokes, comprising 52 hemorrhagic strokes (38 intraparenchymal hemorrhages and 14 subarachnoid hemorrhages) and 98 ischemic strokes (75 lacunar infarctions, 19 large-artery occlusive infarctions, and 4 embolic infarctions). Table 1 shows the risk characteristics of total stroke and each stroke subtype compared with control subjects. The results for embolic infarctions are omitted because of the small number of cases (n=4). The average age was 65 years for total stroke, varying from 63 years for subarachnoid hemorrhage to 69 years for large-artery occlusive infarction. The proportion of men was 53% for total stroke, varying from 21% for subarachnoid hemorrhage to 68% for large-artery occlusive infarction. Systolic and diastolic blood pressure levels and the prevalence of hypertension were higher in patients with stroke than in control subjects; this trend was most evident for hemorrhagic stroke, more specifically intraparenchymal hemorrhage. Alcohol intake tended to be higher in cases than in control subjects for total stroke and other stroke subtypes. Mean serum cholesterol levels tended to be lower in patients with intraparenchymal hemorrhages than in control subjects but were not different between cases and control subjects for total stroke and other stroke subtypes. Mean values of BMI, triglycerides, C-reactive protein, and serum creatinine and the prevalence of smoking were not different between cases and control subjects for total stroke or stroke subtypes. The prevalence of impaired glucose tolerance and diabetes was higher in cases than control subjects, except for hemorrhagic stroke.

Geometric mean values of serum total homocysteine were 0.7 to 0.8 μmol/L higher in total strokes and in ischemic strokes than in control subjects but did not differ significantly between cases and control subjects for other stroke subtypes (Table 2). The proportion of cases with homocysteine ≥11.0 μmol/L was 2-fold higher in cases than in control subjects for total stroke and stroke subtypes other than subarachnoid hemorrhage.

We examined associations between total homocysteine levels and stroke risk factors in control subjects to elucidate potential confounding factors. According the multiple linear regression analysis, homocysteine levels were positively associated with age, male sex, hypertension, smoking, and serum creatinine and inversely associated with impaired glucose intolerance (not shown in the table).

Table 3 shows univariate and multivariate odds ratios (95% CIs) for total stroke and stroke subtypes according to quartiles of total homocysteine and odds ratios associated with a 5-μmol/L increment of total homocysteine. Compared with individuals in the lowest homocysteine quartile, individuals in the highest quartile had an ≈3-fold higher incidence of total stroke. The excess risk was particularly evident for ischemic strokes, more specifically, lacunar infarction, but not for hemorrhagic stroke, either intraparenchymal or subarachnoid hemorrhage. The further adjustment for serum creatinine concentrations strengthened the associations; the multivariate odds ratios were 3.71 (1.79 to 7.68) for total stroke, 5.29 (2.00 to 14.0) for ischemic stroke, 5.03 (1.75 to 14.5) for lacunar infarction, and 2.36 (0.59 to 9.48) for hemorrhagic stroke (not shown in the table).

The multivariate odds ratios of total and ischemic strokes for higher versus lower homocysteine levels (≥11.0 versus <11.0 μmol/L) were examined after stratification for age, sex, smoking, and hypertension status (Table 4). The excess risks of total and ischemic strokes were similarly observed between younger and older age groups, between men and women, between nonsmokers and smokers, and for total stroke between nonhypertensive and hypertensive subjects. The excess risk of ischemic stroke was primarily observed among nonhypertensive subjects, although the interaction of homocysteine with hypertension for ischemic stroke was far from being statistically significant.

Discussion

The major finding of the present study was that moderately elevated concentrations of nonfasting serum total homocysteine were strongly associated with increased incidence of total stroke, ischemic stroke, and, more specifically, lacunar infarction among Japanese: The stroke risk was ≈3-fold higher among persons in the highest quartile of homocysteine values than among those in the lowest quartile values. These associations did not alter materially after adjustment for known cardiovascular risk factors, including smoking and hypertension. There was no significant association between homocysteine level and risk of hemorrhagic stroke. However, effects of elevated homocysteine on risks of intraparenchymal and subarachnoid hemorrhages separately were uncertain because of the limited number of cases.

To our knowledge, the present study is the first to demonstrate a significant association between total homocysteine levels and subsequent risk of total and ischemic strokes in an Asian population. The magnitude of the association between homocysteine and risk of total stroke was consistent with the results from previous prospective studies of whites.^5–8 The multivariate odds ratio of total stroke for the highest versus lowest quartiles of homocysteine in our study was 2.99 (1.51 to 5.93), whereas the corresponding odds ratios were 4.7 (1.1 to 20.0), 2.24 (1.04 to 5.75), 2.53 (1.19 to 5.35), and 1.82 (1.14 to 2.91) in the previous studies of whites.^5–8 The multivariate odds ratio of total stroke associated with a 5-μmol/L increment of homocysteine was 1.40 (1.09 to 1.80), whereas the corresponding summary odds ratio of total stroke, estimated from 5 previous nested case-control studies, was 1.60 (1.40 to 1.83).¹¹

The excess risk of ischemic stroke associated with moderately elevated homocysteine levels did not vary significantly with age or smoking status but was primarily observed among men and nonhypertensive subjects. Several previous studies showed the stronger association of homocysteine with carotid atherosclerosis^18,19 and stroke risk¹⁰ among nonhypertensive subjects than among hypertensive subjects. It is possible that the risk of ischemic stroke associated with hyperhomocysteinemia may be masked by the presence of hypertension, as hypertension is the predominant risk factor for stroke.

