Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
American Journal of EPIDEMIOLOGY Volume 143 Copyright © 1996 by The Johns Hopkins University Number 9 School of Hygiene and Public Health May 1, 1996 Sponsored by the Society for Epldemlologlc Research ORIGINAL CONTRIBUTIONS Homocysteine Metabolism and Risk of Myocardial Infarction: Relation with Vitamins Bfi, B i9 , and Folate Petra Vertioef,1 Meir J. Stampfer, 23 Julie E. Buring,4'5 J. Michael Gaziano,4 Robert H. Allen,6 Sally P. Stabler,6 Robert D. Reynolds,7 Frans J. Kok,1 Charles H. Hennekens,4'5'8 and Walter C. Willett2-3 Elevated plasma homocyst(e)ine levels are an independent risk factor for vascular disease. In a case-control study, the authors studied the associations of fasting plasma homocyst(e)ine and vitamins, which are important cofactors in homocysteine metabolism, with the risk of myocardial infarction. The cases were 130 Boston area patients hospitalized with a first myocardial infarction and 118 population controls, less than 76 years of age, enrolled in 1982 and 1983. Dietary intakes of vitamins B6, B12, and folate were estimated from a food frequency questionnaire. After adjusting for sex and age, the authors found that the geometric mean plasma homocyst(e)ine level was 11 % higher in cases compared with controls (p = 0.006). There was no clear excess of cases with extremely elevated levels. The age- and sex-adjusted odds ratio for each 3-/xmol/liter (approximately 1 standard deviation) increase in plasma homocyst(e)ine was 1.35 (95% confidence interval 1.05-1.75; p trend = 0.007). After further control for several other risk factors, the odds ratio was not affected, but the confidence interval was wider and the p value for trend was less significant. Dietary and plasma levels of vitamin B 6 and folate were lower in cases than in controls, and these vitamins were inversely associated with the risk of myocardial infarction, independently of other potential risk factors. Vitamin B12 showed no clear association with myocardial infarction, although methylmalonic acid levels were significantly higher in cases. Comparing the mean levels of several homocysteine metabolites among cases and controls, the authors found that impairment of remethylation of homocyst(e)ine (dependent on folate and vitamin B 12 rather than on vitamin Be-dependent transsulfuration) was the predominant cause of high homocyst(e)ine levels in cases. Accordingly, plasma folate and, to a lesser extent, plasma vitamin B 12 , but not plasma vitamin B6, correlated inversely with plasma homocyst(e)ine, even for concentrations at the high end of normal values. These data provide further evidence that plasma homomocyst(e)ine is an independent risk factor for myocardial infarction. In this population, folate was the most important determinant of plasma homocyst(e)ine, even in subjects with apparently adequate nutritional status of this vitamin. Am J Epidemiol 1996; 143:845-59. folic acid; homocysteine; myocardial infarction; pyridoxine; vitamin B 12 Received for publication May 17, 1995, and in final form February 12, 1996. Abbreviations: Cl, confidence interval; SE, standard error. 1 Department of Epidemiology and Public Health, Agricultural University, Wageningen, Netherlands. 2 Channing Laboratory, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA. 3 Departments of Epidemiology and Nutrition, Harvard School of Public Health, Boston, MA. 4 Division of Preventive Medicine, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA. 5 Division of Ambulatory Care and Prevention, Brigham and Women's Hospital and Harvard Medical School, Boston, MA 8 Department of Medicine, Division of Hematology, University of Colorado Health Sciences Center, Denver, CO. 7 Department of Human Nutrition and Dietetics, University of Illinois, Chicago, IL 8 Department of Epidemiology, Harvard School of Public Health, Boston, M A Reprint requests to Dr. M. J. Stampfer, Channing Laboratory, 180 Longwood Ave., Boston, MA 02115. 845 846 Vertioef et al. Extremely elevated plasma levels of total homocysteine (generally referred to as homocyst(e)ine, the sum of all homocysteine species in plasma including free and protein-bound forms) are present in patients homozygous for genetic defects in enzymes of homocysteine metabolism, which may lead to premature vascular disease. Plasma homocyst(e)ine can be moderately elevated in individuals heterozygous for these defects or with inadequate intake of vitamins B 6 , B 12 , and folate, which serve as cofactors in the enzymatic pathways of homocysteine metabolism (1, 2). Homocysteine is a thiol-containing amino acid derived from methionine metabolism that can be degraded through two enzymatic pathways: transsulfuration and remethylation (figure 1). In the transsulfuration pathway, homocysteine is condensed with serine to form cystathionine, an irreversible reaction dependent on pyridoxal 5'-phosphate, the active form of vitamin B 6 . Subsequently, cystathionine is converted to cysteine in another vitamin B6-dependent reaction. In remethylation, homocysteine receives a methyl group from N-S-methyltetrahydrofolate (a vitamin B12-dependent reaction) or from betaine to form methionine. In the reaction involving betaine, which is independent of folate and vitamin B 12 , Af,iV-dimethylglycine is formed. The methionine formed is activated by adenosine triphosphate to form 5-adenosylmethionine, which serves as a methyl donor to a variety of acceptors, one being glycine, leading to the formation of yV-methylglycine. 5-Adenosylhomocysteine is formed in a transmethylation reaction, which is then hydrolyzed in homocysteine and adenosine (3, 4). Even moderately elevated plasma homocyst(e)ine levels appear to be an independent risk factor for coronary artery disease (5-7). One prospective study reported a 3.4-fold excess risk of myocardial infarction for men with levels above 15.8 /Ltmol/liter, the 95th percentile for controls in that population, as compared with those with levels below the 90th percentile (8). A recent prospective study confirmed the association and also suggested a graded association rather than a threshold effect (9). Suggested causative mech- Polyamines Acceptor S-AdenosylMethioninc (SAM) CH3-Acceptor ATP -Proteins Dimethylglycine S-AdenosylHomocysteine Choline ' ^ - * " Methionine 3 Serine THF •Glycine MethyleneTHF Adenosine Cystathionine PLP a-Ketobutyrate •*-^ —•• Cysteine I Taurine SO, FIGURE 1. Homocyst(e)ine metabolism in humans and animals. Enzymes: 1, /V-5-methyttetrahydrofoIate:homocysteine methyttransferase; 2, methylenetetrahydrofolate reductase; 3, betalne:homocysteine methyttransferase; 4, choline dehydrogenase; 5, cystathionine 0-synthase; 6, 6-cystathionase. THF, tetrahydrofolate; PLP, pyridoxal 5'-phosphate; ATP, adenosine 5'-triphosphate; B12, vitamin B 12 . (Adapted from J. Selhub and J. W. Miller. Am J Clin Nutr 1992;55:131-8). Am J