cysteine, nins and vascular e
idence m
Petra Verhoef 1 f pNO^O f ßOb
Stellingen
1. Het verband tussen het plasmagehalte van totaal homocysteïne en het risico op hart• en vaatziekten is continu stijgend, zonder een duidelijke drempelwaarde. Dit proefschrift.
2. Homozygotie voor thermolabiliteit van 5,10-methylenetetrahydrofolaat reductase kan vooral in combinatie met een lage foliumzuurstatus leiden tot sterke verhoging van het nuchtere plasma totaal homocysteïnegehalte. Dit proefschrift.
3. Er zijn interventiestudies nodig, waarbij wordt onderzocht in hoeverre verlaging van het plasma totaal homocysteïnegehalte via foliumzuursuppletie (eventueel aangevuld
met vitamines B6 en B12) het risico op hart- en vaatziekten reduceert. Dit proefschrift.
4. Plasma pyridoxal 5'-fosfaat (de actieve vorm van vitamine B6) is, onafhankelijk van plasma totaal homocysteïne, invers geassocieerd met het risico op hart- en vaatziekten. Dit proefschrift.
5. Op basis van de nu beschikbare epidemiologische bevindingen is er voldoende reden om, ter preventie van hart- en vaatziekten, de consumptie te stimuleren van groenten en vruchten, die rijk zijn aan foliumzuur (bijvoorbeeld groene bladgroenten, bietjes, spruitjes, sinaasappelen, bananen).
6. Onderzoek naar de interactie tussen genetische gevoeligheid en leefstijl kan alleen bijdragen aan een effectievere ziektepreventie, indien er goede mogelijkheden zijn voor screening op deze gevoeligheid.
7. De paniek over de hoge kosten van wachtgeld voor aio's die niet op tijd hun proefschrift afronden, staat in geen verhouding tot de onderzoeksproduktie die door hen wordt geleverd, tegen een laag salaris.
8. De uitspraak "Een epidemioloog is een arts die kan tellen" is van toepassing op artsen maar niet op epidemiologen.
9. Het vergelijken van intellectuele prestaties van tweelingen, bijvoorbeeld door ouders of docenten, kan leiden tot identiek carrièreverloop bij de kinderen.
10. Het feit dat de vraag "Hoe gaat het?" steeds meer bij wijze van groet wordt gebruikt, verklaart waarom zo vaak geantwoord wordt met de dooddoener "Druk, druk!". 11. Noorderlingen die laatdunkend spreken over 'mafiana-mentaliteit' van Latijns- Amerikanen, beseffen onvoldoende dat het leven maar kort is en dat de dag, liefst vandaag, geplukt moet worden.
12. "Wie spaarzaam is met zijn woorden, toont verstand en wie zichzelf beheerst, is een man van inzicht." Spreuken 17, vers 27.
Stellingen behorende bij het proefschrift
"Homocysteine, B-vitamins and Cardiovascular Disease: epidemiologie evidence"
Petra Verhoef Wageningen, 25 maart 1996 Homocysteine, B-vitamins and Cardiovascular Disease:
epidemiologic evidence
Petra Verhoef
0000 0670 6101 Promotoren: Dr. F.J. Kok Hoogleraar in de Humane Epidemiologie.
Dr. MJ. Stampfer Professor in Epidemiology and Nutrition, Harvard School of Public Health. Homocysteine, B-vitamins and Cardiovascular Disease:
epidemiologic evidence
Petra Verhoef
Proefschrift
ter verkrijging van de graad van doctor in de landbouw- en milieuwetenschappen op gezag van de rector magnificus, dr. CM. Karssen, in het openbaar te verdedigen op maandag 25 maart 1996 des namiddags om vier uur in de Aula van de Landbouwuniversiteit te Wageningen These Ph.D.-studies were supported by a grant from the Netherlands Organization for Scientific Research (NWO).
Financial support by the Wageningen Agricultural University, the Netherlands Organization for Scientific Research, Bayer AG, and Severi-Med GmbH for the publication of this thesis is gratefully acknowledged.
CIP-DATA KONINKLIJKE BIBIÓTHEEK, DEN HAAG
Verhoef, Petra
Homocysteine, B-vitamins and cardiovascular disease: epidemiologic evidence / Petra Verhoef. -[S.l. : s.n] (Wageningen: Grafisch Service Centrum Van Gils) Thesis Landbouw Universiteit Wageningen. - With ref. - With summary in Dutch. ISBN 90-5485-500-2 Subject headings: homocysteine / B-vitamins / cardiovascular disease.
Cover : Hans Lemmens, Amsterdam Printing : Grafisch Service Centrum Van Gils B.V., Wageningen
© P. Verhoef, Wageningen, The Netherlands, 1996 Voor mijn ouders
Abstract
Homocysteine, B-vitamins and Cardiovascular Disease : epidemiologic evidence
Ph.D.- thesis by Petra Verhoef, Department of Epidemiology and Public Health, Agricultural University, Wageningen, The Netherlands, March 25, 1996.
Background Cardiovascular disease constitutes a major public health problem in the Netherlands and other Western countries. Elevated plasma homocysteine has attracted growing interest as a "new" risk factor for cardiovascular disease. Homocysteine is formed from the essential amino acid methionine. Defective homocysteine metabolism may lead to elevation of plasma total homocysteine (tHcy). Genetic enzyme deficiencies or inadequate intake of vitamins B6, B,2, and folate, all important cofactors in homocysteine metabolism, may result in elevation of tHcy. Accumulation of tHcy can possibly promote atherosclerotic or thrombotic processes. Methods The epidemiologic studies presented in the thesis, aimed to find additional evidence for the hypothesis that elevated plasma tHcy is an independent risk factor for cardiovascular disease. We addressed various disease endpoints, with data of prospective and retrospective studies, from Dutch, European, and US populations. The role of the B-vitamins and of a genetic enzyme defect, predisposing to high tHcy levels, were studied. Results Overall, in line with other findings, most of our studies showed that elevated tHcy is an independent risk factor for cardiovascular disease. Results indicated that the risk increased with rising levels of tHcy, with no threshold effect. The estimated average % increase in risk for 5 umol/L (about 1 SD) increase in fasting tHcy varied between 20% and 60% in the various studies. In a large European case-control study, we found that elevated tHcy was a strong risk factor in women, both in pre- and postmenopausal women. Folate concentrations in plasma or expressed per haematocrit, and dietary folate were found to be important determinants of plasma tHcy in several studies. In one of our studies, in concordance with findings of others, tHcy reached its nadir at a folate intake of 400 ug/day. Furthermore, we observed that homozygosity for a mutation in 5,10- methylenetetrahydrofolate reductase, in combination with low folate status, predisposed to particularly high tHcy levels, and may thereby increase risk of cardiovascular disease. Conclusions & implications Dietary folate intake of a large segment of the general population is lower than 400 ug/day, and tHcy may be generally increased. Several studies have already shown that elevated tHcy can be normalized by supplementation with folate, even at a dose of 650 ug/day. Thus, increased folate intake seems an important way to decrease tHcy in populations, thereby possibly reducing incidence of cardiovascular disease. Large-scale prevention trials are warranted to demonstrate the efficacy of tHcy- lowering, and the minimal folate intake required. At this moment, based on the available epidemiologic evidence, it is advisable to increase consumption of fruits and vegetables in the general population. Results from prevention trials will indicate whether additional measures, such as fortification of food or supplementation are justified as well.
Contents
Chapter 1 General introduction 1
Chapter 2 Plasma total homocysteine, B-vitamins and risk of coronary 13 atherosclerosis. submitted
Chapter 3 Combination of mutated methylenetetrahydrofolate reductase and 31 low folate status is associated with high plasma total homocysteine. submitted
Chapter 4 Homocysteine metabolism and risk of myocardial infarction: 45
relationship with vitamins B6, B12 and folate. Am J Epidemiol, in press
Chapter 5 Plasma total homocysteine and future risk of angina pectoris 69 with evidence of severe coronary atherosclerosis. submitted
Chapter 6 A prospective study of plasma total homocysteine and risk of 87 ischemic stroke. Stroke 1994;25:1924-1930
Chapter 7 Homocysteine and cardiovascular disease: gender differences 103 and effect of menopause. The European Concerted Action Project.
Chapter 8 General discussion 123
Summary 149
Samenvatting 155
Appendix 161
Dankwoord 163
Curriculum vitae 167
1 General Introduction
Introduction
Cardiovascular disease is the main cause of mortality in the Netherlands and other Western countries. In the Netherlands, cardiovascular disease mortality, mostly due to ischemic heart disease, constitutes about 40% of total mortality.1 Many factors may be involved in the etiology of cardiovascular disease, i.e. smoking, high blood pressure, high serum cholesterol levels, and diabetes. Clearly, some of these factors are determined by lifestyle, whereas others have a stronger genetic background. Generally, cardiovascular disease is thought to be multifactorial, and interaction of various factors (for example between inherited properties and a lifestyle factor) or combined occurrence of several risk factors may lead to the disease.2 Despite the advances that have been made in identifying cardiovascular risk factors, a large proportion of cardiovascular disease incidence cannot be explained by well-known risk factors. Elevated plasma homocysteine has been designated a risk factor for cardiovascular disease. Homocysteine is a metabolite of methionine, one of the dietary essential amino acids. Defects in intracellular homocysteine metabolism may lead to elevation of plasma homocysteine. These metabolic defects can have a genetic background, i.e. an inherited enzyme deficiency, or a nutritional background, i.e. an inadequate intake of one or more B-vitamins that serve as cofactors to the enzymes involved. Homocysteine accumulation can possibly promote the formation of atherosclerotic plaques, or affect blood coagulation.
Homocysteine metabolism
Enzymes involved - transsulfuration and remethylation Homocysteine is a sulfur-containing amino acid, formed at demefhylation of the essential amino acid methionine. Intracellular homocysteine is condensed with serine to form cystathionine, a reaction catalyzed by the enzyme cystathionine ß-
1 Chapter 1 synthase (CS, Figure 1-1). Cystathionine is subsequently split to cysteine and a- ketobutyrate, completing the conversion of methionine to cysteine, referred to as the transsulfuration pathway, the major route for homocysteine catabolism. Part of the homocysteine is remethylated to methionine, by the enzyme methionine synthase (MS, Figure l-l).3
Vitamins that serve as cofactors
Three B-vitamins are involved in homocysteine metabolism: vitamins B6, B12, and folate (Figure 1-1). The enzyme cystathionine 8-synfhase requires pyridoxal 5'- phosphate (PLP), the biologic active form of vitamin B6, as a cofactor. Methionine synthase requires vitamin B,2 as a cofactor and 5-methyltetrahydrofolate (methyl- THF) as a substrate. The latter is formed at the reduction of 5,10- mefhylenetetrahydrofolate (methylene-THF), catalyzed by the enzyme 5,10-
3 methylenetetrahydrofolate reductase (MTHFR). Thus, vitamin B6 is important in homocysteine transsulfuration, whereas folate and vitamin B12 play significant roles in homocysteine remethylation, in most cells and tissues.
Cystathionine
Cysteine
Figure 1-1. Condensed scheme of homocysteine metabolism* Enzymes are denoted by boxes, whereas cofactors and cosubstrates are denoted by ellipses. CS = cystathionine fi- synthase; MS = methionine synthase; THF = tetrahydrofolate; methyl-THF = 5- methyltetrahydrofolate; methylene-THF = 5,10-methylenetetrahydrofolaat; MTHFR =
5,10-methylenetetrahydrofolate reductase; PLP = pyridoxal 5'-phosphate; Bn = vitamin
B,2 (methylcobalamine); F = folate.
* A detailed description of homocysteine metabolism is given in Chapter 4.
2 General Introduction
Inborn defects Several inborn errors may lead to accumulation of homocysteine in the cell. Excess homocysteine is exported into the extracellular media, like plasma and urine.3 Thus, subjects with inborn errors have extreme elevations of plasma homocysteine (referred to as hyperhomocysteinemia) and excrete large amounts of homocysteine in the urine (homocystinuria). Examples of inborn errors are defects in cobalamin metabolism, or deficiency of MTHFR.4 The classic form of a genetic disorder in homocysteine metabolism is deficiency of the enzyme cystathionine 8- synthase.5 In individuals who are homozygous for the defect, homocystinuria and hyperhomocysteinemia occur. Heterozygote subjects have normal or only moderately elevated plasma levels of homocysteine in the fasting state. However, their homocysteine levels in response to an oral methionine loading may be abnormally high, compared to control subjects, although overlap between the groups exists.5 Thus, the methionine loading test was developed to discriminate between subjects heterozygous for cystathionine 8-synthase deficiency and normal controls. Usually, peak homocysteine levels, 4 to 8 hours after methionine loading, are measured. Sometimes the test is combined with additional enzyme determination, which appears to improve the identification.6
Inborn metabolic defects and vascular disease
Defective transsulfuration Individuals with homocystinuria, e.g. due to cystathionine B-synfhase deficiency, usually suffer from vascular occlusion and thrombosis in arteries and large veins, sometimes before the age of 20. Coronary, cerebral, and peripheral vessels may all be affected.7 That observation inspired Wilcken and Wilcken to formulate the hypothesis that moderately elevated plasma levels of homocysteine (due to heterozygous cystathionine 8-synthase deficiency) could predispose to cardiovascular disease. In 1976, they published a study in which homocysteine levels after methionine loading were compared between coronary artery disease patients and controls subjects.8 They found plasma homocysteine (measured as homocysteine-cysteine mixed disulfides) to be present in 68% of the patients and 23% of the control subjects. Thereafter, many similar studies have confirmed the finding of a greater proportion of elevated homocysteine levels after methionine loading in cases of cardiovascular disease than in controls.4,910
3 Chapter 1
Defective remethylation Through results of some studies, however, the attention was shifted away from the concept that elevated homocysteine levels due to cystathionine 8-synthase deficiency were a main cause of increased risk of cardiovascular disease. First of all, some studies failed to show that heterozygous cystathionine B-synfhase deficient subjects were at increased risk for cardiovascular disease.11,12 Also, although previous studies had found lower cystathionine 8-synthase activity in vascular patients with an abnormal homocysteine response after methiomne loading,13'14 other investigations could not reproduce these results.15 Furthermore, the prevalence of cystathionine 6-synthase deficiency (either in homozygous or heterozygous form) is too low to account for the number of observed cardiovascular patients with abnormal post-methionine loading responses.16 Clearly, there had to be other reasons for the abnormal responses, and reduced remethylation was one of the possibilities. In fact, Kang et al. reported that a genetic disorder, resulting in defective homocysteine remethylation, was associated with increased risk of cardiovascular disease. They identified thermolability of the enzyme MTHFR, determined by biochemical phenotyping, as a risk factor for cardiovascular disease.17 In contrast to severe MTHFR deficiency, in which the residual enzyme activity in cultured fibroblasts is 0-20%, thermolabile MTHFR has a 50% residual activity. Furthermore, many studies have found that elevated fasting levels of homocysteine are related to cardiovascular disease.18 Since subjects with cystathionine 8-synthase deficiency usually have normal fasting levels, this suggests that impaired remethylation may be related to cardiovascular disease as well. Generally, elevated fasting homocysteine levels are thought to reflect impaired homocysteine remethylation, whereas elevated homocysteine levels in response to a methiomne loading test are thought to reveal reduced transsulfuration.7 Impairment of the pathways could result from inherited enzymatic defects, as described before, or from vitamin inadequacies. Conceivably, combined occurrence of both enzymatic defects and vitamin deficiencies could act synergistically.
B-vitamins
Associations with homocysteine Several studies have shown inverse associations between plasma levels of homocysteine and levels of the vitamins involved in its metabolism, either for dietary intake or plasma/serum levels of the vitamins. Strongest associations have
4 General Introduction been found for folate. Homocysteine elevation does not necessarily occur in subjects with clear deficiencies.19'20 Because of the distinct role of the vitamins in remethylation and transsulfuration, inadequacy of folate and vitamin B12 will generally lead to increased fasting homocysteine levels, whereas vitamin B6 inadequacy will normally lead to increased homocysteine levels (above the fasting level) after methionine loading. For example, a study showed that vitamin BI2 deficiency led to elevated levels of fasting homocysteine, whereas the increase in homocysteine after methionine loading was normal.21 In another study, PLP deficiency was found to cause increased post-load homocysteine levels, but fasting levels remained normal.22
Homocysteine reduction by vitamin supplementation The fact that B-vitamins can reduce plasma homocysteine further emphasizes their important role. In many studies it has been shown, that supplementation with folate can effectively reduce plasma homocysteine (fasting as well as post-load levels) in almost all subjects with elevated homocysteine, even when concentrations
23 27 of folate, vitamin BI2, and PLP are normal. " Intervention with vitamin B6 has very little effect alone, but in combination with folic acid a normalization of
28 29 homocysteine can be achieved in almost all subjects. ' Vitamin BI2 30 supplementation may lower homocysteine levels in vitamin B12-deficient subjects.
Atherogenic and thrombogenic action of homocysteine
Homocysteine is thought to promote both atherogenesis and thrombosis, thus explaining the relationship with vascular disease.** Homocysteine appears to affect the vascular endothelium, platelets, and coagulation proteins.31 Considerable controversy about the mechanisms remains. Atherogenic mechanisms promoted by homocysteine include endothelial cell desquamation, oxidation of LDL cholesterol, and monocyte adhesion to the vascular wall. Also, antithrombotic properties of vascular endothelium may be decreased. Furthermore, there are indications that coagulation factors are affected in a way that predisposes to a thrombotic event. Information on effects of homocysteine on platelet kinetics are controversial.31 In addition, high levels of homocysteine may inhibit growth of endothelial cells and stimulate proliferation of smooth muscle cells, subsequently leading to thickening of
Several chapters address the mechanisms in detail.
5 Chapter 1 arterial walls.3
Homocysteine terminology
In plasma, homocysteine predominantly exists as a complex with protein. Furthermore, there is a free oxidized fraction (disulfides of homocysteine with itself [homocystine] or with cysteine). A small part exists in the free reduced form (sulfhydryl form).3 The total of all the multiple forms is referred to with total homocysteine. This has been measured in all described studies in the thesis. Throughout the rest of the thesis we will use the term total homocysteine, abbreviated as tHcy to indicate either total homocysteine or one of the subtractions, e.g. homocysteine-cysteine mixed disulfides or protein-bound homocysteine. In that way the term tHcy is synonymous with the frequently used term homocysteine. We will use homocysteine when we specifically want to address the amino acid itself, e.g. in "homocysteine metabolism" or when referring to pathofysiologic effects of the amino acid, since those may depend on the presence of the sulfhydryl group of homocysteine.33 tHcy, vitamins and vascular disease - epidemiologic evidence
Since the study of Wilcken and Wilcken,9 in 1976, several epidemiologic studies on plasma tHcy and cardiovascular disease have been conducted. About 30 studies were recently reviewed in a meta-analysis.18 Virtually all studies found that elevated plasma tHcy, either in the fasting state or after methionine loading, is more frequently shown in patients with disease of the cerebrovascular, peripheral, or coronary vessels than in control subjects.18 In the meta-analysis, the association appeared to be strongest for peripheral vascular disease. Elevated tHcy is considered to act as a risk factor for cardiovascular disease, independently of conventional risk factors.4'18 The findings of the meta-analysis and some other studies,34,35 are in favor of a graded effect between tHcy and risk of cardiovascular disease (i.e. rising risk with increasing tHcy levels), but other studies only found increased risks for subjects with tHcy levels above a certain cutoff-point. However, estimation of the size of the effect still remains relatively imprecise. Also, the cause of elevated tHcy levels in vascular disease patients, i.e. genetic or nutritional background, was not addressed in many of the epidemiologic studies. The ones that did study levels of B-vitamins involved in homocysteine metabolism, including folate, did not always
6 General Introduction find lower levels in cardiovascular disease patients than in controls, despite higher tHcy levels in patients.4,36 In most epidemiologic studies, women formed a small part of the study population or were not included at all. Generally, tHcy levels are lower in women than in men.4 However, there are indications that at menopause, due to hormonal changes, tHcy levels may rise, even exceeding levels of men of the same age.24'37 Furthermore, most epidemiologic research on plasma tHcy and risk of cardiovascular disease has been retrospective or cross-sectional, and consequently cannot definitively determine whether increased tHcy levels are the cause or the result of cardiovascular disease. Prospective studies do not have this limitation.38
Outline of the thesis
In 1990, a research proposal was written for a study on tHcy, B-vitamins and risk of angiographically-defined coronary atherosclerosis. Notably, at that time, two studies had indicated a lack of support for elevated tHcy as a risk factor for coronary artery disease.13'39 In 1992, the proposed research started at the Department of Epidemiology and Public Health of the Agricultural University in Wageningen, and was (partly) included in the European Concerted Action Project: Hyperhomocysteinemia and Vascular Disease. Additionally, during the Ph.D.- project, in collaboration with the Departments of Epidemiology and Nutrition of the Harvard University in Boston, the association between elevated tHcy and risk of cardiovascular disease was investigated in three studies, with myocardial infarction, angina pectoris, and ischemic stroke as disease endpoints. The main aim of the present thesis research was to provide additional epidemiologic evidence that elevated tHcy is an independent risk factor for cardiovascular disease. By using different study designs, different endpoints, and different populations, we tried to maximize our contribution. Three specific objectives were:
- To study the dose-response relationship between tHcy and cardiovascular disease, investigating whether there is a clear threshold level, above which risk is increased, or a graded relationship.
