Serum Homocysteine and Risk of Coronary Heart Disease and Cerebrovascular Disease in Elderly Men

Patients

The Zutphen Elderly Study is a longitudinal investigation of risk factors for chronic diseases in elderly men. It is an extension of the Dutch contribution to the Seven Countries Study. The study started in 1960 with a cohort of 878 men then 40 to 59 years old who had lived for at least 5 years in Zutphen, in the eastern Netherlands.¹² In 1985, 555 men of the original cohort were still alive and were invited for new investigations, together with an additional random sample of 711 men of the same age group living in Zutphen and not part of the original cohort. Of those invited, 74% (939/1266) entered the study: 62 had moved or could not be reached, 109 could not be examined because of serious illness, and 156 refused. Complete information was available for 878 men 64 to 84 years old.

Baseline Assessment

The baseline examination took place between March and June 1985.¹² In brief, the body mass index was calculated as weight (kg)/height (m)² . Systolic and diastolic (Korotkoff phase V) blood pressures were measured in duplicate with a random-zero sphygmomanometer, with the men supine. Hypertension was defined as a systolic blood pressure of ≥160 mm Hg, a diastolic blood pressure of ≥95 mm Hg, and/or use of antihypertensive drugs. The diagnosis of myocardial infarction required 2 or more of the following 3 criteria: severe chest pain lasting for >20 minutes that did not disappear in rest, characteristic changes on electrocardiography, and specific enzyme elevations. Information on stroke was collected with a standardized questionnaire. All diagnoses were verified with hospital discharge data and written information from the subjects’ general practitioners and, in the case of stroke, their neurologists.

Venous blood samples were obtained in the nonfasting state. Samples were centrifuged after ≈60 minutes, which is sufficient to prevent increases in serum homocysteine resulting from ex vivo generation of homocysteine by erythrocytes.¹³ Serum was stored at −20°C and assayed in 1995. Total (free plus protein-bound) homocysteine levels are stable in serum or plasma stored for 10 years or more.^{8 14} Serum total homocysteine was measured as previously described in detail.¹⁵ The intra-assay and interassay coefficients of variation are 2.1% and 5.1%. Because the available amount of serum was limited, we performed duplicate assays in only 64 of 878 samples (mean difference, 6.0%). Serum homocysteine levels are consequently given as whole numbers. Serum total and HDL cholesterol were determined with standard methods.¹² Serum creatinine was determined with a modified Jaffé method.

Follow-Up

Information on vital status on December 31, 1994, was obtained. Information on the causes of death was obtained from Statistics Netherlands for deaths that occurred between the baseline assessment and June 1990 and from the subjects’ general practitioners for deaths that occurred thereafter. The causes of death were coded according to the 9th revision of the International Classification of Diseases (ICD) by a single physician. Because the underlying cause of death in the elderly is often difficult to determine, both the primary and the secondary causes of death were considered in the analyses. Death from coronary heart disease and cerebrovascular disease was defined by ICD codes 410 to 414 and 430 to 438. Information on the occurrence of myocardial infarction and stroke between the baseline assessment and December 31, 1994 was obtained in 1990 (at an assessment similar to the baseline assessment; see above), in 1993, and in 1995 (using the Dutch translation of the Rose Questionnaire for myocardial infarction and a standardized questionnaire for stroke). In all cases, myocardial infarction and stroke were defined as described above (see Baseline Assessment). For nonresponders, information on major chronic diseases was obtained from a standardized nonresponse questionnaire filled in by the subjects themselves or their closest relative or caregiver. “First-ever” myocardial infarction or stroke was defined as fatal or nonfatal myocardial infarction or stroke in the absence of a history of myocardial infarction or stroke at baseline. All diagnoses were verified with hospital discharge data and written information from the subjects’ general practitioners and were coded by 2 physicians.

Statistical Methods

SAS statistical programs were used for the analyses (SAS Institute Inc, 1989, version 6.08). All tests were 2-sided. Values of P<0.05 were considered statistically significant. The subjects were categorized according to tertiles of homocysteine level. Differences in their baseline characteristics were then evaluated by use of ANOVA for normally distributed variables, the Kruskal-Wallis test for variables with a skewed distribution, and an overall χ² test for categorical variables. Event rates were calculated as the number of cases divided by the sum of time periods of observation. Logistic regression analysis was used to investigate the associations with the prevalence of myocardial infarction and stroke. Cox’s proportional hazard (survival) analysis was used to investigate the associations with mortality outcomes and with the incidence of first-ever myocardial infarction and stroke. The highest tertile was compared with the 2 lower tertiles if necessary to avoid (near-) empty cells and thus the inability to estimate the relative risk (RR) reliably. Interaction terms were investigated at the 0.10 level. Three persons had moved and were lost to follow-up. The date on which they moved was used as their (censored) end-point date. Unless stated otherwise, adjusted analyses are those in which the effects of major risk factors are taken into account, ie, age, body mass index, systolic blood pressure, total and HDL cholesterol, diabetes mellitus, and cigarette smoking habits.

We chose to categorize the subjects according to tertiles of homocysteine levels, because the alternative approach, ie, survival analysis with homocysteine as a continuous variable, assumes that the relationship between homocysteine level and the risk of vascular disease is linear. It is not clear that this assumption is correct,^{1 2 3 4 5 6 7 8 9 10 16 17 18} but, for comparison, we nevertheless repeated the above analyses with homocysteine as a continuous variable.