The precise mechanisms underlying the apparent adverse effect of hyperhomocysteinemia on the risk of ischemic stroke are not clear at present, although several possibilities can be proposed. Hyperhomocysteinemia may cause a rise in arterial blood pressure,¹² thereby increasing the risk of ischemic stroke. In the present study, a 5-μmol/L higher homocysteine level was associated with 2.6–mm Hg (95% CI, 0.5 to 4.7, P=0.02) higher systolic and 1.1–mm Hg (95% CI, −0.3 to 2.5, P=0.13) higher diastolic blood pressure levels after adjustment for age, sex, BMI, alcohol intake, and use of antihypertensive medication. The positive associations between serum homocysteine and risk of total and ischemic strokes, however, were not explained by a blood pressure–raising effect, as the adjustment for blood pressure category did not eliminate the association.

Elevated total homocysteine induces oxidative injury to vascular endothelial cells and impairs the production of nitric oxide, a strong vascular relaxing factor, from the endothelium.^19,20 Hyperhomocysteinemia also enhances platelet adhesion to endothelial cells,²¹ promotes growth of vascular smooth muscle cells,²² and is associated with higher levels of prothrombotic factors such as β-thromboglobulin, tissue plasminogen activator, and factor VIIc.²³

The strong association observed in our study between homocysteine levels and risk of lacunar infarction should be noted. This finding is concordant with previous findings from clinical studies that homocysteine levels were higher in patients with subcortical vascular encephalopathy, a type of dementia with subcortical diffuse white-matter lesions or multiple subcortical lacunae,²⁴ and patients with silent brain infactions²⁵ than in control subjects. These clinical manifestations involve the cerebral microvascular system. Hyperhomocysteinemia also increases the risk of dementia and Alzheimer’s disease,²⁶ but it remains unknown whether the effects of homocysteine on the risk of lacunar infarction and dementia are manifested by common or different mechanisms.

It is possible that elevated homocysteine concentration is simply a marker of systematic inflammation. However, this possibility is unlikely because the adjustment for C-reactive protein, a sensitive marker of inflammation, did not alter substantially the association between homocysteine and risk of stroke.

The strength of the present study is the large number of strokes confirmed by imaging studies, which allowed us to investigate the relation between total homocysteine levels and risk of total stroke as well as stroke subtypes. We found a strong association between elevated homocysteine and risk of lacunar infarction after adjustment for known cardiovascular risk factors, which extended the evidence from a previous study.⁸

A potential limitation of our study is that we used frozen serum to estimate homocysteine concentrations and we did not examine long-term changes in homocysteine in stored serum samples. However, homocysteine concentrations were reported to be stable at −20°C for 10 years.⁴ Mean values of homocysteine were similar to those reported in our recent study in which the samples were preserved at −80°C for only 3 months until the analysis.¹³

Second, we did not measure nutritional status, that is, serum concentrations of folate and vitamins B₆ and B₁₂, which affect homocysteine metabolism.¹³ Thus, it remains uncertain whether nutritional status affected the risk of ischemic stroke. In this regard, our previous cross-sectional study of Japanese showed that low serum concentrations of these vitamins, in particular vitamin B₁₂ and folate, were strongly associated with high homocysteine concentrations.¹³

Third, the generalizability of our present data to Western countries is unknown. The proportional representation of stroke subtypes in the present study was as follows: hemorrhagic strokes, 35%; lacunar infarctions, 50%; large-artery occlusive infarctions, 12%; and embolic infarctions, 3%. This was similar to previous reports among Japanese subjects, whereas the respective proportions in studies from Western countries were approximately 15% to 20%, 15% to 25%, 50% to 60%, and 10%.²⁷ Because ischemic stroke mostly comprises lacunar infarctions among Japanese subjects and large-artery occlusive infarction among white subjects, the present study reflects the predictive value of serum total homocysteine for the risk of lacunar stroke. The potential effect on large-artery occlusive infarction was uncertain because of the limited number of cases.

In conclusion, our observational study suggests that an elevated serum total homocysteine level may increase the risk of ischemic stroke and lacunar infarction. Because dietary supplementation of folate or combined supplementation with folate and vitamin B₁₂ effectively reduces plasma homocysteine levels,²⁸ our finding suggests the potential importance of dietary intake of folate and vitamin B for the prevention of ischemic stroke. A clinical trial is necessary to confirm the causality between these vitamin intakes and the risk of ischemic stroke.

Footnotes