- To study whether fasting tHcy (associated with homocysteine remefhylation) or post-load tHcy (associated with homocysteine transsulfuration) shows a stronger association with risk of cardiovascular disease.
7 Chapter 1
- To study whether inadequate dietary intake or status of B-vitamins are possible causes of elevated tHcy in patients of cardiovascular disease.
Figure 1-2 schematically depicts the main investigated relationships and chapters in which they were studied.
Genetics Coronary atherosclerosis Ch.2 Myocardial infarction Ch.4 Angina pectoris Ch.5 Ischemic stroke Ch.6 Ch. 4 All CVD Ch. 7 B-vitamins
Figure 1-2. Associations studied in the thesis. MTHFR = 5,10-methylenetetrahydrofolate reductase; tHcy = plasma total homocysteine; CVD = cardiovascular diseases.
In Chapter 2, we describe results of a Dutch case-control study with angiographically-defined coronary atherosclerosis as disease endpoint. We relate tHcy to extent of coronary occlusion. Also, we study whether fasting or post-load tHcy shows a stronger relationship with the disease. Levels of B-vitamins, reflecting recent or long-term dietary intake, are studied as possible causes of tHcy elevation in cases, compared to controls. In Chapter 3, we compare the frequency of a 677C—>T mutation in the gene that codes for the MTHFR enzyme, associated with reduced enzyme activity,40 in patients of coronary atherosclerosis and controls. We describe relationships between genotype and plasma tHcy. Also, the interaction between this mutation and erythrocyte folate is discussed. In Chapter 4, fasting tHcy levels are related to risk of myocardial infarction, in a Boston area case-control study. The role of the B-vitamins (dietary intake and
8 General Introduction plasma levels) is investigated as well, including relationships of the vitamins themselves with risk of myocardial infarction. Angina pectoris is the disease endpoint in Chapter 5, showing results of a case-control study, nested within a prospective study (Physicians' Health Study [PHS]). Chapters 4 and 5 also describe levels of other compounds of homocysteine metabolism in cases and controls. These compounds may give indications of abnormalities in the pathways of homocysteine metabolism, either of genetic or nutritional origin.41 Chapter 6 discusses findings of another case-control study, nested within the PHS, with ischemic stroke as the disease outcome. Possible effect modification by hypertension, as suggested by a previous study,42 is studied. Chapter 7 concentrates on tHcy levels in women, using data of a large multi- center case-control study on tHcy and cardiovascular disease. Relationships of tHcy with risk of cardiovascular disease are compared for the sexes, and for premenopausal and postmenopausal women. Finally, in Chapter 8, we put our findings in perspective, by comparing them to those of other studies. We discuss methodologic issues, make suggestions for future research, and evaluate the possible consequences for public health.
References
1. Dutch Heart Foundation. Cardiovascular diseases in the Netherlands. Morbidity and mortality statistics. The Hague: Dutch Heart Foundation, 1994 (in Dutch).
2. Coronary heart disease epidemiology. From aetiology to public health. Marmot M, Elliott P, eds. New York, NY: Oxford University Press. 1992.
3. 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.
4. Ueland PM, Refsum H, Brattstrom L. Plasma homocysteine and cardiovascular disease. In: Francis RB Jr, ed. Atherosclerotic Cardiovascular Disease, Hemostasis, and Endothelial Function. New York, NY: Marcel Dekker Inc;1992:183-236.
5. Mudd SH, Finkelstein JD, Irreverre F, Laster L. Homocystinuria: an enzymatic defect. Science 1964;143:1443-5.
6. Boers GHJ, Fowler B, Smals AGH, et al. Improved identification of heterozygotes for homocystinuria due to cystathionine synthase deficiency by the combination of methionine loading and enzyme determination in cultured fibroblasts. Hum Genet 1985;69:164-9.
9 Chapter 1
7. Skovby F. Homocystinuria. Clinical, biochemical and genetic aspects of cystathionine B- synthase and its deficiency in man. Acta Paediatr Scand 1985;321:1-21.
8. Wilcken DEL, Wilcken B. The pathogenesis of coronary artery disease: a possible role for methionine metabolism. J Clin Invest 1976;57:1079-82.
9. Bergmark C, Mansoor MA, Swedenborg J, De Faire U, Svardal AM, Ueland PM. Hyperhomocysteinemia in patients operated for lower extremity ischaemia below the age of 50. Effect of smoking and extent of disease. Eur J Vase Surg 1993;7:391-6.
10. Dudman NPB, Wilcken DEL, Wang J, Lynch JF, Macey D, Lundberg P. Disordered methionine/homocysteine metabolism in premature vascular disease. Its occurrence, cofactor therapy, and enzymology. Arterioscler Thromb 1993;13:1253-60.
11. Mudd SH, Havlik R, Levy HL, McKusick VA, Feinleib M. A study of cardiovascular risk in heterozygotes for homocystinuria. Am J Hum Genet 1981;33:883-93.
12. Kluijtmans LAJ, Van den Heuvel LPWJ, Boers GHJ, et al. Molecular genetic analysis in mild hyperhomocysteinemia: a common mutation in the methylenetetrahydrofolate reductase gene is a genetic risk factor for cardiovascular disease. Am J Hum Genet 1996 (in press).
13. Boers GHJ, Smals AGH, Trijbels FJM, et al. Heterozygosity for homocystinuria in premature peripheral and cerebral occlusive arterial disease. N Engl J Med 1985;313:709-15.
14. Clarke R, Daly L, Robinson K, et al. Hyperhomocysteinemia: an independent risk factor for vascular disease. N Engl J Med 1991;324:1149-55.
15. Engbersen AMT, Franken DG, Boers GHJ, Stevens EMB, Trijbels FJM, Blom HJ. Thermolabile 5,10-Methylenetetrahydrofolate reductase as a cause of mild hyperhomocysteinemia. Am J Hum Genet 1995;56:142-50.
16. Daly L, Robinson K, Tan KS, Graham JM. Hyperhomocysteinemia: a metabolic risk factor for coronary heart disease determined by both genetic and environmental influences. Q J Med 1993;86:685-9.
17. Kang SS, Passen EL, Ruggie N, Wong PW, Sora H. Thermolabile defect of methylenetetrahydrofolate reductase in coronary artery disease. Circulation 1993;88:1463-9.
18. Boushey CJ, Beresford SAA, Omenn GS, Motulsky AG. A quantitative assessment of plasma homocysteine as a risk factor for vascular disease. Probable benefits of increasing folic acid intakes. JAMA 1995;274:1049-57.
19. Selhub J, Jacques PF, Wilson PWF, Rush D, Rosenberg IH. Vitamin status and intake as primary determinants of homocysteinemia in an elderly population. JAMA 1993;270:2693-8. General Introduction
20. Jacobsen DW, Gatautis VJ, Green R, et al. Rapid HPLC determination of total homocysteine and other thiols in serum and plasma: sex differences and correlation with cobalamin and folate concentrations in healthy subjects. Clin Chem 1994;40:873-81.
21. Brattstrom L, Israelsson B, Norrving B, et al. Impaired homocysteine metabolism in early- onset cerebral and peripheral occlusive arterial disease. Effects of pyridoxine and folic acid treatment. Atherosclerosis 1990;81:51-60.
22. Miller JW, Nadeau MR, Smith D, Selhub J. Vitamin B-6 deficiency vs. folate deficiency: comparison of responses to methionine loading in rats. Am J Clin Nutr 1994;59:1033-9.
23. Brattstrom LE, Israelsson B, Jeppsson JO, Hultberg BL. Folic acid~an innocuous means to reduce plasma homocysteine. Scand J Clin Lab Invest 1988;48:215-21.
24. Brattstrom LE, Hultberg BL, Hardebo JE. Folic acid responsive postmenopausal homocysteinemia. Metabolism 1985;34:1073-7.
25. Ubbink JB, Van der Merwe A, Vermaak WJH, Delport R. Hyperhomocysteinemia and the response to vitamin supplementation. Clinical Investig 1993;71:993-8.
26. Wilcken DE, Dudman NP, Tyrrell PA, Robertson MR. Folic acid lowers elevated plasma homocysteine in chronic renal insufficiency: possible implications for prevention of vascular disease. Metabolism 1988;37:697-701.
27. Arnadottir M, Brattstrom L, Simonsen O, et al. The effect of high-dose pyridoxine and folic acid supplementation on serum lipid and plasma homocysteine concentrations in dialysis patients. Clin Nephrol 1993;40:236-40.
28. Boers GHJ, Smals AGH, Drayer JIM, Trijbels FJM, Leermakers Al, Kloppenborg PW. Pyridoxine treatment does not prevent homocystinemia after methionine loading in adult homocystinuria patients. Metabolism 1983;32:390-7.
29. Franken DG, Boers GH, Blom HJ, Trijbels FJ, Kloppenborg PW. Treatment of mild hyperhomocysteinemia in vascular disease patients. Arterioscler Thromb 1994;14:465-70.
30. Brattstrom L, Israelsson B, Lindgarde F, Hultberg B. Higher total plasma homocysteine in vitamin B-12 deficiency than in heterozygosity for homocystinuria due to cystathionine B- synthase deficiency. Metabolism 1988;2:175-8.
31. Rees MM, Rodgers GM. Homocysteinemia: association of a metabolic disorder with vascular disease and thrombosis. Thromb Res 1993;71:337-59.
32. 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 USA 1994;91:6369-73.
33. Mudd SH, Levy HL. Plasma homocysteine or homocysteine? N Engl J Med 1995;333:325.
11 Chapter 1
34. Pancharuniti N, Lewis CA, Sauberlich HE, et al. Plasma homocysteine, folate, and
vitamin B12 concentrations and risk of early-onset coronary artery disease. Am J Clin Nutr 1994;59:940-8.
35. Arnesen E, Refsum H, B<|)naa KH, Ueland PM, Fcjirde OH, Nordrehaug JE. Serum total homocysteine and coronary heart disease. Int J Epidemiol 1995;24:704-9.
36. Dalery K, Lussier-Cacan S, Selhub J, Davignon J, Latour Y, Genest J. Homocysteine and
coronary artery disease in French Canadian subjects: relation with vitamins B12, B6, pyridoxal phosphate, and folate. Am J Cardiol 1995;75:1107-11.
37. Boers GH, Smals AG, Trijbels FJ, Leermakers AI, Kloppenborg PW. Unique efficiency of methionine metabolism in premenopausal women may protect against vascular disease in the reproductive years. J Clin Invest 1983;72:1971-6.
38. Malinow MR, Stampfer MJ. Role of plasma homocysteine in arterial occlusive disease. Clin Chem 1994;40:857-8.
39. Wilcken DEL, Reddy SG, Gupta VJ. Homocysteinemia, ischemic heart disease, and the carrier state for homocystinuria. Metabolism 1983;32:363-70.
40. Frosst P, Blom HJ, Milos R, et al. A candidate genetic risk factor for vascular disease: a common mutation in methylenetetrahydrofolate reductase. Nat Genet 1995;10:111-3.
41. Allen RH, Stabler SP, Savage DG, Lindenbaum J. Metabolic abnormalities in cobalamin
(vitamin B12) and folate deficiency. FASEB J 1993;7:1344-53.
42. Araki A, Yoshiyasu S, Fukushima Y, Matsumoto M, Asada T, Kita T. Plasma sulphydryl- containing amino acids in patients with cerebral infarction and in hypertensive patients. Atherosclerosis 1989;79:139-46.
12 Plasma Total Homocysteine, B-vitamins and Risk of Coronary Atherosclerosis
Abstract
Background & methods Epidemiologic research has shown that elevated total homocysteine (tHcy) in plasma is a risk factor for atherosclerotic disease. We investigated whether fasting or post-methionine loading tHcy level was a stronger predictor of coronary atherosclerosis. Furthermore, we studied levels of B-vitamins, involved in homocysteine metabolism. Subjects were recruited from men and women, aged 25 to 65 years, who underwent coronary angiography between June 1992 and June 1994. Cases (n=131) were defined as those with > 90% occlusion in one and > 40% occlusion in a second coronary artery, while control subjects (n=88) had < 50% occlusion in only one coronary vessel. In addition, a population-based control group (n=101) was studied. Results After adjusting for age and gender differences between the groups, cases had higher geometric mean fasting (9%, P=0.01) and post-load tHcy (7%, P=0.04) than the combined control groups. These differences in tHcy between cases and controls were not explained by vitamin status, since geometric mean levels of red cell folate and pyridoxal 5'-phosphate were higher among cases than among controls, whereas plasma vitamin BI2 was only slightly lower in cases. However, within the groups of cases and controls, the vitamins showed significant inverse correlations with tHcy. The odds ratio (OR) of severe coronary atherosclerosis (case status) for subjects with fasting plasma tHcy above 18.5 umol/L (the 95th percentile in the combined control groups), relative to those at or below that level was 1.3 (95% confidence interval [CI] 0.5-3.5) adjusted for sex, age, and other confounders. The corresponding OR for subjects with elevated post-load tHcy (> 59.3 umol/L) was 2.4 (95% CI 0.9-6.1). For each 1 SD increase in fasting (5 umol/L) as well as post-load tHcy (12 umol/L) the risk of coronary atherosclerosis increased borderline significantly with 20%. Conclusions Our data show a positive association between plasma tHcy and risk of coronary atherosclerosis, for both fasting and post-load levels. The data suggest that the association exists over a wide range of tHcy levels and that increased risk is not limited to subjects with elevated levels. Differences in vitamin status could not explain the higher plasma levels of tHcy among cases, suggesting that other factors determine their higher levels in this population.
Petra Verhoef, Frans J. Kok, Dick A.C.M. Kruyssen, Evert G. Schouten, Jacqueline CM. Witteman, Diederick E. Grobbee, Per M. Ueland, Helga Refsum (submitted).
13 Chapter 2
Introduction
Elevated plasma total homocysteine (tHcy) is an independent risk factor for atherosclerotic disease, in the coronary, cerebrovascular, and peripheral vessels.1 Several epidemiologic studies have shown a positive association of either fasting tHcy or levels in response to a methionine load with risk of coronary atherosclerosis.2"8 However, most studies included only a limited number of subjects and studied either fasting or post-load tHcy. Homocysteine is a sulfur amino acid, which is formed from the essential amino acid methionine. Homocysteine is either transsulfurated to cysteine via two vitamin B6-dependent reactions or is remethylated to methionine. Notably, in most cells and tissues the remethylation pathway depends both on vitamin Bn and folate, and low intakes of these vitamins are common causes of elevated plasma tHcy.9"11 It has been suggested that the fasting level may be determined by homocysteine remethylation, while increased post-load tHcy may reflect abnormalities in the transsulfuration pathway.12 In the present investigation, we compared both fasting and post-load plasma tHcy, as well as the increase after methionine loading, in groups of subjects with and without angiographically documented severe coronary occlusion, and apparently healthy subjects with no history of cardiovascular disease. We evaluated whether there was a graded risk of coronary atherosclerosis within the normal range of plasma tHcy, as opposed to a threshold effect, restricted to those with abnormally high levels. Blood levels of pyridoxal 5'-phosphate (PLP), vitamin BI2, and folate were compared for cases and control subjects, and the relationships with plasma tHcy were examined.
Methods
Study population A case-control study was conducted from June 1992 to June 1994. Cases and one control group were selected from patients aged 25 to 65 years, who underwent coronary angiography in the Zuiderziekenhuis Hospital in Rotterdam, the Netherlands. Subjects with either severe coronary occlusions (referred to as cases) or without substantial coronary occlusions (referred to as coronary controls) were included. A second control group was drawn from the general population and comprised subjects with no history of cardiovascular disease (referred to as population-based controls). Exclusion criteria for all groups were diabetic, renal,
14 Homocysteine and Coronary Atherosclerosis hepatic, thyroid or gastro-intestinal disease, cancer, alcohol or drug abuse, and psychiatric illness. At angiography, projections were made of the major coronary vessels using standard catheterization techniques. A trained research nurse selected potential cases and coronary controls, based on the angiography reports. Cases were defined as those having > 90% occlusion in one and > 40% occlusion in one additional coronary artery. Notably, 77.1% of the cases had > 70% occlusion in a second vessel. Coronary controls were defined as those having < 50% occlusion in only one coronary artery. The majority (79.5%), however, had no substantial coronary narrowing in all three arteries, whereas only 5.7% of them had 50% stenosis in a single coronary vessel. Thus, there was a marked contrast between cases and coronary controls, reducing the possibility of disease misclassification. Sixty seven (51.1%) cases and 6 (6.8%) coronary controls had a history of myocardial infarction. In the coronary controls, the myocardial infarctions were due to coronary spasms or other non-atherosclerotic causes. Since coronary atherosclerosis was the endpoint of interest, these controls were not excluded from analyses. The time between myocardial infarction and day of the methionine loading test was at least 2.5 months in all subjects. During the study period, a total of 2,659 patients underwent coronary angiography. Of these, 2,292 were not included mainly because of age over 65 years (n=1122, 49.0%), coronary occlusion outside ranges of case and control definition (n=486, 21.2%), or the presence of one or more other exclusion criteria. Of the 369 subjects that were invited, 353 (95.6%) could be reached, and of those 222 (62.9%) were willing to participate (131 cases, 91 coronary controls). Three of the 91 coronary controls that had originally participated, were excluded from analysis, because a second evaluation of the angiography results revealed too much coronary narrowing. We obtained a population-based control group, from a register of about 10,000 men previously enrolled for participation in a trial of cholesterol-lowering medication. The trial was never conducted, however. Among men with no prior history of cardiovascular disease or diabetes, a random sample of 152 were invited for participation. Fifteen could not be reached, 14 did not meet the inclusion criteria, and 47 were not interested, leaving 76 (61.3%) study subjects for participation. One participant was excluded from analysis because he reported diabetes at the interview. Spouses of 45 male participants were invited to participate of whom 12 were not eligible, 7 were not willing to participate, leaving 26 (78.8%) women for participation. Thus, a total of 101 population-based control subjects was studied. The study protocol was approved by the medical ethical committee. All
15 Chapter 2 participants gave their written informed consent.
Blood sampling and examination At the day of the examinations, venous blood samples were obtained from all subjects between 8:30 and 9:30 AM, after a 10 to 12 hour fast. L-methionine (0.1 g/kg body weight) mixed with orange juice was given orally, together with a standardized low protein breakfast. After breakfast, subjects were interviewed about smoking habits, alcohol consumption, and medication. Interviews were performed by the research nurse or P.V. Subjects received a standardized low protein lunch and were asked not to consume any protein-containing foods, like milk, cheese, or meat. Six hours after methionine administration, a second blood sample was drawn for estimation of plasma tHcy in response to methionine provocation. From one subject we did not obtain a blood sample after methionine loading. Duplicate blood pressure readings were taken before and after the methionine loading test with the subject seated after 5 minutes rest. Height and weight (without shoes and heavy clothing) were measured in the morning. For measurement of whole blood folate, 200 ul of EDTA blood was mixed with 4 ml (1:20) freshly prepared 1% (w/v) ascorbic acid solution. The rest of the EDTA blood, to be used for measurement of tHcy, creatinine, PLP, and cobalamin in plasma was placed on ice immediately and in the dark, and centrifuged at 4° C within one hour. Serum and plasma samples were stored at -80° C for a maximum of 6 months before analysis.