Baseline

The mean (±SD) homocysteine level at baseline was 15.8 (±8.2) μmol/L. Figure 1⇓ shows the distribution of homocysteine levels. The prevalence of homocysteine levels ≥16, 17, and 18 μmol/L was 37.8%, 30.5%, and 25.1%. Table 1⇓ shows that higher homocysteine levels were associated with increasing age, lower levels of HDL cholesterol, higher serum creatinine levels, current smoking, and a history of myocardial infarction and stroke. The associations with myocardial infarction and stroke remained after adjustment for major risk factors (see Statistical Methods). For myocardial infarction, the adjusted odds ratios (ORs) for the second and third tertiles compared with the first were 1.28 (95% confidence interval [CI], 0.76 to 2.15) and 1.81 (95% CI, 1.07 to 3.08; P for trend, 0.03). For stroke, these ORs were 1.74 (95% CI, 0.63 to 4.83) and 4.61 (95% CI, 1.79 to 11.89; P for trend, 0.002).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Distribution of homocysteine levels in 878 elderly men.

Follow-Up

For coronary heart disease, the mortality rate per 100 person-years was 1.45 (98/878). It was highest in subjects with homocysteine levels in the highest tertile (Table 2⇓; Figure 2A⇓). This risk was reduced after adjustment (P for trend, 0.09) and decreased further when we included a history of myocardial infarction at baseline as a potential confounder in the analysis (adjusted RRs for the second and third tertiles compared with the first, 1.11 [95% CI, 0.66 to 1.87] and 1.42 [95% CI, 0.83 to 2.41]; P for trend, 0.20). (Note that such an analysis may result in overadjustment if a history of myocardial infarction is an intermediate in the causal pathway linking a high homocysteine level to a subsequent fatal myocardial infarction.) The incidence rate per 100 person-years of first-ever myocardial infarction was 1.76 (115/761) and was not clearly related to homocysteine levels (adjusted RR for the third compared with the first tertile, 1.17 [95% CI, 0.72 to 1.89]; P for trend, 0.51). There was no evidence for interactions between homocysteine and other risk factors, including hypertension (see below).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

A, Mortality from coronary heart disease according to tertiles of homocysteine levels. Gray line indicates first tertile (6 to 12 μmol/L); black line, second tertile (13 to 16 μmol/L); heavy black line, third tertile (17 to 97 μmol/L); and Prop, proportion. B and C, Mortality from cerebrovascular disease in normotensive (B) and hypertensive (C) subjects. Gray line indicates first and second tertile; black line, third tertile.

For cerebrovascular disease, the mortality rate per 100 person-years was 0.86 (58/878). It was highest in subjects with homocysteine levels in the highest tertile (Table 2⇑). This association was significant only in subjects without hypertension (Table 2⇑ and Figure 2B⇑ and 2C⇑; P for interaction, 0.0003), a finding that remained when we included a history of stroke at baseline as a potential confounder in the analysis (adjusted RR for the highest tertile compared with both lower tertiles combined, 3.92 [95% CI, 1.38 to 11.14]; P=0.01) or when we excluded subjects with stroke at baseline (RR, 4.42; P=0.01). The incidence rate per 100 person-years of first-ever stroke was 1.49 (98/833) and was not clearly related to homocysteine levels (adjusted RR for the third compared with the first tertile, 1.27 [95% CI, 0.78 to 2.07]; P for trend, 0.34). The risk was somewhat higher in normotensive subjects (RR, 1.77 [95% CI, 0.83 to 3.75]; P for trend, 0.14) than in hypertensive subjects (RR, 0.99 [95% CI, 0.52 to 1.91]; P for trend, 0.98; P for interaction, 0.15).

Adjustment for serum creatinine did not materially alter the above risk estimates (data not shown). Multivariate survival analyses with homocysteine as a continuous variable gave results similar to those shown above.

The adjusted RR of mortality from coronary heart disease per 1 μmol/L increase in homocysteine level was 1.014 (95% CI, 0.997 to 1.030; P=0.11) and 1.013 (95% CI, 0.995 to 1.031; P=0.15) after additional adjustment for history of myocardial infarction. For first-ever myocardial infarction, the adjusted RR was 1.010 (95% CI, 0.993 to 1.028; P=0.25). There was no evidence for interactions between homocysteine and other risk factors, including hypertension.

For mortality from cerebrovascular disease, the adjusted RRs were 1.007 (95% CI, 0.984 to 1.032; P=0.55) in the entire group, 0.937 (95% CI, 0.865 to 1.014; P=0.11) in the hypertensive subjects, and 1.024 (95% CI, 1.002 to 1.047; P=0.03) in the normotensive subjects (P for interaction, 0.03). After additional adjustment for a history of stroke, the RR in the normotensive subjects was 1.023 (95% CI, 0.997 to 1.051; P=0.09). For first-ever stroke, the adjusted RRs were 1.001 (95% CI, 0.979 to 1.023; P=0.92) in the entire group, 0.997 (95% CI, 0.961 to 1.033; P=0.85) in the hypertensive subjects, and 1.005 (95% CI, 0.978 to 1.033; P=0.73) in the normotensive subjects (P for interaction, 0.65).