Epidemiol Vol. 143, No. 9, 1996 Homocyst(e)ine, Vitamins, and Myocardial Infarction anisms include increased uptake of low density lipoprotein cholesterol in the vascular wall, promotion of vascular smooth muscle cell growth, and effects on vascular coagulant mechanisms (2). Several studies have demonstrated inverse associations of plasma homocyst(e)ine levels with blood levels of vitamins B 6 , B 12 , and folate and with intake levels of vitamin B 6 and folate (10, 11). Also, in human experimental studies, supplements of these vitamins (especially folate) reduce elevated homocyst(e)ine levels (12, 13). Some, but not all, epidemiologic studies have found inverse associations of the vitamins with vascular disease (5, 14). In the present case-control study, we investigated the associations of plasma homocyst(e)ine, dietary and plasma levels of vitamins B 6 , B 12 , and folate, and dietary methionine with the risk of first myocardial infarction. Also, we studied associations between plasma homocyst(e)ine and the vitamins. Furthermore, we compared plasma levels of several of the homocysteine derivatives shown in figure 1 between cases of first myocardial infarction and control subjects to study enzymatic disturbances in the pathways of homocysteine metabolism, either of nutritional or genetic origin (15). MATERIALS AND METHODS Study population Cases were white men and women less than 76 years of age with no history of previous myocardial infarction or angina and living in the Boston area. All admissions to the coronary or intensive care units of six suburban hospitals between January 1, 1982, and December 31, 1983, were reviewed to identify eligible cases. The diagnosis of myocardial infarction was confirmed from the hospital record, based on clinical history and creatinine kinase rise (16). Informed consent was obtained from the patients after obtaining permission from the admitting physician. A total of 450 cases were eligible for the study. Of these, nine were not invited to participate in the study for lack of physician consent, five could not be contacted after discharge, and 70 (16 percent of those invited to participate in the study) refused cooperation, leaving 366 potential cases. For each case, a control of the same age (within 5 years) and sex was selected at random from the residents' list of the town in which the case resided. Controls were ineligible if they had previous myocardial infarction or angina. Of 741 potential controls, 423 (57 percent) were willing to participate. For 340 cases, controls could be matched with respect to age and sex, leaving 340 case-control pairs (267 men in each group). Of these, sufficient plasma Am J Epidemiol Vol. 143, No. 9, 1996 847 was available for complete information on 130 cases and 118 controls (see below). Study measurements Eligible and willing subjects were visited in their homes by one of two nurses approximately 8 weeks after hospital discharge (cases) or at about the same time as the matching case (controls). These nurses obtained fasting venous blood samples, and information on dietary intake and coronary risk factors related specifically to the time period before the infarction for the cases and before the interview for the controls was obtained. The information included history of diabetes, high blood pressure, and high cholesterol; parental history of ischemic heart disease at or before age 60; weight and height (both self reported); physical activity; cigarette smoking; alcohol consumption; and diet. A physical activity index, expressed in kilocalories per week, was obtained by summing stairs climbed, blocks walked, and recreation and leisure time activity (17). Body mass index was calculated as the weight (kg)/ height (m)2. Blood was drawn into 0.1 percent ethylenediaminetetraacetic acid-containing Vacutainer tubes (Becton, Dickinson & Company, East Rutherford, New Jersey), and fresh plasma was used to determine total cholesterol and high density and low density lipoprotein cholesterol, as previously described (16). The rest of the plasma was stored at — 70°C, with similar storage time for cases and controls. In 1986, plasma pyridoxal 5'-phosphate was determined radiometrically by stimulation of tyrosine apodecarboxylase in 275 cases and 281 controls of the original study population (including 125 cases and 117 controls of the subsample described herein) (18). In 1993, we assayed plasma samples for homocyst(e)ine, several of its derivatives (cystathionine, cysteine, methionine, iV, N-dimethylglycine, Af-methylglycine, serine, and glycine), vitamin B, 2 , and folate, as previously described (15). A subsample of 130 cases (97 men, 33 women) and 118 control subjects (81 men, 33 women) had a sufficient plasma volume left for these assays. In addition to the homocysteine-related metabolites, we measured methylmalonic acid and 2-methylcitric acids I and n, which are usually elevated in subjects with vitamin B 1 2 deficiency (15). We had information on dietary intakes of methionine, the main metabolic precursor of homocysteine, and vitamins B 6 , B 12 , and folate for both the subsample of 130 cases and 118 controls with plasma assays and the original study population of 340 cases and 339 controls (one female control subject was excluded because of unreliable dietary data). Information on diet was collected using a semiquantitative 848 Verhoef et al. food frequency questionnaire, which was an extended and refined version of a previously validated questionnaire (19-22). The questionnaire included 116 food items plus vitamin supplements. For each food, a commonly used unit or portion size was specified, and participants were asked how often on average over the previous year they had consumed that amount. Nine responses were possible, ranging from "less than one time per month" to " six or more times per day." The intakes of methionine, vitamin B 6 , vitamin B 12 , folate, and other nutrients were computed by multiplying the frequency of consumption of each unit of food by the nutrient content of the specified portions. Composition values of foods were primarily obtained from US Department of Agriculture sources (23). Statistical analysis Mean values and proportions of various risk factors for coronary disease were compared for 130 cases of first myocardial infarction and 118 population controls and tested for significance using Student's / test for continuous variables and Pearson's x1 test for proportions. We also studied associations of homocyst(e)ine levels with plasma and dietary variables and with risk factors for coronary disease, controlling for age and sex by calculating correlation coefficients between residuals obtained from linear regression analyses of age and sex on homocyst(e)ine and of age on sex on the respective risk factor or vitamin level. Mean plasma levels of homocyst(e)ine and metabolites, as well as plasma and dietary levels of