Biochemical analyses The first 60 subjects (15 cases, 20 coronary controls,- 25 population-based controls) of our study were included in an European case-control study, 'Hyperhomocysteinemia and Vascular Disease', which had biochemical analyses performed centrally. After tennination of this European study, we continued sending samples to these central laboratories. Plasma tHcy, which refers to the sum of protein-bound, free oxidized and reduced species of homocysteine in plasma, was determined by an automated assay, based on precolumn derivatization with monobromobimane, followed by high pressure liquid chromatography (HPLC) separation and fluorescence detection. The assay was performed at the Department of Clinical Biology, Division of Pharmacology, University of Bergen, Norway and has a coefficient of variation of 3%. Estimations were performed in duplicate. All other determinations were performed at the laboratory of MIMELAB-AB, Soraker, Sweden. For estimation of plasma PLP, enzymatic photometry with HPLC
16 Homocysteine and Coronary Atherosclerosis
separation was used, whereas folate in whole blood and serum plasma vitamin B12 (as cobalamin) were determined by radio immunoassay. We expressed folate concentration per haematocrit, referred to as erythrocyte folate. Creatinine, total cholesterol and HDL cholesterol (after precipitation of LDL and VLDL) were determined in serum with enzymatic photometry.
Data analysis Current smoking was defined as the use of any tobacco at the time of catheterization or at the day of methionine loading (population-based controls). Subjects were diagnosed as hypertensive at a systolic blood pressure > 160 mmHg or diastolic blood pressure > 95 mmHg (measured at day of methionine loading), or when they were using anti-hypertensive medication. Hypercholesterolemia was defined as serum cholesterol > 6.5 mmol/L or use of cholesterol-lowering drugs. Differences in cardiovascular risk factor levels between cases and controls were tested with Student's t test for continuous variables and Pearson's chi-square test for frequency measures. To evaluate potential confounding, we studied associations of the cardiovascular risk factors with plasma tHcy levels among the combined control groups by calculating Spearman correlation coefficients. By subtracting fasting plasma tHcy levels from post-load tHcy levels, we calculated the post-load increase in tHcy for each subject. Means, standard deviations (SDs) and geometric means of tHcy, creatinine and vitamins were calculated for all three study groups. Differences between cases and controls (separate and combined control groups) were examined, adjusting for age and gender by means of linear regression analysis with the variable of interest (i.e. tHcy, vitamins, or creatinine) as the dependent variable and case-control status, age, and gender as the independent variables. By means of logistic regression analysis, the odds ratios (ORs) of severe coronary atherosclerosis (case status) were calculated for those with elevated fasting, post-load, or post-load increase tHcy levels, defined as levels above the 95th percentile of the combined control groups. Subjects with levels at or below the 95th percentile were considered as the reference group. To evaluate a possible graded association of plasma tHcy with coronary atherosclerosis, we computed the ORs per 1 SD increase of plasma tHcy. Multivariate logistic regression analysis was used to simultaneously adjust ORs for age, gender, and other confounding factors. Additionally, we calculated Spearman correlation coefficients of the vitamins with plasma tHcy levels among cases and controls. All reported P-values are two-tailed.
17 Chapter 2
Results
Characteristics of study groups Table 2-1 shows the main characteristics of the cases and the two control groups. The percentage of males and mean age were higher in cases than in both control groups. The majority of subjects in all groups was aged 40 to 60 years. Mean total cholesterol levels were similar among the groups, but cases had significantly lower serum HDL cholesterol levels. The proportion of subjects with hypertension was highest among cases. Mean serum level of triglycerides and body mass index were highest in cases, whereas alcohol consumption was highest in controls. At the time of the examination, a lower percentage of cases was currently smoking (24.4%), but before hospitalization for angiography the percentage of current smokers was about the same in cases and the two control groups. Mean pack years of smoking was higher in cases compared to both control groups.
Table 2-1. Characteristics of cases with severe coronary atherosclerosis and two groups of controls Cases Coronary Population (n=131) controls controls (n=88) (n=101) Age (years) 52.5 ± 7.5 48.2 ± 8.0* 49.9 ± 6.9+ Gender (% male) 84.7 59.1" 74.3 Body mass index (kg/m2) 26.8 ± 3.0 26.0 ± 3.6 26.0 ± 3.7 Total cholesterol (mmol/L) 6.8 ± 1.4 ' 6.8 ± 1.7 6.6 ± 1.7 Total/HDL cholesterol 7.0 ±2.1 5.9 ± 2.5* 6.0 ± 2.7+ Hypercholesterolemic (%) 68.7 61.4 48.5+ Triglycerides (mmol/L) 2.1 ± 1.3 1.7 ± 1.0* 1.4 ± 1.0+ Systolic blood pressure (mmHg) 135.0 ± 13.3 131.1 ± 16.9 134.7 ± 14.4 Diastolic blood pressure (mmHg) 81.3 ± 7.4 78.3 ± 9.3* 81.7 ± 8.0 Hypertensive (%) 85.5 51.1* 16.8+ Currently smoking (%) 38.2 39.8 36.6 Pack years (years) 31.9 ± 27.1 22.4 ± 21.1* 21.6 + 22.6+ Alcohol consumption (glasses/day) 0.8 ± 1.3 1.1 ± 1.7 1.3 ± 1.7+
* P < 0.05, cases versus coronary control subjects. + P < 0.05, cases versus population-based control subjects. Tested with Student's t test for continuous variables (mean ± SD is shown) and Pearson's chi- square test for frequencies.
18 Homocysteine and Coronary Atherosclerosis
30
25 "
8 10 12 14 16 18 >20 % of total
25
Plasma total homocysteine (umol/L)
Figure 2-1. Frequency distribution of fasting and post-load plasma total homocysteine levels (tHcy) among 131 cases with severe coronary atherosclerosis and 189 control subjects (88 coronary controls and 101 population-based controls).
19 Chapter 2
Plasma tHcy levels of study groups Mean and geometric mean levels of plasma tHcy were higher in cases than in both control groups, for both fasting and post-load levels (Table 2-2). Spearman rank correlations between fasting and post-load tHcy levels were 0.56 (P=0.0001) and 0.66 (P=0.0001) among cases and controls, respectively. Thus, since the post- load tHcy level is a combination of the fasting level and the increase after methionine loading, we also investigated the increase after loading (i.e. post-load minus fasting level). This was also highest among the cases. Distributions of tHcy levels were very similar among coronary and population-based controls, therefore the groups were combined in all subsequent analyses. Cases had significantly higher geometric mean fasting and post-load tHcy than the combined control groups, after adjusting for age and gender (Table 2-2). Figure 2-1 shows that for both fasting and post-load tHcy, the distribution among cases was shifted towards the right across the full range of values, compared to controls. This was slightly more apparent for fasting than for post-load tHcy levels. For post-load levels there was an additional upper shift at the far right end among cases. For men separately, similar distributions were observed (data not shown).
Evaluation of confounding by age, gender and coronary risk factors Age was not associated with fasting or post-load tHcy in the combined control groups. Men (n=127) had a 12% higher (P=0.008) geometric mean fasting plasma tHcy level than women (n=62). Geometric mean post-load tHcy and post- load increase of tHcy were 5% (P=0.31) and 12% (P=0.03) higher in women than in men. This was due to a marked response to the methionine loading test in a subfraction of the women, as has been reported previously by others.13,14 Body mass index was positively associated with post-load tHcy (r=0.15, P=0.04) and post-load increase in tHcy (?=0.18, P=0.01). Furthermore, fasting tHcy correlated inversely with alcohol consumption (r=-0.15, P=0.04). Creatinine, which was higher among cases than among controls (Table 2-2), correlated positively with fasting tHcy (r=0.29, P=0.0001) and post-load tHcy (r=0.16, P=0.02) among controls. Smoking, blood pressure, and triglycerides did not correlate with tHcy. Use of anti• hypertensive drugs and lipid-lowering drugs were not related to tHcy levels either. Controls using acetylsalicylic acid or anti-thrombotic drugs (31 coronary controls and 1 population-based control) had 13% lower fasting (P=0.03) and 12% lower post-load tHcy (P=0.04), than control subjects not taking those drugs. Among cases, the majority (92.4%) of the subjects were using these drugs. By matter of convention, age and gender were controlled for in all analyses.
20 Homocysteine and Coronary Atherosclerosis
Furthermore, creatinine and alcohol consumption were controlled for in risk analyses for fasting tHcy and post-load tHcy. Body mass index was controlled for in risk analyses for post-load tHcy and post-load increase tHcy. Since use of acetylsalicylic acid and anti-thrombotic drugs is a consequence of the disease, and hypothetically a consequence of high tHcy levels, it was not considered a true confounder.
Table 2-2. Concentrations of total homocysteine, creatinine, and B-vitamins in cases with severe coronary atherosclerosis and two groups of controls Cases Coronary Population % Case- (n=131) controls controls control (n=88) (n=101) difference* Fasting tHcy (umol/L) 13.7 ± 6.6 11.8 ± 3.6 12.5 ± 5.6 9 (0.01) 12.9 11.4 11.8 Post-load tHcy (umol/L) 41.1 ± 12.6 37.9 ± 11.2 38.9 ± 12.8 7 (0.04) 39.5 36.4 37.3 Post-load increase in tHcy 27.5 ± 9.8 26.1 ± 9.9 26.5 ± 9.1 6 (0.15) (umol/L) 25.9 24.6 25.7 Creatinine (umol/L) 76.9 ± 12.2 72.3 ± 12.8 73.5 ± 11.2 3 (0.11) 75.9 71.1 72.2 Folate (nmol/L) 880 ± 274 780 ± 284 760 ± 282 10 (0.02) 838 734 712
Vitamin B12 (pmol/L) 259 ± 98 287 ± 102 248 ± 97 -1 (0.84) 243 277 229 PLP (nmol/L) 28.8 ± 11.4 29.3 ± 10.8 24.7 ± 8.3 8 (0.08) 26.6 27J 23.4
Means ± SD and geometric means (italics) are shown. tHcy, total homocysteine; PLP, pyridoxal 5'-phosphate. * Age- and gender-adjusted percentage difference in geometric means between cases and combined control groups.
Plasma tHcy and risk of coronary atherosclerosis Table 2-3 shows the number of cases and controls with elevated tHcy levels. A total of 16 (12.2%) cases and 16 (8.5%) controls had either elevated fasting tHcy or elevated post-load tHcy. Of the 22 subjects with elevated post-load tHcy, eight were also defined as having elevated fasting tHcy.
21 Chapter 2
ORs of severe coronary atherosclerosis for subjects with elevated tHcy, relative to subjects with levels at or below the cutoff-points are given in Table 2-3. Age- and gender-adjusted ORs are shown. In addition, ORs were adjusted for body mass index, alcohol consumption, and plasma creatinine where appropriate. Elevated post-load levels were associated with a 2.4-fold increased risk of coronary atherosclerosis, whereas the risks associated with elevated fasting levels or elevated post-load increase levels were lower. The increase in risk of coronary atherosclerosis for each 1 SD increase in fasting tHcy was on average 20%, similarly to the OR associated with 1 SD increase in post-load tHcy or post-load increase tHcy (Table 2-3).
Table 2-3. Odds ratios of severe coronary atherosclerosis for subjects with elevated plasma total homocysteine and per 1 SD increase in plasma total homocysteine Fasting Post-load Increase levels levels after load Elevated tHcy: > 18.5 umoI/L > 59.3 umol/L > 44.8 umol/L
Cases / controls (n) 8 / 10 12/ 10 8 / 10 Cases / controls (%) 6.1 /5.2 9.2 / 5.2 6.1 /5.2
Age- and gender-adjusted* 1.3 (0.5 - 3.5) 2.1 (0.8 - 5.3) 1.4 (0.5 - 3.8) OR (95% CI) Multivariate-adjusted 1.3 (0.5 - 3.5)+ 2.4 (0.9 - 6.1)+ 1.4 (0.5 - 3.8)§ OR (95% CI)
Per 1 SD increase in tHcy: Per 5 umoI/L Per 12 umol/L Per 10 umol/L
Age- and gender-adjusted 1.3 (1.0 - 1.6) 1.3 (1.0 - 1.6) 1.1 (0.9 - 1.5) OR (95% CI) Multivariate-adjusted 1.2 (1.0 - 1.5)+ 1.2 (1.0 - 1.6)* 1.2 (0.9 - 1.5)8 OR (95% CI) tHcy, total homocysteine; SD, standard deviation; OR, odds ratio; CI, confidence interval. * Multivariate logistic regression, adjusting for age and gender. + Adjusted for age, gender, creatinine, and alcohol consumption. + Adjusted for age, gender, creatinine, and alcohol consumption, and body mass index. 5 Adjusted for age, gender, and body mass index.
22 Homocysteine and Coronary Atherosclerosis
Although use of acetylsalicylic acid and other anti-thrombotic drugs was not a true confounder, we included the variable into the multivariate logistic model (for risk analyses per 1 SD tHcy only). Standard errors of the estimates increased by approximately 50%. The ORs (plus 95% CIs) per 1 SD increase in tHcy became 1.5 (95% CI, 1.1-2.2), 1.7 (95% CI, 1.2-2.5), and 1.6 (95% CI, 1.1-2.4), for fasting, post-load, and post-load increase tHcy levels, respectively. These results show that use of these drugs by the majority of cases has weakened the association of tHcy with risk of coronary atherosclerosis.
Degree of coronary atherosclerosis To study whether tHcy related to the number of significantly occluded coronary arteries (0, 1, 2, or 3), we regrouped all 219 angiography patients into four groups (Table 2-4), in a similar way as previously done in epidemiologic research on the association between tHcy and risk of coronary atherosclerosis.5,7 Adjusting for age and gender, there were significant increases in plasma tHcy with increasing number of occluded coronary arteries, both for fasting (P=0.03) and post-load tHcy (P=0.02).
Table 2-4. Plasma total homocysteine, age and gender among subjects with varying severity of coronary atherosclerosis
0 1 2 3 (n=83) (n=10) (n=69) (n=57)
Age (years) 48.1 ± 7.9 48.9 ± 9.6 52.0 ± 8.1+ 53.4 ± 6.6+ Gender (% male) 56.6 90.0+ 85.5+ 84.2+
Fasting tHcy (umol/L) 11.7 ± 3.6 12.7 ± 2.6 13.6 ± 5.3 14.0 ± 8.1 77.2 12.5 72.9+ 13.0 Post-load tHcy (umol/L) 37.2 ± 10.9 43.8 ± 10.3 40.6 ±11.6 42.0 ± 14.1 35.8 42.9* 39.7* 40.7*
Means ± SD and geometric means (italics) are shown. tHcy, total homocysteine. * Group 0 consisted of patients with less than 50% occlusion in any of the three major coronary arteries, group 1 consisted of subjects, who had at least 50% narrowing in only one coronary artery. In group 2, all subjects had at least 50% occlusion in two coronary arteries, whereas in group 3 all three coronary vessels had occlusions of at least 50%. + P-value < 0.05, r test for age and chi-square test for gender comparison, group 0 as a reference. * P-value < 0.05 for age- and gender-adjusted differences in geometric means between each of the groups and group 0.
23 Chapter 2
Myocardial infarction Fasting tHcy did not differ between the cases with and without a history of myocardial infarction. For post-load tHcy, the geometric mean was 12% lower among those who had suffered a myocardial infarction relative to those who had not, after adjustment for age and gender (P=0.02). Levels of tHcy did not differ between the 6 coronary controls with a history of myocardial infarction, and the other coronary controls. When these 6 controls were left out of the analyses, all estimates remained virtually unchanged.
B-vitamins Population-based controls had lowest mean levels of plasma PLP, plasma vitamin B,2, and erythrocyte folate (Table 2-2, lower half). Compared to coronary controls, cases had slightly lower plasma levels of PLP and vitamin B12, but higher erythrocyte folate levels. When pooling the control groups and adjusting for age and gender differences, cases had higher geometric mean levels of erythrocyte folate and plasma PLP than controls, whereas plasma vitamin B12 was slightly lower. The frequency distribution of erythrocyte folate for cases showed a total shift towards the right relative to the distribution for all control subjects, possibly indicating a better overall nutritional folate status in cases (data not shown). The higher folate levels among cases, in comparison with the coronary controls, may have been a result of more frequently changed dietary habits after start of the cardiologic treatment (39.7% in cases versus 22.0% in coronary controls, P < 0.005). Most dietary regimens were meant to lower fat intake, but an increased vegetable or fruit intake may have occurred as well. We did not observe differences in tHcy levels and vitamin levels between cases who had and those who had not changed their diets. However, among coronary controls who had changed their diets, we observed a 17% (P=0.10) higher erythrocyte folate level and a 13% lower post-load tHcy (P=0.09) than among those who did not. Levels of all vitamins correlated inversely with plasma fasting and post-load tHcy, in a similar way for cases and controls. In the group of combined control subjects, the correlation coefficients of fasting plasma tHcy with the vitamins were
-0.22 (P=0.003) for erythrocyte folate, -0.26 (P=0.0004) for plasma vitamin B12, and -0.28 (P=0.0001) for plasma PLP. However, the associations between the vitamins and post-load tHcy were to a large extent explained by the strong intercorrelation of fasting and post-load tHcy, and only plasma PLP correlated inversely with the increase in plasma tHcy after a methionine load (^=-0.15, P=0.04).
