Homocysteine and cardiovascular disease: a 17-year follow-up study in Busselton

Introduction

The strongest support for the relationship between homocysteine and cardiovascular disease arises from case–control studies with the relationship observed in prospective studies being weaker and less consistent [1–10]. Possible explanations for the varying results include a lack of control of confounding factors and the presence of pre-existing cardiovascular disease in the cohort under investigation [2, 8]. Two recently published meta-analyses demonstrate evidence of a modest independent association between homocysteine and cardiovascular disease [9, 10], however the variation in the level of control for confounding factors in the studies included in these reviews raises the possibility that the combined relative risk estimates from these meta-analyses may not be reliable. This study uses a prospective design with long-term cardiovascular follow-up to investigate the relationship between homocysteine and the risk of coronary heart disease (CHD) and stroke in men and women who were free from cardiovascular disease at baseline and investigates the importance of adequately controlling for known confounding factors.

Methods

This study is based on the 1612 men and women aged 40–89 years who participated in the 1981 Busselton Health Survey and had no history of diagnosed CHD or stroke at that time. This Busselton survey cohort and associated methods have been described previously [11]. In brief, survey participants completed a comprehensive health and lifestyle questionnaire, underwent various measurements and tests, and provided a fasting blood sample. The sera were frozen and stored and the homocysteine assays were done on the stored sera in 2001 by immunoassay on an Abbott IMX analyser (Abbott Diagnostics, Melbourne, Australia). The outcomes of interest were time from the 1981 survey to first CHD/stroke event and follow up extended to the end of 1998. Case–cohort sampling was used to reduce costs and preserve stored serum and this involved conducting homocysteine assays for all 215 CHD cases, all 118 stroke cases, and a random sample of 447 of the total cohort. The relationship between homocysteine levels and CHD or stroke was investigated using proportional hazards regression modelling with estimated hazard ratios adjusted for age, gender, body mass index, cholesterol, triglycerides, creatinine, diabetes, treatment for hypertension, systolic blood pressure and smoking. Homo-cysteine level was assessed both as a continuous variable and in gender-specific tertile groups.

OPEN IN VIEWER

Table 1

Hazard ratio and 95% confidence interval after adjusting for age/gender and other cardiovascular disease risk factors (multivariate) for continuous homocysteine and tertile groups

	CHD	Stroke	CHD/Stroke
Homocysteine tertile 1	1.00	1.00	1.00
Homocysteine tertile 2	0.79 (0.49, 1.26)	0.78 (0.43, 1.41)	0.85 (0.56, 1.29)
Homocysteine tertile 3	1.17 (0.74, 1.84)	0.95 (0.52, 1.74)	1.14 (0.75, 1.72)
Homocysteine∗	1.04 (0.88, 1.24)	1.09 (0.87, 1.36)	1.03 (0.88, 1.22)

∗

Hazard ratio per increase of 3 μmol/l. CHD, coronary artery disease.

Results

Table 1 shows the estimated hazard ratios for continuous homocysteine and tertile groups in relation to CHD, stroke, and CHD/stroke combined respectively. There is no evidence of any overall relationship between homocysteine level and risk of cardiovascular disease with all hazard ratios being close to one and none reaching statistical significance at the 5% level. The estimated hazard ratios for homocysteine were significantly different for men and women (P = 0.035 for CHD, P = 0.016 for stroke, P = 0.004 for CHD/stroke combined), but, when analysed separately, none reached statistical significance. For women, the hazard ratio for continuous homocysteine increased appreciably and reached statistical significance when creatinine was dropped from the model [hazard ratio = 1.28 95% confidence interval (CI) 0.99–1.67 for CHD; 1.40 95% CI 1.00–1.96 for stroke; 1.32 95% CI 1.05–1.66 for CHD/stroke].

Discussion

The results of this study provide little support for a significant independent relationship between level of homocysteine and the risk of coronary heart disease or stroke in a population with no evidence of pre-existing cardiovascular disease. Our results are in general agreement with several prospective cohort studies which have investigated populations free from cardiovascular disease and which adjusted for known confounding factors.

Our study design provided 89% power to detect a relative risk of 1.3 for CHD associated with a one SD change (about 3 μmol/l) in homocysteine, however, the results from the recently published meta-analyses suggest that the true relative risk may be closer to 1.15. Our study has only 40% power to detect a relative risk of that magnitude and this may have contributed to our negative findings.

The finding of a significant gender interaction has also been found in other studies [6, 7] and deserves further investigation. Furthermore, our findings indicate that failure to adjust for creatinine could result in an overestimate of risk [12]. The positive association reported by the Homocysteine Studies Collaboration [10] may have resulted from the absence of a study reporting a negative association [3] and the inclusion of two studies demonstrating over a 30% increase in risk, neither of which had controlled for creatinine [4, 7].