vitamin B 6 , vitamin B 12 , and folate, were compared for cases and controls. The mean dietary intake of methionine was compared as well. All distributions were skewed to the right, so we used log-transformation to normalize them. After log-transformation, dietary intakes were adjusted for total energy intake, using regression analysis (24). We also show geometric means of cases and controls, and p values are shown for the casecontrol difference in geometric means, adjusting for age and sex differences between the groups by means of linear regression analysis. We used logistic regression analysis to study the association of plasma homocyst(e)ine with the risk of myocardial infarction. We calculated odds ratios plus 95 percent confidence intervals for both plasma homocyst(e)ine as a continuous variable (per 1 standard deviation increase) and by quintiles (defined according to the distribution among controls, with the lowest quintile as a reference category). Multiple logistic regression analysis was used to control for sex, age, and several other coronary risk factors, some of which were significantly associated with homocyst(e)ine. Tests for trend were performed by adding log- transformed plasma homocyst(e)ine to the logistic regression models continuously. In the same way as for plasma homocyst(e)ine, we assessed the relations of dietary intakes of methionine and dietary and plasma levels of vitamins B 6 , B 12 , and folate with myocardial infarction. Furthermore, we repeated the risk analyses for plasma homocyst(e)ine as a continuous variable (per 1 standard deviation) in strata of several risk factors of coronary heart disease: age, smoking, history of high blood pressure, and history of hypercholesterolemia. All reported p values are two tailed. RESULTS Risk factors for coronary heart disease and associations with homocyst(e)ine Table 1 shows differences in major risk factors of coronary disease between 130 cases and 118 controls. Differences were mostly as expected, with cases having a significantly higher mean level of plasma total/ high density lipoprotein cholesterol, a higher mean intake of saturated fat, a lower mean alcohol intake, and a higher prevalence of hypercholesterolemia and diabetes. Also, among cases there were significantly more current smokers. Overall, the mean levels and proportions were similar to those observed in the original study population of 340 cases and 339 controls (data not shown), except for history of high blood pressure and family history of myocardial infarction. These were significantly lower in the larger group of controls as compared with cases, whereas in the subset controls had relatively high prevalences of a history of high blood pressure and family history of myocardial infarction. Associations of plasma homocyst(e)ine with risk factors of coronary disease and other possible confounding factors were evaluated, controlling for sex and age (table 1). Correlations among cases and controls differed substantially. Homocyst(e)ine levels were lower in women and directly associated with age, especially in cases. In control subjects, we found direct associations of homocyst(e)ine levels with alcohol consumption and with a history of hypercholesterolemia. In cases, but not in controls, a history of high blood pressure correlated directly with homocyst(e)ine. Mean levels of homocyst(e)ine, metabolites, vitamins, and methionine intake Mean and geometric mean plasma levels of homocyst(e)ine and several metabolites are shown in table 2. The geometric mean of plasma homocyst(e)ine was 11 percent higher in cases than in control subjects (p = 0.006). Homocyst(e)ine values of cases were Am J Epidemiol Vol. 143, No. 9, 1996 CO p o * t $ § II 0.01 45.4 25.6 0.001 0.13 0.10 r 0.15 0.27 J[ SmoMngt (current smokers) 0.26 -0.19 2,404 (843) r 2,292 (712) Mean % % -0.16 -0.13 0.07 0.15 15.4 10.2 0.95 0.04 0.22 -0.09 -0.07 r 0.07 -0.15 -0.02 r 0.31 0.44 P value history oi diabetes 38.0(17.9) 33.2 (13.6) Mean 42.6 43.2 % Mean 3,005(3,111) 3,369 (2,947) 0.93 0.19 0.06 r 0.71 0.02 0.04 r 0.83 0.63 0.03 0.51 P value 0.61 0.28 16.9 9.4 % 0.08 -0.04 0.22 r Mean 0.62 0.02 16.9 19.5 % Farrt 6.41 (1.82) 5.18(1.83) Mean p value 0.82 0.0002 P value 0.25 0.06 P value 0.60 -0.19 -O.06 r 0.03 0.53 P value y history of rrlyocarcSaJ 0.0001 0.17 0.11 r Total/HDL * cholesterol ratio 0.08 0.02 0.33 r Alcohol consumption (g/day) 14.2(24.2) 20.0 (28.4) history of high cholesterol 0.35 -0.05 -0.10 r P value Physical adMty (kcal/week) 26.0 (4.3) 26.2 (4.6) Mean P value Body mass Index (kg/(m)i) history of hypertension 0.09 0.81 P value Saturated tat Intake (g/day) 74.6 68.6 Energy Intake (kcaVday) P value 0.03 0.38 r P value 0.30 0.19 0.08 male Sex Matching factor 57.7 (9.3)§ 57.7 (9.1) r$ P value HDL, high density lipoprotein. For calculation of the Pearson correlation coefficient, smoking was coded as follows: 1 =never, 2 = exsmokers, 3 = <1 pack/day, 4 - 1-2 packs/day, and 5 a >2 packs/day. Pearson's correlation with homocyst(e)ine, adjusted for age and sex. Numbers in parentheses, standard deviation. p value of difference in risk factor between cases and controls. p value for case-control difference Cases Controls p value for case-control difference Cases Controls p value for case-control dfferencell Cases Controls Mean Age (years) TABLE 1. Cardiovascular risk factors In 130 cases of myocardlal infarction and 118 controls and correlations of risk factors with plasma homocyst(e)ine, Boston Area Health Study, 1982-1983 1 <; CD .2. i o Lj d. ns, and ial Infarction 850 Verhoef et al. •a 88 g § oi c\i 1O 3 1 d d a CO CO i - 00 •5 T3 73 <d CD CO CO S ° ST eg 8 q d in in I OJ II q d 2-2- I 8 •o c a o d o * e o § 0 g CO a c d « cq in S5 1 | g CD E OJ « - •D g O I8 a OS li oi co 2-52. <q to o o> li II u 3 X I* m as ID c 2 8 if0) 0) .= 5 shifted toward the right across the entire frequency distribution (figure 2). There was no clear excess of cases limited to those with extremely elevated levels. For cysteine and iV, N-dimethylglycine, we observed significantly higher geometric mean plasma levels in cases than controls, whereas methionine and Nmethylglycine levels were lower in cases (all p values below 0.05). Although not statistically significant, geometric mean cystathionine was higher among cases as compared with controls. Geometric mean levels of serine and glycine did not show substantial differences between the groups. The geometric mean level of methylmalonic acid was significantly higher among cases than among control subjects, possibly indicating a higher prevalence of vitamin B I 2 inadequacy among cases. However, the concentration of 2-methylcitric acids I and II did not differ significantly between cases and controls. Table 3 shows the mean dietary intake of methionine, the metabolic precursor of homocysteine, and the mean plasma and intake levels of vitamins B 6 , B 12 , and folate, which are all