24 Homocysteine and Coronary Atherosclerosis
Discussion Our study showed that plasma tHcy is positively associated with risk of severe coronary atherosclerosis, independently of other risk factors for coronary artery disease. The association appeared to be graded, and of similar strength for both fasting and post-load tHcy, except for the fact that increased risk associated with elevated tHcy was observed only for post-load levels. Overall, our analyses suggested that post-load increase in plasma tHcy (i.e. post-load minus fasting tHcy) was somewhat less strongly related to coronary atherosclerosis than absolute post- load tHcy. This may either be the result of double exposure measurement error, possibly causing effect attenuation, or indicate that part of the association observed for post-load tHcy could be attributed to fasting tHcy. Statistical significance was not reached for the OR associated with elevated post-load tHcy, possibly because frequencies of subjects with levels above the 95th percentile of controls were limited. With data of another study15 on coronary artery disease that provided frequencies of subjects with elevated post-load tHcy (> 95fh percentile of controls), we calculated a crude OR of 2.8, concurring with the one in our study. For the risk analyses per 1 SD increase in tHcy, findings were of borderline significance. They are somewhat smaller than those of other case-control studies, showing ORs of 1.3 (95% CI, 1.1-1.7) per 4 umol/L increase in fasting plasma tHcy16 and 1.4 (95% CI, 1.0-2.0) per quartile increase of fasting plasma tHcy,6 respectively, after adjustment for confounding. However, considering the clear shift to the right of the tHcy distributions in cases and the size of the estimate for elevated post-load tHcy, we feel confident to conclude that plasma tHcy shows a positive association with risk of severe coronary atherosclerosis. Several retrospective case-control studies on the association of plasma tHcy with angiographically defined coronary atherosclerosis have demonstrated positive associations, with fasting3"8 or post-load tHcy levels.2 One study has shown a graded association of fasting tHcy with risk of coronary atherosclerosis.6 Our study is the first one with relatively large numbers of cases and controls to demonstrate that both fasting and post-load tHcy levels are positively related to risk of coronary atherosclerosis, over a wide range of values. Our data concur with a study of peripheral vascular disease14 showing that frequency distributions of both fasting and post-load tHcy were displaced to the right in cases relative to control subjects. With few exceptions,17"19 most epidemiologic studies have shown a positive relation between plasma tHcy and risk of myocardial infarction or ischemic coronary disease,15,16,20 23 one of which showed a graded risk of myocardial infarction with increasing levels of fasting tHcy, without a clear threshold.16
25 Chapter 2
The methionine loading test has mainly been used to detect subjects with heterozygosity for cystathionine 6-synthase deficiency,24 but recent data suggest that elevated post-load tHcy is often due to other factors as well.25 Only eight of the 22 subjects (12 cases, 10 controls) with elevated post-load tHcy did also have elevated fasting tHcy. This indicates that by measuring fasting tHcy alone a fair number of subjects with elevated post-load tHcy, which were at increased risk of coronary atherosclerosis in our study, would have stayed undetected, as previously suggested by others.26
We found that levels of PLP, vitamin B,2, and erythrocyte folate were inversely correlated with fasting tHcy, whereas only PLP was inversely related to increase in tHcy after methionine loading. These findings support the hypothesis that fasting level may be determined by homocysteine remethylation, while increased post-load tHcy may reflect abnormalities in the txanssulfuration pathway.9,12 However, we found vitamin levels not to be lower among cases, as compared to combined controls, despite higher tHcy levels in cases. Similar observations were done in another study.8 This raises the possibility that genetic influences or other unmeasured factors are partly responsible for the observation of elevated tHcy levels in cases. For example, a higher prevalence of subjects with a mutation in the enzyme methylenetetrahydrofolate reducase (MTHFR) among cases, as shown in previous studies,25,27 could explain both higher mean fasting and post- load tHcy levels among cases. Controls selected from angiography patients are not representative of all non- diseased persons in the source population. We therefore included controls from the general population, without clinical symptoms of coronary artery disease. Notably, estimation of risks and confidence intervals were similar for the two control series and they were therefore combined to increase the power of the study. The lower vitamin status of population-based controls may in part be accounted for by the fact that both cases and coronary controls may have adopted healthier dietary habits or lifestyle, after start of cardiologie treatment. The frequency of dietary measures after start of the cardiologie treatment was twice as high in cases as in coronary controls, which may have given rise to bias. Fasting tHcy level was not different in those who had and those who had not changed dietary habits. However, in coronary controls who changed their diets, erythrocyte folate levels were higher and post-load tHcy levels were lower than in those who reported to have maintained their previous habits. If similar changes have occurred in cases, this finding may explain the observation of an overall better nutritional folate status in cases than in coronary controls. This may have attenuated the associations of tHcy with risk of coronary atherosclerosis.
26 Homocysteine and Coronary Atherosclerosis
The marked use of drugs in cases is a potential bias, but neither anti• hypertensive drugs nor lipid-lowering drugs appeared to be related to tHcy levels. Conversely, among coronary controls use of acetylsalicylic acid or other anti• thrombotic drugs was associated with significantly lower fasting and post-load tHcy levels. The use of these drugs was high among cases, which may have lead to an underestimation of the effect of tHcy on risk of coronary atherosclerosis in our population. This was indeed indicated by the analyses with use of these drugs included as a covariate in the multivariately adjusted model. A clear advantage of using angiographically-defined coronary atherosclerosis as a disease endpoint is the possibility to grade the vascular disease. Others5'7 have shown that fasting tHcy increased with increasing number of occluded coronary vessels. In our study, we did the same observation, both for fasting and post-load tHcy. This reinforces the hypothesis that elevated tHcy plays a role in the atherosclerotic process. Fasting plasma tHcy was similar and post-load tHcy was lower among cases who had suffered a myocardial infarction, in comparison with those who had not. This suggests an atherogenic effect, independently of a possible thrombogenic effect. The mechanisms for the thrombogenic and atherogenic properties of the amino acid homocysteine are unknown, and hypotheses have mainly been generated by experimental and in vitro studies, mostly in animals. Mechanisms include direct effects on the vascular endothelium, subsequently affecting its anti-thrombotic properties. Furthermore, endothelial cell damage may cause increased uptake of modified LDL cholesterol in the vascular wall.9'28 A recently proposed atherogenic mechanism involves inhibition of growth of endothelial cells and stimulation of proliferation of smooth muscle cells by homocysteine, finally leading to thickening of arterial walls.29 In conclusion, elevated plasma tHcy, fasting and after a methionine load, is an independent risk factor for severe coronary atherosclerosis. The association exists over a wide range of tHcy levels. The methionine loading test may identify additional subjects that are at high risk. The observation that levels of PLP, vitamin
B12 and folate were not lower among cases than among controls in this population, indicates that other important determinants of tHcy should be considered. Thus, further studies are necessary to complement the growing evidence of its important role in cardiovascular disease etiology and to further elucidate the role of genetic, nutritional, and lifestyle factors. In our opinion, reduction of tHcy levels in the general population, and not only in those with clearly abnormal tHcy levels, may reduce the incidence of cardiovascular disease.
27 Chapter 2
Acknowledgments The study was supported by a grant from the Netherlands Organization for Scientific Research. The authors would like to express their gratitude to Drs A.A.A. Bak, S.C. Balduw, G.J. van Beek, M.P. Freericks, F.M.A. Harms, R. van Mechelen, W.M. Muijs van de Moer, and R. Wardeh for their support in selecting the participants. We are very grateful to Annelies Legters, the research assistant. Furthermore, we would like to thank Mariette Penning, Halvard Bergesen, Elfrid Blomdal, Wenche Breyholtz, and coworkers from the Sticares Foundation, for assistance in the research.
References
1. Malinow MR. Homocysteine and arterial occlusive diseases: a mini review. Clin Chem 1994;40:173-6.
2. Wilcken DEL, Wilcken B. The pathogenesis of coronary artery disease: a possible role for methionine metabolism. J Clin Invest 1976;57:1079-82.
3. Kang SS, Wong PW, Cook HY, Norusis M, Messer JV. Protein-bound homocysteine. A possible risk factor for coronary artery disease. J Clin Invest 1986;77:1482-6.
4. Genest JJ Jr, McNamara JR, Salem DN, Wilson PW, Schaefer EJ, Malinow MR. Plasma homocysteine levels in men with premature coronary artery disease. I Am Coll Cardiol 1990;16:1114-9.
5. Ubbink JB, Vermaak WHJ, Bennett JM, Becker PJ, Van Staden DA, Bissbort S. The prevalence of homocysteinemia and hypercholesterolemia in angiographically defined coronary heart disease. Klin Wochenschr 1991;69:527-34.
6. Pancharuniti N, Lewis CA, Sauberlich HE, et al. Plasma homocysteine, folate, and
vitamin B,2 concentrations and risk of early-onset coronary artery disease. Am J Clin Nutr 1994;59:940-8.
7. Von Eckardstein A, Malinow MR, Upson B, et al. Effects of age, lipoproteins, and hemostatic parameters on the role of homocysteinemia as a cardiovascular risk factor in men. Arterioscler Thromb 1994;14:460-4.
8. Dalery K, Lussier-Cacan S, Selhub J, Davignon J, Latour Y, Genest J. Homocysteine and
coronary artery disease in French Canadian subjects: relation with vitamins B]2, B6, pyridoxal phosphate, and folate. Am J Cardiol 1995;75:1107-11.
28 Homocysteine and Coronary Atherosclerosis
9. Ueland PM, Refsum H, Brattstrom L. Plasma homocysteine and cardiovascular disease. In: Francis RB Jr, ed. Atherosclerotic Cardiovascular Disease, Hemostasis, and Endothelial Function. New York, NY: Marcel Dekker Inc; 1992:183-236.
10. Mason JB, Miller JW. The effects of vitamins B]2, B6, and folate on blood homocysteine levels. Ann N Y Acad Sci 1992;30:197-204.
11. Selhub J, Jacques PF, Wilson PWF, Rush D, Rosenberg IH. Vitamin status and intake as primary determinants of homocysteinemia in an elderly population. JAMA 1993;270:2693-8.
12. 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.
13. Andersson A, Brattstrom L, Israelsson B, Isaksson A, Hamfelt A, Hultberg B. Plasma homocysteine before and after methionine loading with regard to age, gender, and menopausal status. Eur J Clin Invest 1992;22:79-87.
14. Mansoor MA, Bergmark C, Svardal AM, L(|)nning PE, Ueland PM. Redox status and protein binding of plasma homocysteine and other aminothiols in patients with early-onset peripheral vascular disease. Arterioscler Thromb Vase Biol 1995;15:232-40.
15. Israelsson B, Brattstrom LE, Hultberg BL. Homocysteine and myocardial infarction. Atherosclerosis 1988;71:227-33.
16. Arnesen E, Refsum H, B(|)naa KH, Ueland PM, F<|)rde OH, Nordrehaug JE. Serum total homocysteine and coronary heart disease. Int J Epidemiol 1995;24:704-9.
17. Wilcken DEL, Reddy SG, Gupta VJ. Homocysteinemia, ischemic heart disease, and the carrier state for homocystinuria. Metabolism 1983;32:363-70.
18. Boers GHJ, Smals AGH, Trijbels FJM, et al. Heterozygosity for homocystinuria in premature peripheral and cerebral occlusive arterial disease. N Engl J Med 1985;313:709-15.
19. Alfthan G, Pekkanen J, Jauhiainen M, et al. Relation of serum homocysteine and lipoprotein(a) concentrations to atherosclerotic disease in a prospective Finnish population based study. Atherosclerosis 1994;106:9-19.
20. Malinow MR, Sexton G, Averbuch M, Grossman M, Wilson D, Upson B. Homocysteine in daily practice: levels in coronary artery disease. Coronary Artery Dis 1990;1:215-20.
21. Clarke R, Daly L, Robinson K, et al. Hyperhomocysteinemia: an independent risk factor for vascular disease. N Engl J Med 1991;324:1149-55.
22. 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.
29 Chapter 2
23. Graham I. Interactions between homocysteinaemia and conventional risk factors in vascular disease. Eur Heart J 1994; 15:530 (abstract).
24. Boers GHJ. Homocystinuria, homozygosity versus heterozygosity. Thesis, University of Nijmegen: ICG Printing BV, Dordrecht, 1985.
25. Engbersen AMT, Franken DG, Boers GHJ, Stevens EMB, Trijbels FJM, Blom HJ. Thermolabile 5,10-MethyIenetetrahydrofolate reductase as a cause of mild hyperhomocysteinemia. Am J Hum Genet 1995;56:142-50.
26. Daly L, Meleady R, Graham I. Fasting or post-methionine load homocysteine: which should be measured in relation to vascular risk? Irish J Med Sci 1995;164(S15):6 (abstract).
27. Kang SS, Passen EL, Ruggie N, Wong PW, Sora H. Thermolabile defect of methylenetetrahydrofolate reductase in coronary artery disease. Circulation 1993;88:1463-9.
28. Rees MM, Rodgers GM. Homocysteinemia: association of a metabolic disorder with vascular disease and thrombosis. Thromb Res 1993;71:337-59.
29. 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 USA 1994;91:6369-73.
30 Combination of Mutated 3 Methylenetetrahydrofolate Reductase and Low Folate Status is associated with High Plasma Total Homocysteine
Abstract Background & methods Thermolability of the enzyme 5,10- methylenetetrahydrofolate reductase (MTFÍFR), due to a 677C->T mutation at the MTHFR locus, is a cause of elevated plasma total homocysteine (tHcy), and may therefore be associated with coronary artery disease. We investigated the frequency of this mutation and its associations with plasma tHcy in 131 cases with severe coronary occlusions, 87 control subjects with angiographically defined normal coronarles, and 100 population- based control subjects. Males and females, aged 25-65 were studied. Results The frequency of homozygosity (+/+) for the MTHFR mutation in cases (10.0%) was not significantly different from that observed in coronary control subjects (11.5%), and in population-based controls (7.0%). In the total study population, plasma fasting tHcy levels were 36% higher and plasma tHcy after methionine loading 25% higher in (+/+) subjects than in homzygous normal subjects, whereas heterozygous mutant subjects had intermediate levels (P = 0.001 for both tests for linear trend). Similar trends were observed within each of the three study groups. For all genotypes, but especially in (+/+) subjects, we observed that erythrocyte folate below the median of the total population (790 nmol/L) was associated with increased tHcy. This finding was most pronounced for fasting tHcy. The (+/+) subjects with low erythrocyte folate had a 77% (95% confidence interval, 27% - 144%) higher geometric mean fasting tHcy (21.4 umol/L) than those with high erythrocyte folate (12.1 umol/L). Conclusions Our study indicates that homozygosity for the MTHFR mutation in combination with low folate status predisposes to particularly high levels of tHcy, and thereby increases risk of coronary artery disease.
Petra Verhoef, Frans J. Kok, Leo A.J. Kluijtmans, Henk J. Blom, Helga Refsum, Per M. Ueland, Dick A.C.M. Kruyssen (submitted).
31 Chapter 3
Introduction
Elevated plasma total homocysteine (tHcy) is an independent risk factor for atherosclerotic vascular disease,1,2 and may partly be genetically determined,3,4 Homocysteine is formed from methionine, and is subsequently catabolized in the vitamin B6-dependent transsulfuration pathway or remethylated to methionine. This latter reaction is catalyzed by the enzyme methionine synthase, which requires 5- methyltetrahydrofolate (5-methyi-THF) as substrate and vitamin B12 as cofactor. 5- Methyl-THF is formed by the reduction of 5,10-methylene-THF, through the action of 5,10-methylene-THF reductase (MTHFR).5 Consequently, deficiencies of MTHFR result in elevation of plasma tHcy. In 1988, Kang et al. discovered a variant of the MTHFR enzyme, characterized by a specific enzyme activity of approximately 50% of the normal activity. The enzyme appeared to be fhermolabile, providing an opportunity to distinguish between this variant and the normal enzyme found in the majority of the population.6,7 Furthermore, they described that thermolability of MTHFR was associated with raised tHcy levels, and increased risk of coronary artery disease.8,9 Engbersen et al.10 found that in 28% of hyperhomocysteinemic vascular patients, abnormal homocysteine metabolism could be attributed to fhermolabile MTHFR. Recently, a 677C—>T mutation was detected in the MTHFR gene, and homozygosity for this mutation was associated with decreased specific enzyme activity, increased thermolability, and elevated tHcy." In previous analyses, we found evidence for a positive association between plasma tHcy and risk of coronary atherosclerosis.12 In the present investigation, we compared the genotype frequency of the 677C->T mutation between groups of subjects with severely occluded coronaries and normal coronaries, and control subjects from the general population. Associations between genotype and tHcy level, before and after methionine loading, were studied. Since an adequate folate status may counterbalance the defective production of 5-methyl-THF in subjects with mutated MTHFR, we also studied the interaction between genotype and erythrocyte folate level.
Methods
Study population A case-control study was conducted from June 1992 to June 1994. Cases and one control group were selected from patients aged 25 to 65 years, who underwent
32 Thermolabile MTHFR, Folate, and Homocysteine coronary angiography in the Zuiderziekenhuis Hospital in Rotterdam,- the Netherlands. Subjects with either severe coronary occlusions (referred to as cases) or without substantial coronary occlusions (referred to as coronary controls) were included. A second control group was drawn from the general population and comprised subjects with no history of cardiovascular disease (referred to as population-based controls). Exclusion criteria for all groups were diabetic, renal, hepatic, thyroid or gastro-intestinal disease, cancer, alcohol or drug abuse, and psychiatric illness. At angiography, projections were made of the major coronary vessels using standard catheterization techniques. A trained research nurse selected potential cases and coronary controls, based on the angiography reports. Cases were defined as those having > 90% occlusion in one and > 40% occlusion in one additional coronary artery. Notably, 77.1% of the cases had > 70% occlusion in a second vessel. Coronary controls were defined as those having < 50% occlusion in only one coronary artery. The majority (79.5%), had no substantial coronary narrowing in all three arteries, whereas only 5.7% of them had 50% stenosis in a single coronary vessel. Thus, there was a marked contrast between cases and coronary controls, reducing the possibility of disease misclassification. A total of 131 cases and 91 coronary controls participated in the study (response rate of 63%). However, three of the coronary controls were excluded from analysis, because a second evaluation of the angiography results revealed too much coronary narrowing. Furthermore, genotyping was not performed for one coronary control subject, leaving 87 coronary controls for analyses. We obtained a population-based control group, from a register of about 10,000 men previously enrolled for participation in a trial of cholesterol-lowering medication. The trial was never conducted, however. Among men with no prior history of cardiovascular disease or diabetes, a random sample was invited for participation. Spouses of participants were invited to participate as well. A total of 101 population-based control subjects was studied (total response rate of 62%). However, genotyping was not performed for one subject, leaving 100 population- based controls for analyses. The study protocol was approved by the medical ethical committee. All participants gave their written informed consent.
Blood sampling, examination, and biochemical analyses At the day of the examinations, venous blood samples were obtained from all subjects between 8:30 and 9:30 AM, after a 10 to 12 hour fast. L-methionine (0.1 g/kg body weight) mixed with orange juice was given orally, together with a
33 Chapter 3 standardized low protein breakfast. After breakfast, subjects were interviewed about medication, smoking habits, alcohol consumption, and other coronary risk factors. Subjects received a standardized low protein lunch and were asked not to consume any protein-containing foods, like milk, cheese, or meat. Six hours after methionine administration, a second blood sample was drawn for estimation of plasma tHcy in response to methionine provocation. From one population-based control subject, we did not obtain a blood sample after methionine loading. Duplicate blood pressure readings were taken before and after the methionine loading test with the subject seated after 5 minutes rest. Height and weight (without shoes and heavy clothing) were measured in the morning. For measurement of whole blood folate, 200 ul of EDTA blood was mixed with 4 ml (1:20) freshly prepared 1% (w/v) ascorbic acid solution. The rest of the EDTA blood, to be used for measurement of tHcy and creatinine in plasma was placed on ice immediately and in the dark, and centrifuged at 4° C within one hour. Serum was obtained for measurement of total and high density lipoprotein (HDL) cholesterol, triglycerides, and creatinine. Plasma tHcy, which refers to the sum of protein-bound, free oxidized and reduced species of homocysteine in plasma, was determined by a modification of the method of Refsum et al.13 Folate in whole blood was determined by radio immunoassay (Diagnostic Products Corporation, USA, Dual-count Solid Phase, no boil assay) by MIMELAB-AB, Soraker, Sweden. We expressed folate concentration per haematocrit, referred to as erythrocyte folate. Folate values were missing in two cases and two coronary controls. Creatinine, total cholesterol and HDL cholesterol (after precipitation of LDL and VLDL) were determined in serum with enzymatic photometry.
Genotyping DNA was obtained from the buffy coat of EDTA blood.14 The mutation involves a C to T mutation at nucleotide 677, which converts an alanine to a valine residue. The alteration creates a HiriFl site, which was used for mutation analysis. The method has been described in detail elsewhere."
