Bostom AG, Silbershatz H, Rosenberg IH, Selhub J, D'Agostino RB, Wolf PA, Jacques PF, Wilson PWF. Nonfasting Plasma Total Homocysteine Levels and All-Cause and Cardiovascular Disease Mortality in Elderly Framingham Men and Women. Arch Intern Med. 1999;159(10):1077-1080. doi:10.1001/archinte.159.10.1077
Copyright 1999 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.1999
Elevated fasting total homocysteine (tHcy) levels were recently shown to confer an independent risk for all-cause and cardiovascular disease (CVD) mortality among selected Norwegian patients with confirmed coronary heart disease. We examined whether elevated fasting plasma tHcy levels were predictive of all-cause and CVD mortality in a large, population-based sample of elderly US women and men.
Nonfasting plasma tHcy levels were determined in 1933 elderly participants (mean age, 70 ± 7 years; 58.9% women) from the original Framingham Study cohort, examined between 1979 and 1982, with follow-up through 1992. Unadjusted and adjusted (ie, for age, sex, diabetes, smoking, systolic blood pressure, total and high-density lipoprotein cholesterol, and creatinine) relative risk estimates (with 95% confidence intervals [CIs]) for total and CVD mortality were generated by proportional hazards modeling, with tHcy levels (quartiles) as the independent variable.
There were 653 total deaths and 244 CVD deaths during a median follow-up of 10.0 years. Proportional hazards modeling revealed that tHcy levels of 14.26 µmol/L or greater (the upper quartile), vs less than 14.26 µmol/L (the lower three quartiles), were associated with relative risk estimates of 2.18 (95% CI, 1.86-2.56) and 2.17 (95% CI, 1.68-2.82) for all-cause and CVD mortality, respectively. The relative risk estimates after adjustment for age, sex, systolic blood pressure, diabetes, smoking, and total and high-density lipoprotein cholesterol levels attenuated these associations, but they remained significant: 1.54 (95% CI, 1.31-1.82) for all-cause mortality; 1.52 (95% CI, 1.16-1.98) for CVD mortality.
Elevated nonfasting plasma tHcy levels are independently associated with increased rates of all-cause and CVD mortality in the elderly.
IN 1969, the observations of McCully,1 who performed routine autopsy studies of children who died of homocystinuria due to distinct inborn errors of metabolism, led him to hypothesize that hyperhomocysteinemia may be linked to precocious arteriosclerosis. Registry data compiled by Mudd and colleagues2 subsequently confirmed that among children and young adults with homocystinuria and marked hyperhomocysteinemia due to cystathionine β-synthase deficiency, there was a dramatic excess risk of premature cardiovascular disease (CVD) morbidity and mortality. By 1976, Wilcken and Wilcken3 reported the first case-control findings indicating that much milder, long-term elevations in homocysteine (Hcy) might be associated with clinical arteriosclerosis in adults. Meta-analyses4- 6 pooling essentially all the published literature from 1975 through 1998 strongly suggest that mild to moderate elevation in fasting, nonfasting, or post–methionine loading Hcy levels confers an independent risk for clinical arteriosclerotic outcomes among adults, comparable in magnitude to a mild to moderate elevation in total cholesterol levels.
In a recent study,7 mild to moderately elevated fasting total homocysteine (tHcy) levels conferred an independent approximately 3- to 5-fold increased risk for subsequent all-cause (n=64 events) or CVD (n=50 events) mortality during 4.6 years of follow-up among 587 selected Norwegian patients (81.4% men; median age, 62 years) with angiographically confirmed coronary heart disease. We examined whether elevated fasting plasma tHcy levels were predictive of all-cause and CVD mortality in a large, population-based sample of elderly women and men.
The study population consisted of the original Framingham Heart Study cohort.8 Baseline examinations for the present analyses took place between May 1979 and May 1982, with follow-up occurring through May 1992. Of 2351 persons examined during the baseline period, 1933 had specimens available for plasma tHcy determinations. Additional baseline covariables assessed for the present analyses included age, sex, systolic blood pressure, cigarette smoking, diabetes, total and HDL cholesterol levels, and creatinine level.