important cofactors in the enzymatic pathways of homocysteine metabolism. Adjusting for differences in sex and age, we observed that both geometric mean dietary (including intake from vitamin supplements) and plasma levels of vitamin B 6 and folate were significantly lower in cases than in controls. There was no statistically significant difference between cases and controls for dietary methionine and plasma or dietary vitamin B 12 , although methionine intake was slightly higher among cases, and plasma vitamin B 1 2 was slightly lower among cases. Associations of homocyst(e)ine with methionine intake and vitamins We calculated correlations between methionine intake and plasma homocyst(e)ine levels. Adjusting for age and sex only, we found that these correlations were -0.28 (p = 0.001) and -0.27 (p = 0.003) for cases and controls, respectively. After additional adjustment for total intakes of vitamins B 6 , B 12 , and folate, these correlations were —0.09 (p = 0.29) and -0.22 (p = 0.01, data not shown). Table 4 shows the correlation coefficients between plasma levels of homocyst(e)ine and plasma and intake levels of vitamins, adjusted for age and sex. Homocyst(e)ine correlated inversely with plasma levels of all three vitamins in cases as well as in controls. Correlations were particularly strong for the association of plasma homocyst(e)ine with plasma folate, with r = -0.38 (p = 0.0001) for cases and r = -0.49 (p = 0.0001) for controls. The association existed across the entire range of folate levels, even in subjects Am J Epidemiol Vol. 143, No. 9, 1996 Homocyst(e)ine, Vitamins, and Myocardial Infarction 851 N u m b e r 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Plasma Homocyst(e)ine (mnol/liter) FIGURE 2. Smoothed frequency distribution of homocyst(e)ine levels (/imol/liter), rounded to the nearest Integer, in plasma from 130 cases of first myocardial Infarction and 118 control subjects, Boston Area Health Study, 1982-1983. with normal to high levels. In cases, but not in controls, an inverse association of plasma homocyst(e)ine with plasma vitamin B 1 2 was observed (r = —0.35, p = 0.0001). In cases, intakes of all three vitamins showed significant inverse relations with plasma homocyst(e)ine, whereas in control subjects these associations were less strong. Because of the use of vitamin supplements and, to a lesser extent, because of similar dietary sources, vitamins were significantly intercorrelated in a similar way for cases and controls. In control subjects, correlation coefficients were 0.44 (p = 0.0001) for plasma pyridoxal 5'-phosphate and plasma folate, 0.30 (p = 0.0008) for plasma levels of pyridoxal 5'-phosphate and vitamin B 12 , and 0.33 (p = 0.0003) for plasma folate and plasma vitamin B 12 . Correlation coefficients were 0.57 (p = 0.0001) for intakes of vitamin B 6 and folate, 0.64 (p = 0.0001) for vitamins B 6 and B 12 , and 0.51 (p = 0.0001) for folate and vitamin B 12 . Thus, we also calculated correlations between levels of homocyst(e)ine with plasma and dietary levels of each of the vitamins, controlling for levels of the other two vitamins simultaneously (table 4). In control subjects, only for folate (plasma and dietary levels) did the association with homocyst(e)ine remain, whereas for vitamins B 6 and B 12 , associations disappeared. For Am J Epidemiol Vol. 143, No. 9, 1996 cases, both plasma folate and vitamin B 1 2 (plasma and dietary levels) remained inversely related to plasma homocyst(e)ine, when controlling for levels of each of the other two vitamins simultaneously. Associations of homocyst(e)ine with myocardial infarction We calculated risks by quintiles of plasma homocyst(e)ine, based on the distribution among controls. For individuals with plasma homocyst(e)ine levels higher than the 80th percentile of controls (11.2 ^i,mol/ liter), as compared with those with levels below the 20th percentile (7.2 /xmol/liter), we observed an ageand sex-adjusted odds ratio of 4.66 (95 percent confidence interval (CI) 1.90-11.40). The corresponding multivariate adjusted odds ratio was 5.09 (95 percent CI 1.84-14.10). For the three quintiles in the middle, the risks were elevated as well. There was a strong direct linear trend of plasma homocyst(e)ine with the risk of myocardial infarction (p = 0.01, table 5). In a logistic regression analysis with untransformed plasma homocyst(e)ine and age and sex as independent variables, we estimated a coefficient of 0.101 ± 0.043 (/3 ± standard error (SE)), corresponding to an odds ratio of 1.35 (95 percent CI 1.05-1.75) for each CO p 0.005 42.4 51.8 2.63 2.69 2.71 (0.67) 2.77 (0.69) 0.57 Geometric mean Mean Food 0.04 3.29(1.37) 5.79 (22.99) 3.05 3.49 Mean 0.60 0.12 10.02 9.09 Geometric moan 0.84 13.07(8.41) 14.20 (23.69) 11.26 11.13 Food plus supplements Geometric Mean moan 0.03 Food 0.01 340.9 (107.0) 372.0 (105.3) Mean 8.23 9.11 Plasma folate (nmol/ilter) Geometric Mean mean 8.76 (3.30) 9.93 (4.93) Dietary vitamin B,, 428 443 Food 11.52(7.56) 9.85 (4.02) 466 (199) 475(178) Food phis supplements Geometric Mean mean Dietary vitamin B, (mg/day) 49.3 (28.7)* 62.2 (40.5) Plasma vitamin B,. (pmol/Hter)t Q so metric Mean mean • PLP, pyrtdoxal 5'-phosphate; five cases and one control had missing values for plasma PLP. t We excluded one case with a plasma B,, value of 2,712 pmol/liter, which was far outside the range of all other subjects. t Numbers in parentheses, standard deviation. § Student's t test, based on the difference in geometric means between cases and controls, adjusting for age and sex differences. p value for case-control difference Cases Controls p value for case-control dlfference§ Cases Controls Plasma PLP* (nmol/ilter) Geometric Mean mean 0.25 323.7 357.7 Geometric moan 2.50 2.41 Geometric mean 0.002 423.0 (212.2) 493.6 (208.0) 377.4 453.7 Food plus supplements Geometric Mean moan Dietary folate (jig/day) 2.58 (0.64) 2.49 (0.63) Mean Dietary methionlne (g/day) TABLE 3. Dietary Intake of methionlne and plasma levels and dietary intake of vitamin Bv vitamin B12> and folate In 130 cases of myocardlai infarction and 118 controls, Boston Area Health Study, 1982-1983 a Homocyst(e)ine, Vitamins, and Myocardial Infarction 3-^xmol/liter (about 1 standard deviation) increase in plasma homocyst(e)ine. After additional adjustment for standard risk factors, including a history of high blood pressure, diabetes, plasma total/high density lipoprotein cholesterol, family history of ischemic heart disease, cigarette smoking, alcohol consumption, body mass index, and energy intake, we estimated a coefficient of 0.100 ± 0.051, corresponding to an odds ratio of 1.35 (95 percent CI 1.00-1.82). This can be interpreted as an average 35 percent increase for each 1 standard deviation increase in plasma homocyst(e)ine. To evaluate the possible effect modification of four major risk factors of myocardial infarction (age, history of high blood