Statistical analyses Differences in cardiovascular risk factor levels between cases and the two control groups were tested with Student's t test for continuous variables and Pearson's chi-square test for frequency measures. To study whether the genotype frequency differed among the cases and the control groups, we used Pearson's chi-
34 Thermolabile MTHFR, Folate, and Homocysteine square test. Plasma tHcy showed positive skewness, therefore log-transformed tHcy was used in all analyses. Differences in geometric mean tHcy levels were tested with Student's t test. Tests for trend were performed with linear regression analysis. All reported P-values are two-tailed.
Results
Characteristics and tHcy Age, gender, coronary risk factors and plasma tHcy levels of the cases and the two control groups are shown in Table 3-1. Cases had the highest percentage of males, the highest mean age, the highest proportion of hypertensive subjects, and the lowest serum HDL cholesterol levels. Mean serum level of triglycerides, mean plasma level of creatinine and body mass index were highest in cases, whereas alcohol consumption was highest in controls. Mean pack years of smoking was higher in cases compared to both control groups. Cases had higher geometric mean fasting (9%, P=0.02) and post-load tHcy (7%, P=0.04) than the combined control groups, after adjusting for age and gender. The increase after loading (i.e. post-load minus fasting level) was 6% (P=0.15) higher in cases relative to the combined control groups.
MTHFR genotype In the total study population, the allele frequency of the mutation was 33%. The numbers (%) of individuals that were homozygous for the MTHFR mutation (+/+) were 13 (10.0%) among cases, 10 (11.5%) among coronary controls, and 7 (7.0%) among population-based controls (Table 3-2). The frequency of homozygosity was 9.1% for the combined control groups, i.e. virtually the same as for cases. The prevalences in cases were not statistically significantly different from those in either of the two control groups. The odds ratio (OR) of severe coronary atherosclerosis for (+/+) subjects relative to. heterozygous mutant (+/-) and homozygous normal (-/-) subjects was 1.1 (95% confidence interval [CI], 0.5-2.4). Gender ratio, age, serum creatinine and risk factors for coronary artery disease did not differ between the genotype subgroups, in any of the three study groups, nor in the total study population.
35 Chapter 3
Table 3-1. Characteristics and plasma total homocysteine of cases with severe coronary atherosclerosis and two groups of controls
Cases Coronary Population (n=131) controls controls (n=87) (n=100)
Age (years) 52.5 ± 7.5 48.1 + 8.0* 49.9 ± 6.9+ Gender (% male) 84.7 59.8* 74.0 Body mass index (kg/m2) 26.8 ± 3.0 26.0 + 3.6 26.0 ± 3.7 Total cholesterol (mmol/L) 6.8 ± 1.4 6.8 ± 1.7 6.7 ± 1.7 Total/HDL cholesterol 7.0 + 2.1 6.0 + 2.5* 6.0 ± 2.7+ Hypercholesterolemic (%) 68.7 60.9 49.0+ Triglycerides (mmol/L) 2.1 ± 1.3 1.7 + 1.0* 1.4 + 1.0+ Systolic blood pressure (mmHg) 135.0 + 13.3 131.4 ± 16.9 134.8 ± 14.4 Diastolic blood pressure (mmHg) 81.3 ± 7.4 78.4 ± 9.3* 81.7 ± 8.0 Hypertensive (%) 85.5 51.7* 17.0+ Currently smoking (%) 38.2 40.2 37.0 Pack years (years) 31.9 + 27.1 22.7 ± 21.1 21.8 ± 22.6 Alcohol consumption (glasses/day) 0.8 ± 1.3 1.1 ± 1.7 1.4 ± 1/7+
Fasting tHcy (umol/L) 13.7 ± 6.6 11.8 ± 3.6 12.5 ± 5.7 Post-load tHcy (umol/L) 41.1 ± 12.6 37.9 ±11.3* 39.0 ± 12.9 Post-load increase in tHcy (umol/L) 27.5 ± 9.8 26.0 ± 9.9 26.5 ± 9.2
tHcy, total homocysteine. * P < 0.05, cases versus coronary control subjects. + P < 0.05, cases versus population-based control subjects. + P < 0.05, cases versus coronary control subjects, adjusted for sex and gender.
Table 3-2. Frequencies of thermolabile MTHFR genotype among cases with severe coronary atherosclerosis and two groups of controls
Cases Coronary Population (n=131) controls controls (n=87) (n=100) Genotype
% (n) Homozygous normal (-/-) 45.0 (n=59) 39.1 (n=34) 45.0 (n=45) Heterozygous mutant (+/-) 45.0 (n=59) 49.4 (n=43) 48.0 (n=48) Homozygous mutant (+/+) 10.0 (n=13) 11.5 (n=10) 7.0 (n=7)
36 Thermolabile MTHFR, Folate, and Homocysteine
Levels of tHcy by MTHFR genotype Within groups of cases and controls, geometric mean levels of tHcy were highest among the (+/+) subjects and lowest among the (-/-) subjects, whereas (+/-) individuals had intermediate levels (Table 3-3). Ratios of geometric mean tHcy levels of (+/+) subjects to levels of (-/-) subjects were highest for fasting tHcy and lowest for the increase in tHcy after methionine loading, in all three study groups (data not shown). When combining all subjects, these ratios were 1.36 for fasting tHcy, 1.25 for post-load tHcy, and 1.18 for post-load increase in tHcy. In the combined study groups, tests for linear trend were highly statistically significant, for all three tHcy measurements.
Table 3-3. Associations between thermolabile MTHFR genotype and plasma total homocysteine (tHcy) among cases with severe coronary atherosclerosis and two groups of controls
Cases Coronary Population All (n=131) controls controls Subjects (n=87) (n=100) (n=318)
umol/L Fasting tHcy (-/-) 12.4 10.3 10.9 11.4 (+/-) 12.7 11.9+ 12.1+ 12.3+ (+/+) 16.6 13.1 17.4 15.5* P* = 0.001 Post-load tHcy (-/-) 36.4 34.5 34.7 35.4 (+/-) 41.3+ 36.8 38.9+ 39.2+ (+/+) 46.4 42.0* 43.6 44.2+ P = 0.001 Increase in tHcy (-/-) 23.6 23.7 23.6 23.6 (+/-) 27.9+ 24.4 26.5 26.4+ (+/+) 28.4 28.1 25.8 27.7+ P = 0.002
* Test for linear trend. + P < 0.05 for (+/-) versus (-/-). + P < 0.05 for (+/+) versus (-/-). Results are expressed as geometric means.
37 Chapter 3
Erythrocyte folate Erythrocyte folate did not differ between the genotypes, either in the separate study groups, or in the pooled study groups (data not shown). In the entire study population, we compared plasma tHcy levels of subgroups with low and high erythrocyte folate, per genotype (Figure 3-1). We chose the median erythrocyte level (790 nmol/L) of the entire study population as a cutoff-point for low folate status. Geometric mean plasma tHcy was higher for the subjects with low compared to those with high erythrocyte folate, especially in the (+/+) group. Furthermore, this relation was most evident for fasting tHcy, as shown in Figure 1. Table 3-4 shows the differences in tHcy between individuals with low and high folate status in the (+/+) group. Finally, we calculated ORs of coronary atherosclerosis for (+/+) subjects relative to (+/-) and (-/-) subjects, for strata of low and high erythrocyte folate level. The ORs were 2.2 (95% CI, 0.7-6.8) and 0.6 (95% CI, 0.2-1.7) among subjects with low and high folate levels, respectively.
60 r Post-load
50
40 Increase
30 l-'asiinji
i I •351
20 B Low folate
High folate 10 0' -/- +/-+/+ -/- +/-+/+ -/- +/-+/+
Genotype
Figure 3-1. Geometric mean total homocysteine levels (tHcy) for genotypes of thermolabile MTHFR mutation, stratified by erythrocyte folate (cutoff-point 790 nmol/L).
38 Thermolabile MTHFR, Folate, and Homocysteine
Table 3-4. Plasma total homocysteine among subjects homozygous for thermolabile MTHFR, stratified by median erythrocyte folate level (790 nmol/L) Low Folate High Folate Difference in (n=13) (n=17) geometric means
umol/L Percent (95% CI) Fasting tHcy Geometric mean 21.4 12.1 77% (27% to 144%) Mean ± SD 25.7 ± 18.2 12.6 + 3.3 Post-load tHcy Geometric mean 51.2 39.5 30% (-2% to 70%) Mean ± SD 56.3 + 26.2 41.6 ± 15.4 Increase in tHcy Geometric mean 28.9 26.8 8% (-18% to 42%) Mean ± SD 30.6 + 10.5 29.1 ± 14.0 tHcy, total homocysteine; CI, confidence interval.
Discussion
In the present investigation, we showed that homozygosity for the 677C—>T mutation in the MTHFR gene was associated with raised plasma tHcy levels. However, homozygosity for this mutation was not associated with increased risk of severe coronary atherosclerosis. The combination of low erythrocyte folate and (+/+) genotype was associated with a particularly high plasma tHcy, possibly infering an increased risk of coronary artery disease. The frequency of homozygosity for thermolabile MTHFR is high in the general population," and frequencies may be even higher in subjects with cardiovascular disease.8,15,16 Kang et al., in an enzymatic study, reported a frequency of 17% of thermolabile MTHFR, in patients with coronary artery disease, and of 5% among subjects without clinical evidence of atherosclerotic vascular disease.8 In a DNA study, Frosst et al. observed a frequency of 12% for the 677C-»T mutation in unselected French-Canadian chromosomes." In another study, also in French- Canadians, the frequency of the 677C->T mutation was 15% in coronary artery disease patients, but no information was provided on the frequency in controls.15 Kluijtmans et al. found prevalences of 15% and 5% for the same mutation in Dutch patients of vascular disease and controls, respectively16. Apparently, frequencies of the (+/+) genotype for the 677C-»T mutation may vary between populations. The
39 Chapter 3
(+/+) genotype may be slightly more common in the French-Canadian population," in which a frequency of 12% was observed, compared to 7% in the present study (presuming that our population-based control sample comes closest to an unselected group). Conceivably, different frequencies of the (+/+) genotype in different populations of cardiovascular disease patients may be related to the inclusion criteria, including age and gender, but also to ethnicity and the folate status of the source population. It is not unlikely that the (+/+) genotype will only emerge as a risk factor for cardiovascular disease in a source population with a low folate status. Our observation of highest tHcy levels in homozygous mutant subjects and lowest tHcy levels in homozygous normal subjects, is supported by other studies.8,11'15'16 We found that this trend was most apparent for fasting tHcy levels. Also, we found the interaction between erythrocyte folate and the mutation to be most marked for fasting tHcy. These findings support the idea that the fasting tHcy level is to a large extent determined by homocysteine remethylation, whereas increased post-load tHcy may reflect abnormalities in the transsulfuration pathway.17 We observed higher tHcy levels among subjects with low compared to high erythrocyte folate, which is in agreement with results published by others.18 Furthermore, we found a dose-response relation between the 677C—>T mutation and plasma tHcy, but no difference in erythrocyte folate levels among the various genotypes. These data suggest that cellular folate status and MTHFR activity are independent, but interactive determinants of plasma tHcy. This interrelation may be expected as both defects cause metabolic events converging at a point of insufficient 5-methyl-THF for optimal homocysteine remethylation. Several reports have demonstrated a strong inverse association of dietary folate,19 serum folate,19'20 and erythrocyte folate18 with fasting plasma tHcy, and a reduction of tHcy after intake of folic acid.21"23 We have no data on folate intake. However, erythrocyte folate is a measure of tissue folate stores, likely to reflect the balance between folate intake, absorption, metabolism, and utilization.24 Provided that the determinants of folate homeostasis other than folate intake are similar within groups of genotype, our findings suggest that adequate intake is essential for maintaining low tHcy, particularly in (+/+) subjects. In conclusion, we have demonstrated a positive interaction between the 677C—>T mutation in MTHFR and low erythrocyte folate, which both result in elevated plasma tHcy. Subjects who are homozygous for this common mutation are particularly susceptible to elevation of plasma tHcy, under conditions of inadequate folate status, which is often related to low folate intake. There is ample evidence that elevated tHcy is a risk factor for cardiovascular disease.25 Thus, the results presented here suggest that increased dietary folate intake is a means to prevent
40 Thermolabile MTHFR, Folate, and Homocysteine cardiovascular disease in subjects with the MTHFR mutation. Future studies are warranted to obtain the mutation frequency, but also plasma and erythrocyte folate, together with estimates on folate intake in large populations of healthy subjects and in patients with cardiovascular disease.
Acknowledgments
The study was supported by a grant from the Netherlands Organization for Scientific Research. The authors would like to express their gratitude to Drs A.A.A. Bak, S.C. Balduw, G.J. van Beek, M.P. Freericks, F.M.A. Harms, R. van Mechelen, W.M. Muijs van de Moer, and R. Wardeh for their support in selecting the participants. We are very grateful to Annelies Legters, the research assistant. Furthermore, we would like to thank Sandra G. Heil for expert technical assistance. Mariette Penning, Halvard Bergesen, Elfrid Blomdal, Wenche Breyholtz, and coworkers from the Sticares Foundation, are thanked for their assistance in the research.
References
1. Malinow MR. Homocysteine and arterial occlusive diseases (Frontiers in Medicine). J Intern Med 1994;236:603-17.
2. Malinow MR, Stampfer MJ. Role of plasma homocysteine in arterial occlusive disease. Clin Chem 1994;40:857-8.
3. Genest JJ Jr, McNamara JR, Upson B, et al. Prevalence of familial hyperhomocyst(e)inemia in men with premature coronary artery disease. Arterioscler Thromb 1991;11:1129-36.
4. Wu LL, Wu J, Hunt SC, James BC, et al. Plasma homocysteine as a risk factor for early familial coronary artery disease. Clin Chem 1994;40:552-61.
5. Ueland PM, Refsum H, Brattström L. Plasma homocysteine and cardiovascular disease. In: Francis RB Jr, ed. Atherosclerotic Cardiovascular Disease, Hemostasis, and Endothelial Function. New York, NY: Marcel Dekker Inc; 1992:183-236.
6. Kang SS, Wong PWK, Zhou J, et al. Thermolabile methylenetetrahydrofolate reductase in patients with coronary artery disease. Metabolism 1988;37:611-13.
41 Chapter 3
7. Kang SS, Zhou J, Wong PWK, Kowalisyn J, Strokosch G. Intermediate homocysteinemia: a thermolabile variant of methylenetetrahydrofolate reductase. Am J Hum Genet 1988;43:414-21.
8. Kang SS, Wong PWK, Susmano A, Sora J, Norusis M, Ruggie N. Thermolabile methylenetetrahydrofolate reductase: an inherited risk factor for coronary artery disease. Am J Hum Genet 1991;48:536-45.
9. Kang SS, Passen EL, Ruggie N, Wong PW, Sora H. Thermolabile defect of methylenetetrahydrofolate reductase in coronary artery disease. Circulation 1993;88:1463-9.
10. Engbersen AMT, Franken DG, Boers GHJ, Stevens EMB, Trijbels FJM, Blom HJ. Thermolabile 5,10-methylenetetrahydrofolate reductase as a cause of mild hyperhomocysteinemia. Am J Hum Genet 1995;56:142-50.
11. Fresst P, Blom HJ, Milos R, et al. A candidate genetic risk factor for vascular disease: a common mutation in methylenetetrahydrofolate reductase. Nat Genet 1995;10:111-3.
12. Verhoef P, Kok FJ, Kruyssen HACM, et al. Plasma total homocysteine, B-vitamins and risk of coronary atherosclerosis (submitted).
13. Fiskerstrand T% Refsum H, Kvalheim G, Ueland PM. Homocysteine and other thiols in plasma and urine; automated determination and sample stability. Clin Chem 1993;39:263-71.
14. Miller SA, Dykes DD, Polesky HF. A simple salting out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res 1988;16:1215.
15. Frosst P, Christensen B, Goyette P, Rosenblatt DS, Genest J, Rozen R. Metylenetetrahydrofolate reductase (MTHFR) in coronary artery disease patients. Irish J Med Sei 1995;164(S15):17 (abstract).
16. Kluijtmans LAJ, Van den Heuvel LPWJ, Boers GHJ, et al. Molecular genetic analysis in mild hyperhomocysteinemia: a common mutation in the methylenetetrahydrofolate reductase gene is a genetic risk factor for cardiovascular disease. Am J Hum Genet 1996 (in press).
17. Brattström L, Israelsson B, Norrving B, et al. Impaired homocysteine metabolism in early- onset cerebral and peripheral occlusive arterial disease. Atherosclerosis 1990;81:51-60.
18. Chadefaux B, Cooper BA, Gilfix BM, et al. Homocysteine: relationship to cobalamin, serum folate, erythrocyte folate, and lobation of neutrophils. Clin Invest Med 1994;17:540-50.
19. Seihub J, Jacques PF, Wilson PWF, Rush D, Rosenberg IH. Vitamin status and intake as primary determinants of homocysteinemia in an elderly population. JAMA 1993;270:2693-8.
42 Thermolabile MTHFR, Folate, and Homocysteine
20. Jacobsen DW, Gatautis VJ, Green R, et al. Rapid HPLC determination of total homocysteine and other thiols in serum and plasma: sex differences and correlation with cobalamin and folate concentrations in healthy subjects. Clin Chem 1994;40:873-81.
21. Brattstrom LE, Israelsson B, Jeppsson JO, Hultberg BL. Folic acid-an innocuous means to reduce plasma homocysteine. Scand J Clin Lab Invest 1988;48:215-21.
22. Ubbink JB, Van der Merwe A, Vermaak WJH, Delport R. Hyperhomocysteinemia and the response to vitamin supplementation. Clinical Investig 1993;71:993-8.
23. Franken DG, Boers GHJ, Blom HJ, Trijbels JMF, Kloppenborg PWC. Treatment of mild hyperhomocysteinemia in vascular disease patients. Arterioscler Thromb 1994;14:465-70.
24. Davidson S, Passmore R, Eastwood MA. Human Nutrition and Dietetics. -8th ed. Hong Kong: Longman Group Limited, 1986.
25. Boushey CJ, Beresford SAA, Omenn GS, Motulsky AG. A quantitative assessment of plasma homocysteine as a risk factor for vascular disease. Probable benefits of increasing folic acid intakes. JAMA 1995;274:1049-57.
43
Homocysteine Metabolism and 4 Risk of Myocardial Infarction: Relationship
with Vitamins B6, B12, and Folate
Abstract
Background & methods Elevated plasma tHcy levels are an independent risk factor for vascular disease. In a case-control study, the authors studied the associations of fasting plasma tHcy and vitamins, that are important cofactors in homocysteine metabolism, with 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. Results After adjusting for sex and age, geometric mean plasma tHcy was 11 % higher in cases, as compared to controls (P=0.006). There was no clear excess of cases with extremely elevated levels. The age- and sex-adjusted odds ratio [OR] for each 3 umol/L (approximately 1 standard deviation) increase in plasma tHcy was 1.35 (95 % confidence interval [CI], 1.05-1.75; P trend=0.007). After further control for several other risk factors, the OR was not affected, but the CI was wider and the P-value for trend was less significant. Dietary and plasma levels of vitamin B6 and folate were lower in cases than in controls and these vitamins were inversely associated with risk of myocardial infarction, independently of other potential risk factors. Vitamin B12 showed no clear association with myocardial infarction, although methylmalonic acid levels were significantiy higher in cases. Comparing mean levels of several homocysteine metabolites among cases and controls, it appeared that impairment of remethylation of homocysteine, dependent of folate and vitamin B12, rather than vitamin B6-dependent transsulfuration, was the predominant cause of high tHcy levels in cases. Accordingly, plasma folate (and to a lesser extent plasma vitamin B12), but not plasma vitamin B6, correlated inversely with plasma tHcy, even for concentrations at the high end of normal vitamin values. Conclusions These data provide further evidence that plasma tHcy is an independent risk factor for myocardial infarction. In this population, folate was the most important determinant of plasma tHcy, even in subjects with apparently adequate nutritional status of this vitamin.