Total homocysteine level was determined by high-performance liquid chromatography with fluorescence detection,9 using nonfasting plasma aliquots stored at −20°C from the baseline examination period, until mid-1997. Such long-term storage conditions have been validated for tHcy determinations in plasma or serum.9 Total and HDL cholesterol levels were assessed from fresh, nonfasting plasma samples by standard Lipid Research Clinics techniques,8 and creatinine level was measured in nonfasting plasma samples by the Jaffe method.10
Diabetes was operationally defined as use of insulin preparations or oral hypoglycemic agents, or any recorded blood glucose level of 11.1 mmol/L or higher (≥200 mg/dL), and smoking was defined as current cigarette smoking.
Information for outcome ascertainment was obtained from records supplied by hospitals, attending physicians, pathologists, medical examiners, and families. Cardiovascular disease death consisted (primarily) of coronary heart disease death and stroke death, with a very minor contribution from "other" CVD death (eg, hypertensive heart disease). As per all events in the Framingham Study, a panel of physicians reviewed the pooled evidence to arrive at the cause of death.8
The skewed tHcy data were natural log transformed, and comparisons of differences in geometric mean tHcy levels by sex, diabetes, and smoking status were performed using unpaired t tests. Spearman ρ was used to assess unadjusted rank-order correlations between untransformed tHcy levels, and age, creatinine, systolic blood pressure, total cholesterol, and HDL cholesterol. Unadjusted and adjusted (ie, for age, sex, diabetes, smoking, systolic blood pressure, total and HDL cholesterol) relative risk estimates (with 95% confidence intervals) for total and CVD mortality were generated by proportional hazards modeling, with tHcy levels (natural log or quartiles) as the independent variable.
Subject characteristics at the baseline examination are depicted in Table 1. Geometric mean tHcy levels were higher in men than in women (12.40 vs 11.27 µmol/L; P<.001), but did not differ according to the presence or absence of diabetes (11.67 vs 11.73 µmol/L, P=.86), or among current cigarette smokers vs nonsmokers (11.76 vs 11.71, P=.67). Weak, but significant Spearman correlations were observed between tHcy levels and age (+0.215, P<.001), creatinine (+0.186, P<.001), systolic blood pressure (+0.114, P<.001), HDL cholesterol (−0.114, P<.001), and total cholesterol (−0.049, P=.03). Quartiles of tHcy (in micromoles per liter) were as follows: Q1, 4.13-9.25; Q2, 9.26-11.43; Q3, 11.44-14.25; and Q4, 14.26-219.84.
There were 653 total deaths and 244 CVD deaths, during a median follow-up of 10.0 years. Although natural log tHcy (as a continuous variable) was associated with both total and CVD mortality in unadjusted and multivariable-adjusted proportional hazards analyses, the excess risk for these outcomes was largely confined to the uppermost quartile. Furthermore, the sex × tHcy (quartile analyses) interaction term was nonsignificant, so the total and CVD mortality analyses were not stratified by sex. Results of the total and CVD mortality analyses are presented in Table 2. A nonfasting plasma tHcy level of 14.26 µmol/L or greater vs less than 14.26 µmol/L was associated with an approximately 2.2-fold increased risk for both total and CVD mortality. Following multivariable adjustment, these associations persisted, but were attenuated to an approximately 1.5-fold increased risk for total and CVD mortality. Further adjustment for creatinine level (as a continuous variable, or as creatinine level ≥144 µmol/L vs <144 µmol/L) did not change these relationships (data not shown). Lastly, there was no evidence that the assumption of constant proportional hazards was violated (P>.2, data not shown).