pressure, history of hypercholesterolemia, smoking), we assessed the risk associated with an increase of 3 /xmol/liter in plasma homocyst(e)ine among subjects stratified for these factors. We did not observe substantial differences among the strata of these risk factors (data not shown). CM o o co p p o o o n o O CO in p d d m JO o a o o 00 •a c a o in p r-- o d d o d 3 r-~ oo co p •a a 2S O <N s O d d O 853 _ CO Associations of methionine intake and vitamins with myocardial infarction o en CN i CM f- •a c a £ m o o p cl d § o i- O p p d d 3 1 CO o c O CM O CO d d I o o .c a eg p °° « a. ° o> en a> CD co || E TJT3 co ca cd p p d d *f co cn d d (D r O CO IT ii LLJ r- aj a Am J Epidemiol 8I 58 ..055 8S to c 3(3 Vol. 143, No. 9, 1996 111 10 For plasma pyridoxal 5'-phosphate, the highest two quintiles showed a trend toward an inverse relation with the risk of myocardial infarction, even after adjustment for strong confounders like plasma total/high density lipoprotein cholesterol and smoking habits (table 5) (p trend = 0.12). For plasma folate, the association was inverse as well, with a p value for trend of 0.09 after control for other factors. Plasma vitamin B 1 2 was not associated with the risk of myocardial infarction. For plasma pyridoxal 5'-phosphate, we did the same analyses in the original study group, of which 275 cases and 281 controls had pyridoxal 5'-phosphate levels measured. Both the age- and sexadjusted effect and the multivariate adjusted effect were stronger in the smaller study group as compared with the larger group. Also, there seemed to be more confounding in the larger group, especially from smoking and plasma total/high density lipoprotein cholesterol (data not shown). Table 6 shows odds ratios of myocardial infarction by quintiles of energy-adjusted daily intakes of methionine and vitamins B 6 , B, 2 , and folate (with and without intake from supplements). Intake of methionine showed no substantial association with the risk of myocardial infarction. For folate intake, and to a lesser extent for vitamin B 6 , we observed significant inverse associations with myocardial infarction risk, after adjustment for age and sex, as well as in a multivariate analysis, adjusting for possible confounders. For total vitamin B 6 intake, the estimates were rather inconsistent. Intake of vitamin B 1 2 was not materially associated with the risk of myocardial infarction; odds ratios CO CO CO o 1.29 1.19(0.53-2.69) £8.31 £6.56 1 1 0.71 (0.29-1.77) 1 £399 0.60 £321 1 0.97 0.97 (0.40-2.33) 1 1 £47.0 £29.7 0.41 0.47(0.18-1.22) £9.95 Plasma folate (nmol/llter) 1.13 1.07 (0.46-2.48) £503 Plasma vitamin B,, (pmot/Ilter) 0.94 1.68 (0.70-4.04) £63.1 Plasma PLPt (nmoVUter) 3.23 4.61 (1.61-13.20) £9.7 0.44 0.32 (0.12-0.84) £12.0 0.55 0.57(0.22-1.49) £610 0.37 0.37(0.14-1.01) £88.9 2.10 2.50 (0.85-7.32) £11.1 4 >610 0.53 0.57 (0.22-1.42) >12.0 0.97 (0.41-2.28) 0.94 0.32 0.51 (0.19-1.36) >88.9 4.66 5.09(1.84-14.10) >11.1 5 -0.804 ± 0.365 -0.707 ±0.417 -0.175 ±0.328 -0.151 ± 0.364 -0.644 ± 0.233 -0.407 ± 0.259 1.258 ±0.467 1.352 ±0.548 Pt±SEt P 0.03 0.09 0.59 0.68 0.006 0.12 0.007 0.01 value * Quintiles are based on the distribution of control subjects; real scale boundaries of quintiles are given, t Based on logistic regression analysis with log-transformed variables, i SE, standard error; PLP, pyrldoxal 5'-phosphate. § Adjusted for age, sex, energy intake, history of diabetes, history of high blood pressure, family history of Ischemic heart disease, total/high density lipoprotein cholesterol ratio, body mass index, cigarette smoking, and alcohol consumption. II Numbers in parentheses, 95% confidence interval. Age and sex adjusted MultJvariate§ adjusted Age and sex adjusted Multjvariate§ adjusted Age and sex adjusted Multivariate§ adjusted 3 Qulntlle* of plasma level Plasma homocyst(e)ine (nmol/liter) 1.81 1.80(0.62-5.21)11 £8.2 £7.2 1 1 2 1 Odda ratios of myocardlal Infarction by plasma levels of homocyst(e)ine, pyrldoxal 5 -phosphate, vitamin B12, and folate, Boston Area Health Study, 1982-1983 Age and sex adjusted Multivariate§ adjusted TABLE 5. CO p 1.51 1.60(0.65-3.91) £319 1 1 £282 1 1 0.72 0.63(055-1.62) 0.88 0.66(057-1.63) 0.67 0.48 (0.19-155) £3.79 >4.38 1.00 1.05(0.43-2.57) 44.38 0.17 0.17 (0.05-0.59) 1.09 1.17(0.46-3.01) £9.81 >12.28 1.67 1.77(0.73-4.34) £12.28 1.45 1.30(0.51-3.29) 0.67 0.70 (0.30-1.62) £391 Folate Mate, food only (no'day) 1.36 1.13(0.46-2.78) £12.61 >467 0.30 0.30(0.11-0.81) £467 1.42 1.46(0.61-3.51) 0.72 0.73(058-1.92) 0.66 0.57(054-1.35) >16.09 £16.09 0.45 0.44 (0.17-1.09) £490 >682 0.43 0.38(0.15-0.95) £682 0.31 054 (0.09-0.62) Fotate Mate, tood and supplements (w>May) 0.61 0.63 (0.27-1.49) £391 0.37 0.28 (0.11-0.74) £9.70 £7.06 £310 >3.26 £3.28 Vitamin B1t Intake, tood and supplements (ntfday) 0.96 0.84(0.31-254) 1 1 1 1.00 1.00(0.43-2.36) £257 Vtamki B1t Make, tood only (wjttay) 1.44 2.05 (0.86-4.90) £3.07 £8.32 1 1.43 1.28(0.52-3.11) 1.37 156(0.52-3.09) 5 >256 4 £256 Vtamln B, Nate, tood only (mgAJay) 1.18 1.04(0.41-2.62) £2.62 MetNonfrM Inlate (gAtay) Vitamin B, Make, tood and supplements (mgMay) 0.55 0.53(0.21-1.33) ££50 0.77 0.58(0.22-1.53)11 4657 1 1 £2.31 1 1 £2.12 1 1 ££24 £154 3 QuWlls' of nutrient intake -0.924 ± 0.302 -1.012 ±0.347 -1.115 ±0.454 -1.086 ±0.514 0.048 ± 0 5 3 3 0.052 ± 0 5 6 0 0.442 ± 0.283 0.522 ± 0.321 -O.651 ± 0.324 -0.729 ± 0.377 -0.301 ± 0.523 -0.438 ± 0.597 0.584 ±0.510 0.552 ± 0.578 Pt±S£* 0.002 0.004 0.01 0.03 0.84 0.84 0.12 0.10 0.04 0.05 0.57 0.46 0.25 0.34 P value * Quintiles are based on the dstribution of control subjects; rsal scale boundaries of qulntles are given, t Baaed on logistic regression analysis with log-transtormed variables. t SE, standard error. § Adjusted tor age, sax, energy intake, history of dtebetes, history of high Wood pmsure, family history of ischemic heart disease, total/high density Upoprotein cholesterol ratio body mass index, cigarette smoking, and alcohol consumption. ' II Numbers In parentheses, 95% confidence interval. Age and sex adjusted Multivariate§ adjusted Age and sex adjusted Mutth/ariate§ adjusted Age and sex adjusted Multivariate§ adjusted Age and sex adjusted Multivariate§ adjusted Age and sex adjusted Muluvariate§ adjusted Age and sex adjusted Multfvaiiate§ adjusted Age and sex adjusted Multivariate§ adjusted 2 1 TABLE 6. Odda ratio* of myocardW infarction by level of energy-adjusted daily intake of methionine, vitamin B 8 , vitamin B12< and folate, Boston Area Health Study, 1982-1983 856 Verhoef et at. for all quintiles were close to or above unity. Generally, for all risk analyses per quintiles, the small number of subjects for each quintile made the estimates quite unstable and the confidence intervals wide. We repeated the analyses shown for the smaller group in table 6 in the original study population of 340 cases and 339 controls; the associations were in the same direction but weaker than those in the