Petra Verhoef, Meir J. Stampfer, Julie E. Buring, J. Michael Gaziano, Robert H. Allen, Sally P. Stabler, Robert D. Reynolds, Frans J. Kok, Charles H. Hennekens, Walter C. Willett (in press, Am J Epidemiol 1996).
45 Chapter 4
Introduction
Extremely elevated plasma levels of total homocysteine (referred to as tHcy, 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 tHcy can be moderately elevated in individuals heterozygous for these defects, or with inadequate intake of vitamins B6, B12, and folate, which serve as cofactors in the enzymatic pathways of homocysteine metabolism.1,2 Homocysteine is a fhiol-containing amino acid derived from methionine, that can be metabolized through two enzymatic pathways: transsulfuration and remethylation (Figure 4-1). In the transsulfuration pathway, homocysteine is condensed with serine to form cystathionine, an irreversible reaction, dependent on pyridoxal 5'-phosphate (PLP), the active form of vitamin B6. Subsequently, cystathionine is converted to cysteine, in another vitamin B6-dependent reaction. In remethylation, homocysteine receives a methyl group from 5-mefhyltetrahydrofolate
(a vitamin BI2-dependent reaction) or from betaine to form methionine. In the reaction involving betaine, which is independent of folate and vitamin B,2, dimethylglycine is formed. The methionine formed is activated by ATP to form S- adenosylmethionine, which serves as a methyl donor to a variety of acceptors, one being glycine, leading to the formation of methylglycine. S-adenosylhomocysteine is formed in a transmethylation reaction, which is then hydrolyzed in homocysteine and adenosine.3,4 Even moderately elevated plasma tHcy levels appear to be an independent risk factor for coronary artery disease.5,6,7 One prospective study reported a 3.4-fold excess risk of myocardial infarction for men with levels above 15.8 umol/L, the 95th percentile for controls in that population, as compared to 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 mechanisms include increased uptake of LDL 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 tHcy levels with blood levels of vitamins B6, B12, and folate, and with intake levels of vitamin 10,11 B6 and folate. Also, in human experimental studies, supplements of these vitamins (especially folate) reduce elevated tHcy levels.1213 Some, but not all epidemiologic studies have found inverse associations of the vitamins with vascular disease.5,14
46 Homocysteine, B-vitamins and Myocardial Infarction
In the present case-control study, we investigated the associations of plasma tHcy, dietary and plasma levels of vitamins B6, B,2, and folate, and dietary methionine with risk of first myocardial infarction. Also, we studied associations between plasma tHcy and the vitamins. Furthermore, we compared plasma levels of several of the homocysteine derivatives shown in Figure 4-1 among 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
Polyamines
CH3-.
Cystathionine
PLP 6
a-Ketobutyrate Cysteine
Taurine—»~ —»~ S04
Figure 4-1. Homocysteine metabolism in man and animals. Enzymes: 1=5- methyltetrahydrofolate:homocy steine methyltransferase; 2 = 5,10- methylenetetrahydrofolate reductase; 3 = betaine:homocysteine methyltransferase; 4 - choline dehydrogenase; 5 = cystathionine ß-synthase; 6 = y-cystathionase. THE = tetrahydrofolate; PLP = pyridoxal 5'-phosphate, Selhub and Miller (reference 3).
47 Chapter 4
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 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 eight 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 were obtained. The information included history of diabetes, high blood pressure, and high cholesterol, parental history of ischemic heart disease < 60 years, weight, 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 (weight in kilograms)/(height in metres squared). Blood was drawn into 0.1% EDTA vacutainers, and fresh plasma was used to determine total, high density lipoprotein, and low density lipoprotein cholesterol, as previously described.16 The rest of the plasma was stored at -70 0 C, with similar
48 Homocysteine, B-vitamins and Myocardial Infarction storage time for cases and controls. In 1986, plasma PLP 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 in this publication).18 In 1993, we assayed plasma samples for tHcy, several of its derivatives (cystathionine, cysteine, methionine, dimethylglycine, methylglycine, serine, and glycine), vitamin BI2, and folate, as previously described.15 A subsample of 130 cases (97 men, 33 women) and 118 control subjects (81 men and 37 women) had a sufficient plasma volume left for these assays. In addition to the homocysteine-related metabolites, we measured methylmalonic acid and methylcitric acids I and II, which are usually elevated in
15 subjects with vitamin B12 deficiency. We had information on dietary intakes of methionine, the main metabolic precursor of homocysteine, vitamins B6, B12, 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 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 B6, vitamin B[2, 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 U.S. 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 t test for continuous variables and Pearson's chi-square test for proportions. We also studied associations of tHcy 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 tHcy and of age on sex on the respective risk factor or vitamin level. Mean plasma levels of tHcy and metabolites, as well as plasma and dietary
49 Chapter 4
levels of vitamin B6, vitamin B12, and folate, were compared for cases and controls. 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 case-control 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 tHcy with risk of myocardial infarction. We calculated odds ratios (OR) plus 95% confidence intervals (CI), for both plasma tHcy as a continuous variable (per 1 standard deviation [SD] 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 tHcy. Tests for trend were performed by adding log-transformed plasma tHcy to the logistic regression models continuously. In the same way as for plasma tHcy, we assessed relationships of dietary intakes of methionine, and dietary and plasma levels of vitamins B6, B12, and folate with myocardial infarction. Furthermore, we repeated the risk analyses for plasma tHcy as a continuous variable (per 1 SD increase) 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 - associations with tHcy Table 4-1 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/HDL 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 to cases, whereas in the subset controls had relatively high prevalences of history of high blood pressure and family history of myocardial
50 Homocysteine, B-vitamins and Myocardial Infarction
infarction. Associations of plasma tHcy with risk factors of coronary disease and other possible confounding factors were evaluated, controlling for sex and age (Table 4-1). Correlations were small and differed substantially among cases and controls. Levels of tHcy were lower in women and directly associated to age, in cases. In control subjects, we found direct associations of tHcy levels with alcohol consumption and with history of hypercholesterolemia. In cases, but not in controls, history of high blood pressure correlated directly to tHcy, whereas energy intake and family history of MI correlated inversely with levels of tHcy.
Table 4-1. Characteristics of cases of myocardial infarction and controls, and correlations of characteristics with plasma total homocysteine
Cases (n==130 ) Controls (n =118)
Mean Correlation Mean Correlation or SD with tHcy: or SD with tHcy: % * P % r P r Age (years) 57.7 9.3 0.19 0.03 57.7 9.1 0.08 0.38 Sex (% male) 74.6 0.16 0.07 68.6 0.13 0.15 Body mass index (kg/m2) 26.0 4.3 0.02 0.83 26.2 4.6 0.04 0.63 Total/HDL cholesterol+ 6.41 1.82 0.17 0.06 5.18 1.83 0.11 0.25 Hist, of high cholest. (%)+ 16.9 -0.04 0.62 9.4 0.22 0.02 Hist, of hypertension (%) 42.6 0.19 0.03 43.2 0.06 0.51 Energy intake (kcal/day) 2404 843 -0.19 0.04 2292 712 0.01 0.95 Saturated fat intake (g/day)+ 38.0 17.9 -0.15 0.09 33.2 13.6 -0.02 0.81 Currently smoking§ (%)+ 45.4 0.13 0.15 25.6 0.10 0.27 Alcohol intake (g/day)+ 14.2 24.2 0.02 0.82 20.0 28.4 0.33 <0.01 Physical activity (kcal/day) 3005 3111 -0.05 0.61 3369 2947 -0.10 0.28 History of diabetes (%) 15.4 -0.09 0.31 10.2 -0.07 0.44 Family history of MI (%) 16.9 -0.19 0.03 19.5 -0.06 0.53 tHcy, total homocysteine; HDL, high density lipoprotein. * Pearson's correlation coefficient, adjusted for age and sex. + P < 0.05, difference in characteristic between cases and controls. * P < 0.10, difference in characteristic between cases and controls. 8 For calculation of the Pearson correlation coefficient smoking was coded as: l=never, 2=ex-smoker, 3=less than 1 pack/day, 4=1-2 pack/day, 5=2 or more packs/day.
51 Chapter 4
Mean levels of tHcy, metabolites, vitamins and methionine intake Mean and geometric mean plasma levels of tHcy and several metabolites are shown in Table 4-2. The geometric mean of plasma tHcy was 11% higher in cases than in control subjects (P=0.006). Values of tHcy in cases were shifted towards the right across the entire frequency distribution (Figure 4-2). There was no clear excess of cases limited to those with extremely elevated levels. For cysteine and dimethylglycine we observed significantly higher geometric mean plasma levels in cases than controls, whereas methionine and methylglycine levels were lower in cases (all P-values below 0.05). Although not statistically significant, geometric mean cystathionine was higher among cases, as compared to 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 B12-inadequacy among cases. However, the concentration of methylcitric acids I and II did not differ significantly between cases and controls (Table 4-2).
30 -
25
N 20 u m b e r
Ox -nf-i? 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 tHcy umol/L
Cases Controls
Figure 4-2. Smoothed frequency distribution of total homocysteine (tHcy) levels (pmol/L), rounded to the nearest integer, in plasma from 130 cases of first myocardial infarction and 118 control subjects.
52 Homocysteine, B-vitamins and Myocardial Infarction
Table 4-2. Concentrations of total homocysteine and compounds reflecting homocysteine metabolism and vitamin Bl2 status in cases of myocardial infarction and controls
Cases (n= 130) Controls (n==118 )
Mean SD GM Mean SD GM P'
Total homocysteine (umol/L) 10.6 3.2 10.2 9.6 3.3 9.1 0.006 Cystathionine (nmol/L) 199 195 170 167 72 154 0.11 Cysteine (|imol/L) 304 43 301 291 43 288 0.01 Methionine (umol/L) 21.3 4.5 20.9 22.6 4.5 22.2 0.007 Dimethylglycine (umol/L) 3.1 1.4 2.9 2.8 0.9 2.7 0.04 Methylglycine (umol/L) 1.8 0.6 1.7 1.9 0.6 1.9 0.02 Serine (nmol/L) 90.8 21.8 i 88.4 91.5 20.6 89.1 0.91 Glycine (nmol/L) 213 54 207 226 67 218 0.20 Methylmalonic acid (nmol/L) 200 119 181 174 70 163 0.02 MC acids I & II (nmol/L) 138 72 127 139 47 132 0.25
GM, geometric mean; MC, methylcytric. * Student's t test, based on difference in geometric means between cases and controls, adjusting for age and sex differences.
Table 4-3 shows mean dietary intake of methionine, the metabolic precursor of homocysteine, and mean plasma and intake levels of vitamins J36, B12, 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 B6 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 Bl2, although methionine intake was slightly higher among cases, and plasma vitamin B12 slightly lower among cases.
Associations of tHcy with methionine intake and vitamins We calculated correlations between methionine intake and plasma tHcy levels. Adjusting for age and sex only, 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 B6, B12 and folate, these correlations were -0.09 (P=0.29) and -0.22 (P=0.01).
53 Chapter 4
Table 4-3. Dietary intake of methionine, and plasma and dietary intake levels of vitamin
11$, vitamin B,2, and folate in cases of myocardial infarction and controls
Cases (n=130) Controls (n=l 18)
Mean SD GM Mean SD GM P'
Plasma: PLP+ (nmol/L) 49.3 28.7 42.4 62.2 40.5 51.8 0.005
+ Vitamin Bn (pmol/L) 466 199 428 475 178 443 0.60 Folate (nmol/L) 8.76 3.30 8.23 9.93 4.93 9.11 0.03 Diet: Methionine (g/day) 2.58 0.64 2.50 2.49 0.63 2.41 0.25
Vitamin B6 (mg/day) Food 2.71 0.67 2.63 2.77 0.69 2.69 0.57 Food & supplements 3.29 1.37 3.05 5.79 22.99 3.49 0.04
Vitamin B,2 (ug/day) Food 11.52 7.56 10.02 9.85 4.02 9.09 0.12 Food & supplements 13.07 8.41 11.26 14.20 23.69 11.13 0.84 Folate (ug/day) Food 340.9 107.0 323.7 372.0 105.3 357.7 0.01 Food & supplements 423.0 212.2 377.4 493.6 208.0 453.7 0.002
GM, geometric mean. * Student's t test, based on difference in geometric means between cases and controls, adjusting for age and sex differences. + PLP, pyridoxal 5'-phosphate; five cases and one control had missing values for plasma PLP.
We excluded one case with a plasma vitamin B12 value of 2712 pmol/L, which was far outside the range of other subjects.
Table 4-4 shows the correlation coefficients between plasma levels of tHcy and plasma and intake levels of vitamins, adjusted for age and sex (first and third columns). tHcy correlated inversely with plasma levels of all three vitamins, in cases as well in controls. Correlations were particularly strong for the association of plasma tHcy with plasma folate, with ^=-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 with normal to high levels. In cases, but not in controls, an inverse association of plasma tHcy with plasma vitamin B[2 was observed (r=-0.35, P=0.0001). In cases, intakes of all three vitamins showed significant inverse relationships with plasma tHcy, whereas in control subjects these associations were less strong.
54 Homocysteine, B-vitamins and Myocardial Infarction
Table 4-4. Correlation coefficients of total homocysteine with vitamin vitamin B,2, and folate in cases of myocardial infarction and controls
Cases (n=130 ) Controls (n=118)
r" (P-value) r+ (P-value) r (P-value) r (P-value) Plasma:
PLP* -0.22 (0.01) -0.08 (0.37) -0.23 (0.01) -0.01 (0.94)
§ Vitamin BI2 -0.35 (0.0001) -0.25 (0.006) -0.19 (0.04) -0.02 (0.82)
Folate -0.38 (0.0001) -0.23 (0.01) -0.49 (0.0001) -0.45 (0.0001) Diet:
Vitamin B6 -0.29 (0.0009) -0.04 (0.63) -0.12 (0.21) -0.02 (0.83)
Vitamin B12 -0.37 (0.0001) -0.24 (0.007) -0.08 (0.39) 0.03 (0.75)
Folate -0.32 (0.0002) -0.06 (0.50) -0.19 (0.03) -0.16 (0.08)
* Using log-transformed variables, adjusting for age and sex. + Using log-transformed variables, adjusting for age, sex, and levels of the other two vitamins. + PLP, pyridoxal 5'-phosphate; five cases and one control had missing values for plasma PLP. § We excluded one case with a plasma vitamin B,2 value of 2712 pmol/L, which was far outside the range of other subjects.
Because of 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 PLP and plasma folate, 0.30 (P=0.0008) for plasma levels of
PLP and vitamin BI2, and 0.33 (P=0.0003) for plasma folate and plasma vitamin
B12. Correlation coefficients were 0.57 (P=0.0001) for intakes of vitamin B6 and
folate, 0.64 (P=0.0001) for vitamins B6 and B12, and 0.51 (P=0.0001) for folate and
vitamin B,2. Thus, we also calculated correlations between levels of tHcy with plasma and dietary levels of each of the vitamins, controlling for levels of the other two vitamins simultaneously (Table 4-4, second and fourth columns). In control subjects, only for folate (plasma and dietary levels) did the association with tHcy
remain, whereas for vitamin B6 and B12 associations disappeared. For cases, both
plasma folate and vitamin B12 (plasma and dietary levels) remained inversely related to plasma tHcy, when controlling for levels of each of the other two vitamins simultaneously.
55 Chapter 4
Associations of tHcy with myocardial infarction We calculated risks by quintiles of plasma tHcy, based on the distribution among controls. For individuals with plasma tHcy levels higher than the 80th percentile of controls (11.2 umol/L), as compared to those with levels below the 20th percentile (7.2 umol/L), we observed an age- and sex-adjusted OR of 4.66 (95 % CI, 1.90 - 11.40). The corresponding multivariate-adjusted OR was 5.09 (95 % CI, 1.84 - 14.10). For the three quintiles in the middle, risks were elevated as well. There was a strong direct linear trend of plasma tHcy with risk of myocardial infarction (P=0.01, Table 4-5). In a logistic regression analysis with untransformed plasma tHcy, age and sex as independent variables, we estimated a coefficient of 0.101 ± 0.043 (8 ± SE), corresponding to an OR of 1.35 (95 % CI, 1.05 - 1.75) for each 3 umol/L (about 1 SD) increase in plasma tHcy. After additional adjustment for standard risk factors, including history of high blood pressure, and diabetes, plasma total/HDL 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 OR of 1.35 (95 % CI, 1.00 - 1.82). This can be interpretated as an average 35 percent increase in risk for each 1 SD increase in plasma tHcy. To evaluate possible effect modification by 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 umol/L in plasma tHcy among subjects stratified for these factors. We did not observe substantial differences among the strata of these risk factors (data not shown).
Associations of methionine intake and vitamins with myocardial infarction For plasma PLP, the highest two quintiles showed a trend towards an inverse relation with risk of myocardial infarction, even after adjustment for strong confounders like plasma total/HDL cholesterol and smoking habits (Table 4-5) (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 B12 was not associated with risk of myocardial infarction. For plasma PLP we did the same analyses in the original study group, of which 275 cases and 281 controls had PLP levels measured. Both the age- and sex-adjusted effect and the multivariate-adjusted effect were stronger in the smaller study group compared to the larger group. Also, there seemed to be more confounding in the larger group, especially from smoking and plasma total/HDL cholesterol (data not shown).