The present findings are consistent with recently reported data from women and men participants aged 65 years and older in the Cardiovascular Health Study11 indicating that age, sex, elevated (brachial) systolic blood pressure, and elevated creatinine levels, were independently predictive of 5-year all-cause mortality. We further report the initial population-based evidence that elevated nonfasting tHcy levels may be independently predictive of all-cause and CVD mortality in elderly women and men. Our findings are in turn consistent with an earlier report indicating that elevated fasting tHcy levels were a powerful independent predictor of all-cause and CVD mortality in a younger, select group of predominantly male Norwegian patients with established coronary artery disease.7 Considerable subject population differences (ie, distribution of age, sex, and other key comorbidities and CVD risk factors) likely account for the sizable disparity in the magnitude of total and CVD mortality risk conferred by tHcy levels, comparing the present and previous study.7
The best estimate from meta-analyses4,5 updated through early 1998,6 for the increased risk of coronary heart disease morbidity and mortality comparing tHcy levels of greater than 15 µmol/L with levels less than 10 µmol/L, after adjustment for the traditional CVD risk factors, appears to be 1.4. This estimate does not change when only prospective studies are included (Omenn et al6; S.A.A. Beresford, PhD, written communication, November 1998). Simultaneous pursuit of 2 related areas of investigation will be required to confirm a causal relationship between hyperhomocysteinemia and CVD: (1) randomized, placebo-controlled trials of the effect of tHcy-lowering treatment on recurrent and de novo CVD outcomes and (2) elucidation of the basic pathomechanisms linking hyperhomocysteinemia to arteriosclerosis. Recently it has been proposed that clinical or even subclinical arteriosclerosis may play an important etiologic role in the development of hyperhomocysteinemia, so-called "reverse causality."12 This hypothesis appears untenable in light of the following published findings from both human and animal studies: (1) Despite the absence of any traditional CVD risk factors, 50% of untreated children and young adults with homocystinuria due to cystathionine synthase deficiency experience a major atherothrombotic event by age 30 years.2,13 Furthermore, strategies designed solely to reduce tHcy levels in these patients have been shown to decrease atherothrombotic event rates.2,13 (2) In adults (n=38; mean ± SD age, 58 ± 12 years) with mild hyperhomocysteinemia, tHcy-lowering treatment appears to have reduced the rate of progression of ultrasound-determined extracranial carotid artery plaque area.14 (3) Young, healthy subjects, free of clinical arteriosclerosis or CVD risk factors, who have normal baseline flow-mediated brachial artery reactivity, experience a dramatic, "dose-response" reduction in their flow-mediated brachial artery reactivity following acute hyperhomocysteinemia produced by an oral L-methionine load.15 (4) Randomized, controlled studies have revealed that mild dietary-induced hyperhomocysteinemia resulted in abnormal vascular reactivity among nonhuman primates,16 as well as increased arterial stiffness, and frank atherothrombotic sequelae, in minipigs.17
Our data confirm that tHcy levels are increased in elderly persons.9,18 Potential etiologic factors accounting for this age-related increase in tHcy levels include suboptimal intake and absorption of the key vitamin B cofactors or substrates for Hcy metabolism18; reduced activity of Hcy-metabolizing enzymes (for example, as demonstrated for cystathionine synthase19); and declining renal,20 and possibly thyroid21 function. Indeed, key limitations of the present analyses include the absence of data on intake or status of folic acid, vitamin B6, vitamin B12, and riboflavin, and the lack of a specific index of renal function such as glomerular filtration rate. A further potential limitation of these analyses is their relevance to the current era of folic acid fortification of cereal grain flour (140 µg per 100 g of flour) in the United States, begun voluntarily in March 1996, and by mandate on January 1, 1998.22
We conclude that, in the elderly, elevated nonfasting tHcy levels are independently associated with increased rates of all-cause and CVD mortality. Our findings require confirmation in other large population-based cohorts of elderly women and men.
Accepted for publication August 27, 1998.
Support for this work was provided by the US Department of Agriculture contract No. 53-3K06-5-10, the National Institutes of Health contract No. N01-HC-38038, the National Heart, Lung, and Blood Institute grant No. RO1-HL-40423-05, and the National Institute of Neurological Disorders and Stroke grant No. 2-RO1-NS-17950-12.
Reprints: Andrew G. Bostom, MD, MS, Division of General Internal Medicine, Memorial Hospital of Rhode Island, 111 Brewster St, Pawtucket, RI 02860 (e-mail: email@example.com).