smaller population for whom we had plasma. Intake of folate and, to a lesser extent, of vitamin B 6 was higher in the smaller control group as compared with the original control group (data not shown). Because vitamin intakes were intercorrelated, all variables were entered into the multivariate adjusted model continuously as log-transformed variables. Association of the total dietary intake of folate with myocardial infarction risk became somewhat stronger (/3 ± SE: -1.378 ± 0.483, p = 0.004) when the intakes of vitamin B 6 and vitamin B 12 were added to the logistic model as continuous variables. Adjustment for folate and vitamin B 1 2 weakened the association of total intake of vitamin B 6 with myocardial infarction (/3 ± SE: -0.3872 ± 0.4266, p = 0.36). This illustrates that increased intakes of both vitamins are associated with the decreased risk of myocardial infarction, but that part of the observed effect of vitamin B 6 is due to intercorrelation with folate. Finally, when plasma folate, plasma pyridoxal 5'phosphate, and plasma vitamin B 1 2 were entered into the logistic model together with plasma homocyst(e)ine (all as log-transformed continuous variables), the coefficient for homocyst(e)ine became somewhat smaller (/3 ± SE: 0.998 ± 0.620, p = 0.11) and for folate, much smaller (/3 ± SE: -0.242 ± 0.511, p = 0.22). The coefficient of pyridoxal 5'-phosphate was almost not affected (/3 ± SE: -0.311 ± 0.282, p = 0.27). This seems to indicate that folate and homocyst(e)ine are part of the same causal pathway (low folate levels leading to elevated plasma homocyst(e)ine which subsequently may lead to increased risk of myocardial infarction), whereas in this population vitamin B 6 possibly exhibits an association with myocardial infarction through mechanisms other than influencing homocyst(e)ine levels. DISCUSSION In this study, we found plasma homocyst(e)ine to be an independent risk factor for first myocardial infarction, and the effect seemed to be graded. Elevation of homocyst(e)ine in cases was associated with increased levels of cystathionine and cysteine, indicating increased catabolism of homocysteine through the transsulfuration pathway, which is dependent on vitamin B 6 . Remethylation of homocysteine to methionine seemed to be impaired in cases, as reflected by their lower plasma methionine levels. This was most Likely due to lower dietary intake of folate or subclinical vitamin B 12 deficiency among cases as compared with controls. Higher levels of AfJV-dimethylglycine suggest that there was increased demethylation of betaine to A^yV-dimethylglycine, a reaction step which is independent of folate and vitamin B ] 2 , at which homocysteine is remethylated to methionine. Accordingly, folate exhibited a strong inverse association with plasma homocyst(e)ine in cases as well as controls, whereas vitamin B 6 showed no association with plasma homocyst(e)ine, controlling for the other two vitamins. Therefore, although dietary and intake levels of vitamin B 6 were lower in cases than controls, this did not appear to be the primary reason for elevated fasting homocyst(e)ine levels. This is in line with the concept that the fasting homocyst(e)ine level is usually determined by homocysteine remethylation and not by abnormalities in the transsulfuration pathway (25). Total dietary intakes and plasma levels of vitamin B 6 and folate showed inverse relations with the risk of myocardial infarction, whereas for vitamin B 12 , no clear associations with the risk of myocardial infarction were observed. Vitamin levels were intercorrelated, however, and the effect of dietary vitamin B 6 was attenuated after correction for dietary folate and vitamin B 12 . The association between plasma pyridoxal 5'-phosphate and myocardial infarction risk was independent of plasma folate and homocyst(e)ine, suggesting that protective mechanisms of vitamin B 6 , other than its role in homocysteine metabolism, may be more important in this study population. Several possible mechanisms have been proposed to explain an apparent protective effect of vitamin B 6 against coronary disease (26). Also, it may be possible that low pyridoxal 5'-phosphate levels in cases are associated with reduced transsulfuration, maybe leading to increased homocyst(e)ine levels after protein-rich meals. However, without having measured plasma homocyst(e)ine in response to a methionine load, we cannot draw any conclusion about this. Among controls, we observed an inverse association between methionine intake and plasma homocyst(e)ine, taking into account confounding by age, sex, and intakes of vitamins B 6 , B 12 , and folate. One might expect a positive association, since methionine is the main precursor of homocysteine. However, a methionine-rich diet may induce a more efficient catabolism, as has been shown in rats (27). This might occur in humans as well, as suggested by our data. This may explain why we did not find a strong direct association of methionine intake with the risk of myocardial infarction in this population. Am J Epidemiol Vol. 143, No. 9, 1996 Homocyst(e)ine, Vitamins, and Myocardial Infarction The effects of folate and, to some extent, of vitamin B 6 (especially plasma pyridoxal 5'-phosphate) were stronger in the subsample shown herein than in the original study population, because of higher intake of these vitamins among controls of the subsample as compared with the original control group. Subjects in the subsample were selected on the criterion of having a sufficient volume of stored plasma. Since the same amount of blood was drawn from each subject, independently of case-control status or dietary habits, chance is the most plausible explanation for this observation. Another possibility may be residual confounding by age and sex, because cases and controls were not matched by these factors in the subsample, as was the case for the original sample. The selection bias may in part explain the observed associations for folate and vitamin B 6 and maybe plasma homocyst(e)ine with the risk of myocardial infarction. To get some indication of the effect of this bias, we estimated the expected homocyst(e)ine levels in cases and controls of the original study group that had plasma levels of pyridoxal 5'-phosphate available (281 cases, 275 controls), based on a linear regression equation obtained from data in the subsample. Included independent variables were the intake of the vitamins, plasma pyridoxal 5'-phosphate, age, sex, alcohol consumption, plasma total/high density lipoprotein cholesterol, methionine intake, and history of high blood pressure. In the original study group, estimated homocyst(e)ine levels were significantly higher in cases than in controls. Also, the increase in risk per 1 standard deviation increase in plasma homocyst(e)ine was similar to the one observed in the smaller study population. Hence, the selection bias might