56 Table 4-5. Odds ratios of myocardial infarction by plasma levels of total homocysteine, pyridoxal 5'-phosphate, vitamin B12 and folate
Quintile* of plasma level
~ — B*-±SB (P)
Plasma tHcy (umol/L) <7.2 <8.2 <9.7 ¿11.1 > 11.1 Age- and sex-adjusted 1 1.81 3.23 2.10 4.66 1.258 ± 0.467 (0.007) Multivariate-adjusted4 1 1.80 4.61 2.50 5.09 1.352 ± 0.548 (0.01) 95% confidence interval (0.62 - 5.21) (1.61 - 13.20) (0.85 - 7.32) (1.84 - 14.10)
Plasma PLP (nmol/L) <29.7 <47.0 <63.1 <88.9 > 88.9 Age- and sex-adjusted 1 0.97 0.94 0.37 0.32 -0.644 ± 0.233 (0.006) Multivariate-adjusted+ 1 0.97 1.68 0.37 0.51 -0.407 ± 0.259 (0.12) 95% confidence interval (0.40- 2.33) (0.70-4.04) (0.14- 1.01) (0.19 - 1.36)
Plasma vitamin BI2 (pmol/L) < 321 < 399 < 503 < 610 > 610 Age- and sex-adjusted 1 0.60 1.13 0.55 0.94 -0.175 ± 0.328 (0.59) Multivariate-adjusted4, 1 0.71 1.07 0.57 0.97 -0.151 ± 0.364 (0.68) 95% confidence interval (0.29 - 1.77) (0.46 - 2.48) (0.22 - 1.49) (0.41 - 2.28)
Plasma folate (nmol/L) < 6.56 < 8.31 < 9.95 < 12.0 > 12.0 Age- and sex-adjusted 1 1.29 0.41 0.44 0.53 -0.804 ± 0.365 (0.03) Multivariate-adjusted4 1 1.19 0.47 0.32 0.57 -0.707 ± 0.417 (0.09) 95% confidence interval (0.53 - 2.69) (0.18 - 1.22) (0.12 - 0.84) (0.22 - 1.42) tHcy, total homocysteine; PLP, pyridoxal 5'-phosphate. * Quintiles are based on distribution of control subjects; real scale boundaries of quintiles are given. + Based on logistic regression analysis with log-transformed variables. + Adjusted for age, sex, energy intake, history of diabetes, history of high blood pressure, family history of ischemic heart disease, total/HDL cholesterol, body mass index, cigarette smoking, and alcohol consumption. 00
Table 4-6. Odds ratios of myocardial infarction by level of energy adjusted daily intake of methionine, vitamins B^ Bl2 and folate
Quintile* of nutrient intake
13+ ± SE (P)
Methionine intake (g/d) < 1.94 < 2.24 < 2.62 < 2.96 > 2.96 Age- and sex-adjusted 1 0.77 1.18 1.37 1.43 0.584 ± 0.510 (0.25) Multivariate-adjusted4* 1 0.58 1.04 1.26 1.28 0.552 ± 0.578 (0.34) 95% confidence interval (0.22- 1.53) (0.41 - 2.62) (0.52 - 3.09) (0.52 - 3.11)
Vitamin B6 intake (mg/d)
Food only: ¿2.12 < 2.50 ¿2.97 < 3.26 > 3.26 Age- and sex-adjusted 1 0.55 1.00 0.88 0.72 -0.301 ± 0.523 (0.57) Multivariate-adjusted4, 1 0.53 1.00 0.66 0.63 -0.438 ± 0.597 (0.46) 95% confidence interval (0.21 - 1.33) (0.43 - 2.36) (0.27 - 1.63) (0.25 - 1.62)
Food and supplements: < 2.31 ¿3.07 ¿ 3.79 ¿ 4.38 > 4.38 Age- and sex-adjusted 1 1.44 0.67 0.17 1.00 -0.651 ± 0.324 (0.04) Multivariate-adjusted4, 1 2.05 0.48 0.17 1.05 -0.729 ± 0.377 (0.05) 95% confidence interval (0.86 - 4.90) (0.19 - 1.25) (0.05 - 0.59) (0.43 - 2.57) Table 4-6. Continued
Vitamin B]2 intake (ug/day)
Food only: < 6.57 <8.32 <9.81 < 12.28 > 12.28 Age- and sex-adjusted 1 0.96 1.09 1.45 1.67 0.442 ± 0.283 (0.12) Multivariate-adjusted4, 1 0.84 1.17 1.30 1.77 0.522 ± 0.321 (0.10) 95% confidence interval (0.31 - 2.24) (0.46 - 3.01) (0.51 - 3.29) (0.73 - 4.34)
Food and supplements: < 7.06 <9.70 < 12.61 < 16.09 > 16.09 Age- and sex-adjusted 1 1.51 1.36 0.72 1.42 0.048 ± 0.233 (0.84) Multivariate-adjusted4' 1 1.60 1.13 0.73 1.46 0.052 ± 0.260 (0.84) 95% confidence interval (0.65 - 3.91) (0.46 - 2.78) (0.28 - 1.92) (0.61 - 3.51)
Folate intake (pg/d)
Food only: <282 < 319 <391 <467 > 467 Age- and sex-adjusted 1 0.37 0.67 0.66 0.30 -1.115 ± 0.454 (0.01) Multivariate-adjusted4, 1 0.28 0.70 0.57 0.30 -1.086 ± 0.514 (0.03) 95% confidence interval (0.11 - 0.74) (0.30 - 1.62) (0.24 - 1.35) (0.11 - 0.81)
Food and supplements: <310 < 391 <490 <682 > 682 Age-and sex-adjusted 1 0.61 0.45 0.31 0.43 -0.924 ± 0.302 (0.002) Multivariate-adjusted4, 1 0.63 0.44 0.24 0.38 -1.0i2 ± 0.347 (0.004) 95% confidence interval (0.27 - 1.49) (0.17 - 1.09) (0.09 -0.62) (0.15 - 0.95)
* Quintiles are based on distribution of control subjects; real scale boundaries of quintiles are given. + Based on logistic regression analysis with log-transformed variables. + Adjusted for age, sex, energy intake, history of diabetes, history of high blood pressure, family history of ischemic heart disease, total/HDL cholesterol, body mass index, cigarette smoking, and alcohol consumption. Chapter 4
Table 4-6 shows ORs of myocardial infarction by quintiles of energy- adjusted daily intakes of methionine, vitamins B6, B12 and folate (with and without intake from supplements). Intake of methionine showed no substantial association with risk of myocardial infarction. For folate intake, and to a lesser extent for vitamin B6, 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 B6 intake the estimates were rather inconsistent. Intake of vitamin B]2 was not materially associated with risk of myocardial infarction: ORs for all quintiles were close 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 CIs wide. We repeated the analyses shown for the smaller group in Table 4-6, 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 B6 was higher in the smaller control group as compared to 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. The association of total dietary intake of folate with myocardial infarction risk became somewhat stronger (8 ± SE: -1.378 ± 0.483, P=0.004) when intake of vitamin B6 and vitamin B12 were added to the logistic model as continuous variables.
Adjustment for folate and BI2 weakened the association of total intake of vitamin B6 with myocardial infarction (8 ± SE: -0.3872 ± 0.4266, P=0.36). This illustrates that increased intakes of both vitamins are associated with decreased risk of myocardial infarction, but that part of the observed effect of vitamin B6 is due to intercorrelation with folate.
Finally, when plasma folate, plasma PLP and plasma B12 were entered into the logistic model together with plasma tHcy (all as log-transformed continuous variables), the coefficient for tHcy became somewhat smaller (8 ± SE: 0.998 ± 0.620, P=0.11) and for folate much smaller (8 ± SE: -0.242 ±0.511, P=0.22). The coefficient of PLP was almost not affected (8 ± SE: -0.311 ± 0.282, P=0.27). This seems to indicate that folate and tHcy are part of the same causal pathway (low folate levels leading to elevated plasma tHcy, which subsequently may lead to increased risk of myocardial infarction), whereas in this population vitamin B6 possibly exhibits an association with myocardial infarction through other mechanisms than influencing tHcy levels.
60 Homocysteine, B-vitamins and Myocardial Infarction
Discussion
In this study, we found plasma tHcy to be an independent risk factor for first myocardial infarction and the effect seemed to be graded. Elevation of tHcy in cases was associated with increased levels of cystathionine and cysteine, indicating increased catabolism of homocysteine through the transsulfuration pathway, which is dependent of vitamin B6. 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 BI2 deficiency among cases, as compared to controls. Higher levels of dimethylglycine suggest that there was increased demethylation of betaine to dimethylglycine, a reactionstep which is independent of folate and vitamin B,2, at which homocysteine is remethylated to methionine. Folate exhibited a strong inverse association with plasma tHcy, in cases as well as controls, whereas vitamin B6 showed no association with plasma tHcy, controlling for the other two vitamins. Therefore,
although dietary and intake levels of vitamin B6 were lower in cases than controls, this did not appear to be the primary reason for elevated fasting tHcy levels. This is in line with the concept that fasting tHcy level is usually determined by homocysteine remethylation, and not by abnormalities in the transsulfuration pathway.25
Total dietary intakes and plasma levels of vitamin B6 and folate showed
inverse relationships with risk of myocardial infarction, whereas for vitamin B12 no clear associations with risk of myocardial infarction were observed. Vitamin levels
were intercorrelated, however, and the effect of dietary vitamin B6 was attenuated
after correction for dietary folate and vitamin BI2. The association between plasma PLP and myocardial infarction risk was independent of plasma folate and tHcy,
suggesting that protective mechanisms of vitamin B6, 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
26 vitamin B6 against coronary disease. Also it may be possible that low PLP levels in cases are associated with reduced transsulfuration, maybe leading to increased tHcy levels after protein-rich meals. However, without having measured plasma tHcy in response to a methionine load we cannot draw any conclusion on this. Among controls, we observed an inverse association between methionine intake and plasma tHcy, taking into account confounding by age, sex, and intake of
vitamins B6, B12, 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
61 Chapter 4 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 risk of myocardial infarction in this population.
Effects of folate, and to some extent of vitamin B6 (especially plasma PLP), were stronger in the subsample shown in this publication than in the original study population, due to higher intake of these vitamins among controls of the subsample, as compared to the orginal control group. Subjects in the subsample were selected on the criterion of having a sufficient volume of stored plasma left. 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 B6, and maybe plasma tHcy, with risk of myocardial infarction. To get some indication of the effect of this bias, we estimated expected tHcy levels in cases and controls of the original study group, that had plasma levels of PLP available (281 cases, 275 controls), based on a linear regression equation obtained from data in the subsample. Included independent variables were intake of the vitamins, plasma PLP, age, sex, alcohol consumption, plasma total/HDL cholesterol, methionine intake, and history of high blood pressure. In the original study group, estimated tHcy levels were significantly higher in cases than in controls. Also, the increase in risk per 1 SD increase in plasma tHcy was similar to the one observed in the smaller study population. Hence, the selection bias might have strenghtened the association of tHcy with risk of myocardial infarction, but can not have been totally responsible for the observed effect. The case-control design of the study leaves the possibility that the higher tHcy 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 tHcy or vitamins. We tested this in 106 cases, for whom we had plasma levels of tHcy 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 tHcy to be significantly lower (-1.4 ± 0.25 [SE] umol/L, P=0.0001) than the values measured in blood obtained 8 weeks after the event. Consequently, any temporary decrease in tHcy values in cases could only have caused the association with myocardial infarction to be weaker, but certainly not stronger. For plasma PLP the hospital values were significantly lower than the levels measured in blood that was drawn at the time of the interview. Therefore,
62 Homocysteine, B-vitamins and Myocardial Infarction any temporary decrease in plasma PLP may have strenghtened the inverse association of plasma PLP with myocardial infarction. Plasma levels of vitamin B12 and folate did not differ between bloods drawn in the hospital and at the interview. Medication used by subjects could have affected their plasma levels of tHcy or vitamins. However, most drugs that are known to have an effect on plasma tHcy or vitamins, like methotrexate, anti-convulsants, or penicillamine1 are not prescribed more frequently in myocardial infarction patients. Recall bias or changes in dietary habits of cases since the event could have influenced tHcy-myocardial infarction or vitamin-myocardial infarction relationships. However, we observed very similar correlations between dietary and plasma levels of vitamin B6, BI2 and folate for cases and controls, which reduces the possibility that either of those two biases has occurred. Since dietary assessment by means of a questionnaire is rather crude, this may explain why the associations of plasma tHcy 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 tHcy and vitamins, which might have had an effect on tHcy 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 tHcy and the similarity with findings from other case-control (including prospective ones), and cross-sectional studies on tHcy and risk of coronary heart disease,8,9'28"32,33,34 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 tHcy with coronary heart disease. They observed a significant 32 percent increase for each 1 SD (4 umol/L) increase in plasma tHcy, after adjustment for established coronary risk factors, similar to the 35 percent increase for each 3 umol/L in our study. Stampfer et al.,8 in their study with data of the Physicians' Health Study, probably a better nourished study population than that of Arnesen et al., found a threshold level above risk was increased. They observed that the higher mean tHcy level in cases was due to an excess of cases with tHcy values above the 95th percentile of control subjects (15.8 umol/L). Whether the level of nourishment of the studied population influences the shape of the relationship of tHcy 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,
63 Chapter 4 two women) and one male control subject had folate intakes less than 200 ug/d, the current RDA 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/L, US NHANES II). When plotting folate intakes against plasma tHcy levels, we observed that tHcy reached its nadir at folate intakes higher than 350 to 400 ug/d, similar to the findings of Selhub et al.10 Plasma folate was inversely associated with plasma tHcy across the entire distribution of plasma folate levels, also above the critical level of 6.8 nmol/L. These findings indicate that for optimalizing tHcy levels, recommended folate intake should be at least 300-400 ug/d. Several explanations for the positive association of tHcy with vascular disease have been proposed. Homocysteine thiolactone, a reactive form of homocysteine, can modify low density lipoprotein (LDL), leading to aggregation and increased uptake of LDL by macrophages. Homocysteine released from the LDL 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 tHcy is an important independent risk factor for coronary disease. The graded effect suggests that lowering of plasma tHcy 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 towards normalizing tHcy levels. Supplementation with folate has been shown to succesfully reduce tHcy levels within a few weeks, even in subjects with genetically caused hyperhomocysteinemia.12 Whether this will also reduce risk of coronary heart disease can only be shown by observational studies and randomized trials relating intake and status to incidence of coronary heart disease.37 In the meantime it might be worthwhile to reconsider the RDA of 200 ug/d for folate, and perhaps restore the previous value of 400 ug/d.
64 Homocysteine, B-vitamins and Myocardial Infarction
Acknowledgements
The research was supported by grants HL-24423 and HL-21006 and institutional training grant HL-07575 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 (Dr. Marvin H. Kendrick), Framingham Union Hospital (Dr. Marvin Adner), Leonard Morse Hospital (Dr. L. Frederick Kaplan), Mount Auburn Hospital (Dr. Leonard Zir), Newton-Wellesley Hospital (Dr. James Sidd), and Walfham Hospital (Dr. Solomon Gabbay). We would like to thank Stephanie Bechtel, Tom Gaziano, 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. Thromb Res 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 Bl2, B6, and folate on blood homocysteine levels. Ann N Y Acad Sci 1992;30:197-204.
5. Ueland PM, Refsum H, Brattstrom 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. Homocysteine 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 homocysteine and risk of myocardial infarction in US physicians. JAMA 1992;268:877-81.
9. Arnesen E, Refsum H, Bnaa KH, Ueland PM, F 65 Chapter 4 10. Selhub J, Jacques PF, Wilson PWF, Rush D, Rosenberg IH. 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, Becker PJ. Vitamin B12, vitamin B6, and folate nutritional status in men with hyperhomocysteinemia. Am J Clin Nutr 1993;57:47-53. 12. Brattstrom LE, Israelsson B, Jeppsson JO, Hultberg BL. 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, Delport R. Hyperhomocysteinemia and the response to vitamin supplementation. Clinical Investig 1993;71:993-8. 14. Pancharaniti N, Lewis CA, Sauberlich HE, et al. Plasma homocysteine, folate, and vitamin B12 concentrations and risk for early-onset coronary artery disease. Am J Clin Nutr 1994;59:940-8. 15. Allen RH, Stabler SP, Savage DG, Lindenbaum J. 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 HDL2 and HDL3 cholesterol, Apo A-I and Apo A-II, and increased risk of myocardial infarction. Circulation 1992;85:22-9. 17. Paffenbarger RS, 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 B6 analysis. In: Pesce AJ, Kaplan LA, eds. Methods in Clinical Chemistry. St Louis: CV Mosby, 1987:558-68. 19. Willett WC, Sampson, L, Stampfer MJ, et al. Reproducibility and validity of a semiquanti• tative food frequency questionnaire. Am J Epidemiol 1985;122:51-65. 20. Willett WC, Sampson, 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 reproducibility of a dietary questionnaire. Int J Epidemiol 1987;16:392-8. 22. Willett WC. Nutritional Epidemiology. New York, NY: Oxford University Press, 1989. 23. Department of Agriculture. Agriculturural handbook no. 8 series. Composition of food - raw, processed, and prepared. Washington, D.C.: Government Printing Office, 1963-1988. 24. Willett W, Stampfer MJ. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol 1986;124:17-27. 66 Homocysteine, B- vitamins and Myocardial Infarction 25. Brattstrom L, Israelsson B, Norrving B, Bergqvist D, Thome J, Hultberg B, Hamfelt A. Impaired homocysteine metabolism in early-onset cerebral and peripheral occlusive arterial disease. Atherosclerosis 1990;81:51-60. 26. Willett WC. Does low vitamin B6 intake increase the risk of coronary heart disease? In: Reynolds RD, Leklem JE eds. Vitamin B6: Its Role in Health and Disease. New York, NY: 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, Grossman M, Wilson D, Upson B. Homocysteine in daily practice: levels in coronary artery disease. Coronary Artery Dis 1990;1:215-20. 30. Ubbink JB, Vermaak WHJ, Bennett JM, Becker PJ, Van Staden DA, Bissbort S. 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, Wilson PW, Schaefer EJ, Malinow MR. Plasma homocysteine levels in men with premature coronary artery disease. J Am Coll Cardiol 1990;16:1114-9. 32. Kang SS, Wong PW, Cook HY, Norusis M, Messer JV. Protein-bound homocysteine. 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 homocysteinemia as a cardiovascular risk factor in men. Arterioscler Thromb 1994;14:460-4. 34. Dalery K, Lussier-Cacan S, Selhub J, Davignon J, Latour Y, Genest J. Homocysteine and coronary artery disease in French Canadian subjects: relation with vitamins B12, B6, 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 USA 1994;91:6369-73. 37. Stampfer MJ, Willett WC. Homocysteine and marginal vitamin deficiency. The importance of adequate vitamin intake. JAMA 1993;270:2726-7. 67 Plasma Total Homocysteine and Future Risk 5 of Angina Pectoris with Evidence of Severe Coronary Atherosclerosis Abstract Background Homocysteine is an amino acid with suggested atherogenic and thrombogenic properties. Epidemiologic studies have demonstrated that elevated plasma total homocysteine (tHcy) is associated with increased risk of atherosclerotic vascular disease. We assessed prospectively the risk of angina pectoris associated with elevated plasma tHcy levels. Methods The Physicians' Health Study is a randomized, double-blind, placebo- controlled trial of aspirin and 6-carotene in 22 071 US male physicians. A total of 14 916 subjects, aged 40 to 84 years, with no prior history of cardiovascular disease provided blood samples at baseline and were followed for 5 years, with 99.7% morbidity and 100% mortality follow-up. We measured tHcy and several metabolites in plasma samples from 218 men who developed angina pectoris and had coronary artery bypass surgery, significant occlusion at angiography, or other evidence of severe coronary atherosclerosis during the follow-up period, and from 218 apparently healthy control men, matched for age and smoking. Results The mean plasma level of tHcy was slightly higher in cases of angina pectoris (10.6 ± 5.8 [SD] umol/L) compared to controls (10.2 ± 4.1 umol/L, P=0.33). There were no significant differences even at plasma levels above the 95th percentile (16.6 umol/L) of controls. Mean plasma levels of cysteine, cystathionine, methionine, dimethylglycine, serine and glycine did not differ significantly between cases and control subjects either. Taking the lowest quintile of plasma tHcy (< 7.3 umol/L) as the reference, odds ratios of angina pectoris for the upper four quintiles were 1.6, 2.7, 1.5, and 1.6, respectively, controlling for confounding factors. However, all confidence intervals, except for the estimate of 2.7, included one. Conclusions These findings are much less striking than other epidemiologic observations, which may be related to the fact that this is a well-nourished population. However, in a previous nested case-control study within the Physicians' Health Study, we observed a positive association between tHcy and risk of myocardial infarction. This raises the possibility that any effect of moderately elevated tHcy is more likely to be thrombogenic than atherogenic. Petra Verhoef, Charles H. Hennekens, Robert H. Allen, Sally P. Stabler, Walter C. Willett, Meir J. Stampfer (submitted). 