have strengthened the association of homocyst(e)ine with the risk of myocardial infarction, but it cannot have been totally responsible for the observed effect. The case-control design of the study leaves the possibility that the higher homocyst(e)ine level or lower vitamin levels were caused by the disease or its treatment. Cases were interviewed approximately 8 weeks after the event, and blood was drawn at the same time. This time span of almost 2 months should probably be enough to exclude any acute effect of the myocardial infarction on plasma homocyst(e)ine or vitamins. We tested this in 106 cases, for whom we had plasma levels of homocyst(e)ine and vitamins measured in blood that was drawn when they were still in the hospital. In a paired analysis, we found the hospital values of homocyst(e)ine to be significantly lower (-1.4 ± 0.25 (SE) ^mol/liter, p = 0.0001) than the values measured in blood obtained 8 weeks after the event. Consequently, any temporary decrease in homocyst(e)ine values in cases could only have caused Am J Epidemiol Vol. 143, No. 9, 1996 857 the association with myocardial infarction to be weaker but certainly not stronger. For plasma pyridoxal 5'-phosphate, the hospital values were significantly lower than the levels measured in blood that was drawn at the time of the interview. Therefore, any temporary decrease in plasma pyridoxal 5'-phosphate may have strengthened the inverse association of plasma pyridoxal 5'-phosphate with myocardial infarction. Plasma levels of vitamin B 1 2 and folate did not differ between blood samples drawn in the hospital and at the interview. The medications used by subjects could have affected their plasma levels of homocyst(e)ine or vitamins. However, most drugs that are known to have an effect on plasma homocyst(e)ine or vitamins, such as methotrexate, anticonvulsants, or penicillamine (1), are not prescribed more frequently in myocardial infarction patients. Recall bias or changes in dietary habits of the cases since the event could have influenced homocyst(e)inemyocardial infarction or vitamin-myocardial infarction relations. However, we observed very similar correlations between dietary and plasma levels of vitamin B 6 , vitamin B 12 , and folate for cases and controls, which reduces the possibility that either of those two biases had occurred. Since dietary assessment by means of a questionnaire is rather crude, this may explain why the associations of plasma homocyst(e)ine with dietary intakes of the vitamins were weak, as opposed to associations with plasma levels of the vitamins. Blood was stored for approximately 10 years at —70°C before samples were assayed for homocyst(e)ine and vitamins, which might have had an effect on homocyst(e)ine levels. However, because of similar storage time, sample treatment, and laboratory measurement for samples of cases and controls, chances are small that this could have affected the association. Also, the strength of the observed effect of homocyst(e)ine and the similarity with findings from other case-control (including prospective ones) and crosssectional studies on homocyst(e)ine and the risk of coronary heart disease (8, 9, 28-34) make it unlikely that chance or bias explain our findings. In a general population in Norway, Arnesen et al. (9) showed a graded association of homocyst(e)ine with coronary heart disease. They observed a significant 32 percent increase for each 1 standard deviation (4 /unol/liter) increase in plasma homocyst(e)ine, after adjustment for established coronary risk factors, similar to the 35 percent increase for each 3 jtimol/liter in our study. Stampfer et al. (8), in the Physicians' Health Study, probably a better nourished study population than that of Arnesen et al., found a threshold level above which 858 Vertioef et al. risk was increased. They observed that the higher mean homocyst(e)ine level in cases was due to an excess of cases with homocyst(e)ine values above the 95th percentile of control subjects (15.8 /xxnol/liter). Whether the level of nourishment of the studied population influences the shape of the relation of homocyst(e)ine with risk of vascular disease needs to be studied further. Our study population represented a well-nourished segment of a general population as well; both cases and controls lived in upper middle class neighborhoods in the Boston suburbs and were more highly educated than the general population. Only 10 subjects with first myocardial infarction (eight men, two women) and one male control subject had folate intakes less than 200 /xg/day, the current recommended daily allowance for men. The nutritional status of folate was less adequate, however. Thirty-nine cases (30 percent, 27 men and 12 women) and 27 controls (23 percent, 16 men and 11 women) had inadequate folate levels (^6.8 nmol/liter, US National Health and Nutrition Examination Survey II (NHANES II)). When plotting folate intakes against plasma homocyst(e)ine levels, we observed that homocyst(e)ine reached its nadir at folate intakes higher than 350-400 jig/day, similar to the findings of Selhub et al. (10). Plasma folate was inversely associated with plasma homocyst(e)ine across the entire distribution of plasma folate levels, also above the critical level of 6.8 nmol/ liter. These findings indicate that, for optimizing homocyst(e)ine levels, recommended folate intake should be at least 300-400 /xg/day. Several explanations for the positive association of homocyst(e)ine with vascular disease have been proposed. Homocysteine thiolactone, a reactive form of homocysteine, can modify low density lipoprotein, leading to aggregation and increased uptake of low density lipoprotein by macrophages. Homocysteine released from the low density lipoprotein within the vascular wall promotes intimal injury and influences vascular coagulant mechanisms (35). Furthermore, homocysteine may enhance the binding of lipoprotein(a) to fibrin, diminishing fibrinolysis (2). A recent publication points at the possibility that homocysteine promotes vascular smooth muscle cell growth and inhibits endothelial cell growth, both predisposing to atherosclerosis (36). In conclusion, our study adds further epidemiologic evidence to the hypothesis that plasma homocyst(e)ine is an important independent risk factor for coronary disease. The graded effect suggests that lowering of plasma homocyst(e)ine can have important public health effects for a large segment of the adult population. Furthermore, our study supports the view that adequate folate intake from diet or supplements might be an important step toward normalizing homocyst(e)ine levels. Supplementation with folate has been shown to successfully reduce homocyst(e)ine levels within a few weeks, even in subjects with genetically caused hyperhomocysteinemia (12). Whether this will also reduce the risk of coronary heart disease can only be shown by observational studies and randomized trials relating intake and status to the incidence of coronary heart disease (37). In the meantime, it might be worthwhile to reconsider the recommended daily allowance of 200 pig/day for folate and perhaps restore the previous value of 400 /xg/day. ACKNOWLEDGMENTS The research was supported by grants HL24423 and HL21006 and institutional training grant HL07575 from the National Heart, Lung, and Blood Institute, Bethesda, MD. The authors thank the six Boston area hospitals that participated in this study: Emerson Hospital, Concord (Dr. Marvin H. Kendrick), Framingham Union Hospital, Framingham (Dr. Marvin Adner), Leonard Morse Hospital, Natick (Dr. L. Frederick Kaplan), Mount Auburn Hospital, Cambridge (Dr. Leonard Zir), Newton-Wellesley Hospital, Newton (Dr. James Sidd), and Waltham Hospital, Waltham (Dr. Solomon Gabbay). They also thank Stephanie Parker, Tom Gaziano, Kathie Schneider, and Marty Vandenburgh, who assisted in the research. REFERENCES 1. Ueland PM, Refsum H. Plasma homocysteine, a risk factor for vascular disease: plasma levels in health, disease, and drug therapy. J Lab Clin Med 1989;114:473-501. 2. Rees MM, Rodgers GM. Homocysteinemia: association of a metabolic disorder with vascular disease and thrombosis. ThrombRes 1993;71:337-59. 3. Selhub J, Miller JW. The pathogenesis of homocysteinemia: interruption of the coordinate regulation by S-adenosylmethionine of the remethylation and transsulfuration of homocysteine. Am J Clin Nutr 1992;55:131-8. 4. Mason JB, Miller JW. The effects of vitamins B, 2 , B 6 , and folate on blood homocysteine levels. Ann N Y Acad Sci 1992;30:197-204. 5. Ueland PM, Refsum H, Brattstrbm L. Plasma homocysteine and cardiovascular disease. In: Francis RBJ, ed. Atherosclerotic cardiovascular disease, hemostasis, and endothelial function. New York: Marcel Dekker, Inc, 1992:183-236. 6. Malinow MR. Hyperhomocyst(e)inemia. A common and easily reversible risk factor for occlusive atherosclerosis. Circulation 1990;81:2004-6. 7. Malinow MR. Homocyst(e)ine and arterial occlusive disease: a mini review. Clin Chem 1994;40:173-6. 8. Stampfer MJ, Malinow MR, Willett WC, et al. A prospective study of plasma homocyst(e)ine and risk of myocardial infarction in US physicians. JAMA 1992;268:877-81. 9. Amesen E, Refsum H, Bonaa KH, et al. Serum total homocysteine and coronary heart disease. Int J Epidemiol 1995;24: 704-9. Am J Epidemiol Vol. 143, No. 9, 1996 Homocyst(e)ine, Vitamins, and Myocardial Infarction 10. Selhub J, Jacques PF, Wilson PWF, ct al. Vitamin status and intake as primary determinants of homocysteinemia in an elderly population. JAMA 1993,270:2693-8. 11. Ubbink JB, Vermaak WJH, van der Merwe A, et al. Vitamin B| 2 , vitamin B 6 , and folate nutritional status in men with hyperhomocysteinemia. Am J Clin Nutr 1993;57:47-53. 12. BrattstrOm LE, Israelsson B, Jeppsson JO, et al. Folic acid: an innocuous means to reduce plasma homocysteine. Scand J Clin Lab Invest 1988;48:215-21. 13. Ubbink JB, van der Merwe A, Vermaak WJH, et al. Hyperhomocysteinemia and the response to vitamin supplementation. Clin Investig 1993;71:993-8. 14. Pancharuniti N, Lewis CA, Sauberlich HE, et al. Plasma homocyst(e)ine, folate, and vitamin B 1 2 concentrations and risk for early-onset coronary artery disease. Am J Clin Nutr 1994;59:940-8. 15. Allen RH, Stabler SP, Savage DG, et al. Metabolic abnormalities in cobalamin (vitamin B12) and folate deficiency. FASEB J 1993;7:1344-53. 16. Buring JE, O'Connor GT, Goldhaber SZ, et al. Decreased HDLj and HDL, cholesterol, Apo A-I and Apo A-U, and increased risk of myocardial infarction. Circulation 1992;85: 22-9. 17. Paffenbarger RS Jr, Wing AL, Hyde RT. Physical activity as an index of heart attack risk in college alumni. Am J Epidemiol 1978; 108:161-75. 18. Reynolds RD. Vitamin B 6 analysis. In: Pesce AJ, Kaplan LA, eds. Methods in clinical chemistry. St. Louis, MO: CV Mosby, 1987:558-68. 19. Willett WC, Sampson L, Stampfer MJ, et al. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol 1985;122:51-65. 20. Willett WC, Sampson L, Browne ML, et al. The use of a self-administered questionnaire to assess diet four years in the past. Am J Epidemiol 1988;127:188-99. 21. Colditz GA, Willett WC, Stampfer MJ, et al. The influence of age, relative weight, smoking, and alcohol intake on the rcproducibility of a dietary questionnaire. Int J Epidemiol 1987; 16:392-8. 22. Willett WC. Nutritional epidemiology. New York: Oxford University Press, 1989. 23. US Department of Agriculture. Composition of foods: raw, processed, prepared. Washington, DC: US GPO, 1963-1988. (USDA handbook no. 8 series). Am J Epidemiol Vol. 143, No. 9, 1996 859 24. Willett W, Stampfer MJ. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol 1986;124:17-27. 25. BrattstrOm L, Israelsson B, Norrving B, et al. Impaired homocysteine metabolism in early-onset cerebral and peripheral occlusive arterial disease. Atherosclerosis 1990;81:51-60. 26. Willett WC. Does low vitamin B 6 intake increase the risk of coronary heart disease? In: Reynolds RD, Leklem JE, eds. Vitamin B 6 : its role in health and disease. New York: Alan R Liss, Inc, 1985:337-46. 27. Finkelstein JD, Martin JJ. Methionine metabolism in mammals. Adaptation to methionine excess. J Biol Chem 1986; 261:1582-7. 28. Israelsson B, BrattstrOm LE, Hultberg BL. Homocysteine and myocardial infarction. Atherosclerosis 1988;71:227-33. 29. Malinow MR, Sexton G, Averbuch M, et al. Homocyst(e)ine in daily practice: levels in coronary artery disease. Coron Artery Dis 1990; 1:215-20. 30. Ubbink JB, Vermaak WHJ, Bennett JM, et al. The prevalence of homocysteinemia and hypercholesterolemia in angiographically defined coronary heart disease. Klin Wochenschr 1991;69:527-34. 31. Genest JJ Jr, McNamara JR, Salem DN, et al. Plasma homocyst(e)ine levels in men with premature coronary artery disease. J Am Coll Cardiol 1990;16:l 114-19. 32. Kang SS, Wong PW, Cook HY, et al. Protein-bound homocyst(e)ine. A possible risk factor for coronary artery disease. J Clin Invest 1986;77:1482-6. 33. Von Eckardstein A, Malinow MR, Upson B, et al. Effects of age, lipoproteins, and hemostatic parameters on the role of homocyst(e)inemia as a cardiovascular risk factor in men. Arterioscler Thromb 1994;14:460-4. 34. Dalery K, Lussier-Cacan S, Selhub J, et al. Homocysteine and coronary artery disease in French-Canadian subjects: relation with vitamins B 12 , B 6 , pyridoxal phosphate, and folate. Am J Cardiol 1995,75:1107-11. 35. McCully KS. Chemical pathology of homocysteine. I. Atherogenesis. Ann Clin Lab Sci 1993;23:477-93. 36. Tsai JC, Perrella MA, Yoshizumi M, et al. Promotion of vascular smooth muscle cell growth by homocysteine: a link to atherosclerosis. Proc Natl Acad Sci U S A 1994;91: 6369-73. 37. Stampfer MJ, Willett WC. Homocysteine and marginal vitamin deficiency. The importance of adequate vitamin intake. (Editorial). JAMA 1993;270:2726-7.