69 Chapter 5 Introduction Homocysteine is a thiol-containing amino acid, formed through degradation of methionine. Several atherogenic and thrombogenic effects have been postulated for homocysteine.1 Extremely high levels of plasma total homocysteine (designated as tHcy), are present in subjects with inborn errors in homocysteine, cobalamin (vitamin B12), or folate metabolism. The classic genetic disorder homocystinuria is caused by a homozygous defect in the enzyme cystathionine B-synfhase and is associated with premature vascular occlusive disease.2 Recent epidemiologic research has indicated that moderate elevations of plasma tHcy, either caused by less severe genetic defects or low intake of vitamins that serve as cofactors to metabolizing enzymes, may also be independently associated with increased risk of coronary heart disease.3 However, most of these studies were retrospective or cross- sectional studies. Three prospective studies on tHcy and risk of coronary heart disease have been reported: one observing no association,4 and two others observing a positive association.5'6 A variety of mechanisms has been described to explain atherogenic and thrombogenic effects of elevated plasma tHcy, none of which has been demonstrated uniformely, however.1 Homocysteine may reduce the resistance of vascular endothelial cells to thrombosis (e.g. by decreasing the function of tissue plasminogen activator, thrombomodulin and protein C),7"9 affect other hemostatic factors (e.g. by increasing factor VIIc activity or platelet aggregation),10'11 or promote oxidative modification and uptake of low density lipoprotein (LDL) cholesterol in the vascular wall.12,13 Furthermore, homocysteine may inhibit growth of endothelial cells and stimulate proliferation of smooth muscle cells, subsequently leading to thickening of arterial walls.14 These mechanisms may explain how homocysteine could promote both the development of atherosclerotic plaques, as well as the formation of thrombi, which may finally result in a clinical cardiovascular event.1 In the present study, we prospectively investigated the association of plasma tHcy with risk of angina pectoris, supported by evidence of severe coronary atherosclerosis. Since angina pectoris largely represents the atherosclerotic aspect of coronary heart disease, this investigation might indicate the relative importance of the atherogenic effect of moderately elevated plasma tHcy, as opposed to the thrombogenic effect. In addition to tHcy, we measured plasma levels of cysteine, cystathionine, methionine, serine, glycine, and dimethylglycine, which are amino acids of homocysteine metabolism. Abnormal levels of these amino acids may give indications of metabolic disturbances caused by genetic enzyme defects or by low 70 Homocysteine and Angina Pectoris nutritional status of vitamins B6, B,2, or folate, which are cofactors in homocysteine metabolism.15,16 Methods Design and study population We conducted a prospective nested case-control analysis in the Physicians' Health Study in which blood samples were collected at baseline. The Physicians' Health Study is an ongoing randomized, double blind, placebo-controlled 2x2 factorial trial of aspirin and B-carotene.1718 A total of 22 071 US male physicians, aged 40-84 years in 1982, were enrolled. Men were excluded if they had a prior history of cardiovascular disease (including angina pectoris), cancer (except non- melanoma skin cancer), current renal or liver disease, peptic ulcer or gout, contrain• dication to aspirin, or current use of aspirin, other platelet active agents, or vitamin A supplements. At enrollment, study participants completed questionnaires concerning medical history, behavior, and use of medications and vitamin supplements. A limited food-frequency questionnaire, mainly focusing on vitamin A and carotene, was also included. Before randomization, between August 1982 and December 1984, we sent kits for blood sampling to the participants who were instructed to have their blood drawn into EDTA Vacutainer tubes, to centrifuge them, and to return the plasma in polypropylene cryopreservation vials by prepaid overnight courier. The kit included a cold pack to keep the specimens cool (but not frozen) until receipt at Charming Laboratory the next morning, when they were aliquoted and stored at -80°C. During storage, no specimen thawed or warmed substantially. We received specimens from 14 916 (68%) of the randomized physicians, over 70% between September and November 1982. The study participants reported cardiovascular events and other acute illnesses annually. Persistent nonresponders to the questionnaires were telephoned. Deaths were usually reported by the families or postal authorities. The follow-up for morbidity was 99.7% complete and for mortality was 100% complete. Subjects with angina pectoris and additional evidence of severe coronary atherosclerosis were selected as cases. The evidence was: coronary artery bypass surgery (n=T49), significant occlusion at coronary angiography (n=39), positive treadmill test (n=21), positive thallium scan (n=8), and positive radionuclide ventriculography (n=l). The End Points Committee of physicians blinded to treatment status, confirmed reported angina pectoris after reviewing additional evidence in all available medical records. 71 Chapter 5 Levels of tHcy were measured in plasma samples from 218 men with confirmed angina pectoris and 218 control men, matched individually for age and smoking habits. Control men were free from any type of clinical coronary heart disease at the time of the case's diagnosis. They were randomly selected from participants who met the matching criteria of age (± 1 year), smoking status (current, past, or never smoker), and length of follow-up (by 6-monfh intervals). Blood samples of cases and control subjects were treated identically and stored at - 80 ° C for the same amount of time. Laboratory methods Levels of tHcy and the metabolites cysteine, cystathionine, methionine, dimethylglycine, serine and glycine were measured in the laboratory of Dr. R.H. Allen and Dr. S.P. Stabler. Total homocysteine, referred to as tHcy - the sum of homocysteine, the homocysteinyl moieties of homocystine, and cysteine- homocysteine mixed disulfide, whether free or protein-bound - and the metabolites were assayed by capillary gas chromatography-mass spectrometry. A stable isotope- labeled form of each of the metabolites was added to samples followed by partial purification using ion exchange resins. The metabolites were then converted to their rerf-butyldimethylsilyl derivatives and analyzed by capillary gas chromatography- mass spectrometry using selected ion monitoring and positive ion electron impact ionization.16 Samples from case-control pairs were handled together, blindly and identically throughout processing and analysis, and the position within pairs was varied randomly. Blind paired quality-control samples (n=9) were interspersed at random among the specimens. The quality-control samples were aliquots of a large, well-mixed plasma pool from healthy volunteers that were treated identically to the samples collected from the participants. The mean within-pair coefficient of variation in these paired quality control samples was 7.0% for tHcy and varied between 4.1% and 6.8% for the other metabolites. Plasma levels of total and high- density lipoprotein (HDL) cholesterol were measured in the laboratory of Dr. F.M. Sacks.19 Statistical analysis Risk factors for coronary heart disease as well as plasma levels of tHcy and other metabolites were compared for cases and controls. Differences were tested for significance by using a paired Student's t test for continuous variables and a McNemar paired chi-square test for proportions. In control men, Spearman rank correlation coefficients for the association of tHcy with several possible risk factors 72 Homocysteine and Angina Pectoris of coronary heart disease were calculated. The association of plasma fHcy with risk of angina pectoris was evaluated by calculating odds ratios (OR) and 95% confidence intervals (CI), using conditional logistic regression (conditional on age and smoking habits) to control for several risk factors for coronary heart disease, simultaneously. We calculated ORs in three ways: 1. for subjects with hyperhomocysteinemia (defined as plasma fHcy levels above the 95th percentile [16.6 umol/L] of controls); 2. by quintiles of plasma fHcy, based on the distribution among controls; 3. continuously, per 1 SD (approximately 5 umol/L) increase in plasma fHcy. Unmatched analyses, using multiple logistic regression (including age and smoking habits as independent variables), were performed by strata of age (above or below median age), total/HDL cholesterol (above or below median level), aspirin treatment (aspirin versus placebo), and history of high blood pressure (yes versus no). Subjects with systolic blood pressure > 150 mm Hg, diastolic blood pressure > 90 mm Hg, or on treatment for hypertension, were considered to be hypertensive. The EGRET package (version 0.25.1) and SAS package (version 6.09) were used for conditional and multivariate logistic regression analyses, respectively. All P-values are two-tailed. Results Coronary risk factors Risk factors for coronary heart disease and other characteristics at enrollment are shown for 218 pairs of confirmed angina pectoris cases and controls, matched for age and smoking habits (Table 5-1). The mean total/HDL cholesterol ratio, and mean systolic and diastolic blood pressure were significantly higher in cases than controls. Cases had significantly higher proportions of men with history of diabetes, hypertension, and family history of myocardial infarction. Quetelet's index was nonsignificantly higher, and alcohol consumption nonsignificantly lower in cases, as compared to controls. Age and cigarette smoking status, both matching variables, were identical among cases and controls. Among controls, we calculated Spearman correlation coefficients relating plasma tHcy to individual risk factors and other characteristics. Plasma tHcy was significantly positively correlated with alcohol consumption. All other correlations were small and nonsignificant. 73 Chapter 5 Table 5-1. Characteristics of 218 cases of angina pectoris and 218 controls, including correlations of risk factors with plasma total homocysteine Cases Controls Mean Mean Correlation P-vatue+ or SD or SD with tHcy: case-control % % r* P difference Age (years) 58.3 8.6 58.3 8.6 0.08 0.27 MV Body mass index (kg/m2) 25.2 2.6 24.9 3.1 0.00 0.99 0.19 Total/HDL cholesterol 5.97 3.22 4.86 1.55 0.08 0.27 0.0001 Systolic BP (mm Hg) 131 12 128 12 0.05 0.52 0.06 Diastolic BP (mm Hg) 81 8 80 7 0.05 0.48 0.02 Hist, of hypertension (%) 28.4 19.7 -0.10 0.12 0.04 Smoking Current (%) 8.3 8.3 0.07 0.28 MV Past (%) 48.2 48.2 Never (%) 43.5 43.5 Alcohol intake (drinks/day) 0.46 0.46 0.53 0.46 0.17 0.01 0.09 History of diabetes (%) 6.5 0.9 -0.05 0.51 0.006 Family history of MI (%) 18.4 10.1 0.05 0.44 0.02 tHcy, total homocysteine; HDL, high density lipoprotein; BP, blood pressure; MV, matching variable. * Spearman correlation coefficient of plasma tHcy with risk factors, for controls only. + P-value of difference in risk factor between cases and controls, paired t test for continuous values and McNemar chi-square test for proportions. Levels of tHcy The mean level of plasma tHcy was shghtly, but nonsignificantly, higher among the 218 men with confirmed angina pectoris than among the matched controls (Table 5-2). Besides reported angina pectoris, all cases had additional evidence of coronary atherosclerosis, of which coronary artery bypass surgery could be considered the strongest one. The mean tHcy level of the 149 cases who underwent coronary artery bypass surgery was nonsignificantly higher (10.9 ± 6.6 [SD] umol/L) than among the cases with other additional evidence (10.2 ± 3.2 umol/1, P unpaired t test on log-transformed tHcy=0.62). When restricting analyses to those 149 cases and their matched controls, the case-control difference for plasma tHcy was 0.4 umol/L (P=0.51) (data not shown). Eight cases and 11 controls had hyperhomocysteinemia (defined as levels 74 Homocysteine and Angina Pectoris above the 95th percentile of controls [16.6 umol/L]). When using the 95th percentile (15.8 umol/L) of control subjects of a previous study of tHcy and risk of myocardial infarction in the Physicians' Health Study, the number of subjects with hyperhomocysteinemia was 13 for both cases and controls. However, cases showed a tendency to have more extreme values than controls: among cases, the highest tHcy level was 63.1 umol/L, followed by 53.5 umol/L, and 39.3 umol/L, whereas for control subjects the highest three values were 34.8, 30.0 and 28.6 umol/L. Table 5-2. Plasma levels of total homocysteine and several metabolites in cases of angina pectoris and controls Cases (n=218 ) Controls (n=218) Mean SD GM Mean SD GM P* tHcy (umol/L) 10.6 5.8 9.9 10.2 4.1 9.6 0.33 Cystathionine (nmol/L) 226 95 208 223 97 206 0.71 Cysteine (umol/L) 316 47 313 309 46 305 0.07 Methionine (umol/L) 27.1 7.6 26.2 28.1 8.5 27.0 0.15 Dimethylglycine (umol/L) 4.61 1.21 4.47 4.60 1.42 4.43 0.98 Serine (nmol/L) 90.8 18.9 89 91.8 21.8 89 0.62 Glycine (nmol/L) 206 42 202 211 46 206 0.30 GM, geometric mean; tHcy, total homocysteine. * Case-control difference, using paired t test. Levels of homocysteine metabolites Mean plasma levels of cysteine, cystathionine, methionine, dimethylglycine, serine and glycine did not differ significantly between cases and control subjects (Table 5-2). For cases (n=8) and controls (n=ll) with tHcy levels higher than 16.6 umol/L, we separately show plasma levels of tHcy and other metabolites (Table 5- 3). Despite the hyperhomocysteinemia, the levels of most other metabolites were similar to those observed in the whole study population. Hyperhomocysteinemic cases had relatively high levels of cystathionine and dimethylglycine, and relatively low methionine in comparison with hyperhomocysteinemic controls. Only for cys• tathionine was the difference statistically significant. 75 Chapter 5 Table 5-3. Plasma levels of total homocysteine and several metabolites in hyperhomocysteinemic cases of angina pectoris and controls Cases (n=8) Controls (n=10) Mean SD GM Mean SD GM P* tHcy (umol/L) 32.1 18.0 28.2 23.4 6.2 22.7 0.31 Cystathionine (nmol/L) 292 76 284 206 72 194 0.02 Cysteine (umol/L) 298 63 293 320 52 317 0.35 Methionine (umol/L) 25.6 6.5 24.9 29.0 8.9 27.2 0.44 Dimethylglycine (umol/L) 4.70 1.12 4.61 4.71 1.73 4.48 0.83 Serine (nmol/L) 85.4 22.1 83 98.6 23.0 96 0.23 Glycine (nmol/L) 216 59 211 222 44 218 0.75 GM, geometric mean; tHcy, total homocysteine. * Case-control difference, unpaired t test with log-transformed variables. Vitamin use Current use of multivitamins was slightly more frequent among controls than among cases (22.9% versus 18.4%, P=0.81). Among controls, mean plasma tHcy was 16% lower in current users of multivitamins, as compared to nonusers (P unpaired t test on log-transformed tHcy=0.0008). Among cases the difference was -7% (P=0.21). None of the cases and only one control subject with hyperhomocysteinemia were using multivitamin supplements. Risk of angina pectoris The OR of angina pectoris, conditional on age and smoking habits, for individuals with levels above the 95th percentile of controls, compared to those with levels below the 90th percentile (13.7 umol/L) was 0.7 (95% CI, 0.3-1.8). When further adjusting for diabetes, high blood pressure, Quetelet's index, aspirin assignment, total cholesterol/HDL cholesterol ratio, alcohol consumption and fasting time, the OR was 0.6 (95% CI, 0.2-1.9). We also calculated ORs for quintiles of plasma tHcy, based on the distribution among controls, with the lowest quintile as a reference group (Table 5-4). ORs of angina pectoris were above unity for all four upper quintiles, but most CIs included the null value and there was no graded effect. The OR of angina pectoris for men in the upper quintile of plasma tHcy, compared to those in the bottom quintile, conditional on age and smoking habits, was 1.5 (95% CI, 0.8-2.9). 76 Homocysteine and Angina Pectoris The OR was 1.6 (95% CI, 0.7-3.5) after controlling for other possible risk factors. Although analyses by quintiles did not show a graded effect, we tested whether there was an overall trend of increasing risk of angina pectoris with increasing levels of plasma tHcy. The relative risk for each 5 umdl/L (approximately 1 SD) increase in plasma tHcy was 1.1 (95% CI, 0.9-1.3), controlling for age and smoking habits in a conditional regression analysis. After controlling for other factors, the OR was 1.0 (95% CI, 0.8-1.3). Table 5-4. Risk of angina pectoris by quintiles of plasma total homocysteine Quintiles* 1 2 3 4 5 <7.3 <8.6 < 10.2 < 12.1 > 12.1 Age-and smoking-adj.+OR 1 1.5 2.1 1.7 1.5 95% confidence interval (0.8-2.8) (1.1-3.9) (0.9-3.3) (0.8-2.9) Multivariate-adj.+OR 1 1.6 2.7 1.5 1.6 95% confidence interval (0.8-3.3) (1.2-5.9) (0.7-3.2) (0.7-3.5) * Quintiles are based on distribution of control subjects; boundaries (umol/L) of quintiles are given. + Matched logistic regression, conditional on age and cigarette smoking. + Adjusted for age, history of diabetes, history of high blood pressure, family history of ischemic heart disease, total/HDL cholesterol, body mass index, cigarette smoking, alcohol consumption, aspirin assignment and fasting time. We wished to evaluate whether any association between plasma tHcy and risk of angina pectoris differed by strata of three major risk factors for coronary heart disease, i.e. age, blood pressure, and total/HDL ratio, or by aspirin assignment. Therefore, we repeated the risk analyses by quintiles among strata of these factors, using logistic regression analysis, with age, smoking habits and other possible confounders as independent variables in the model. We could use this unmatched approach, since matched analyses yielded virtually the same ORs for quintiles, conditionally on age and smoking habits. The association of plasma tHcy with risk of angina pectoris did not significantly differ between strata of age (younger versus 11 Chapter 5 older than median age at enrollment [60 years]), total/HDL cholesterol ratio (lower versus higher than median), history of hypertension (no versus yes), or aspirin assignment (placebo versus active) (data not shown). Also, we found no substantial differences in risks by quintiles of plasma tHcy for strata of cases who were diagnosed with angina pectoris in 1986 or before (n=96) and those diagnosed after 1986 (n=122), including all controls in both analyses, and adjusting for effects of age and smoking habits and other factors simultaneously (data not shown). Discussion In this prospective, nested case-control study, we observed virtually no association between plasma tHcy and risk of angina pectoris with evidence of severe coronary atherosclerosis, not even at hyperhomocysteinemic levels. Relative to men with tHcy levels in the lowest quintile, the ORs of angina pectoris for men with higher tHcy values were above 1, but most CIs included the null value and there was no dose-response effect. Previous findings in Physicians' Health Study Contrastingly, in a previous report on a case-control study nested within the Physicians' Health Study, over the same period . of follow-up we observed a statistically significant 3.4-fold relative risk of myocardial infarction for men with hyperhomocysteinemia (> 15.8 umol/L, the 95th percentile of controls in that population), compared to those with normal levels (lower than the 90th percentile of controls).5 Elevated risk appeared to be limited to those with hyperhomocys• teinemia, although the data were compatible with a graded effect as well. In the present population, tHcy levels above that cutoff-point were not associated with increased risk of confirmed angina pectoris. The treatment of blood samples in both studies was exactly the same. Although methods of tHcy analysis differed between the studies, both methods yield very comparable tHcy values.20 The mean wifhin- pair coefficient of variation for tHcy determination in the present study was somewhat higher than in the previous study (7.0% versus 3.2%). However, there is no reason to believe that there has been non-differential exposure misclassification in the present study. 78 Homocysteine and Angina Pectoris Disease classification The endpoint of angina pectoris required additional evidence of severe coronary atherosclerosis. Most cases had undergone coronary artery bypass surgery or had significant occlusions at coronary angiography. However, in control subjects there was no objective information on the extent of coronary atherosclerosis. Although one can expect some coronary atherosclerosis in middle-aged men, the fact that control men had no clinical signs of coronary artery disease must have provided enough contrast between the study groups. However, there is a possibility that the control subjects with hyperhomocysteinemia had in fact silent, but substantial coronary artery disease, associated with coronary atherosclerosis. Another explanation for our null finding may be that moderate elevation of plasma tHcy might be more strongly related to thrombotic events, such as a myocardial infarction, than to atherosclerotic events, such as angina pectoris. Other epidemiologic studies All previously reported epidemiologic studies with angiographically defined coronary atherosclerosis as an endpoint, have reported positive associations with plasma tHcy levels.21"28 However, unlike our cases of angina pectoris, many cases in these studies had prior myocardial infarction, which has a strong thrombotic component. Therefore, a strong conclusion about the possible atherogenic effect of moderately elevated tHcy cannot always be drawn from these studies. This does not apply to studies of carotid artery thickening, however. Recently, Selhub et al.,29 with data of the Framingham Heart Study, showed an OR of 2.0 (95% CI, 1.4-2.9) of having 25% or more stenosis in the extracranial carotid-artery for subjects with tHcy concentrations higher or equal to 14.4 umol/L. In another case-control study among asymptomatic subjects, tHcy was positively associated with carotid-artery thickening as well.30 Carotid-artery thickening may be associated with coronary atherosclerosis.31 Some of the above mentioned studies found a linear trend of increasing tHcy levels associated with severity of atherosclerosis,26,29 providing additional evidence for an atherogenic effect of moderately increased plasma tHcy. Most epidemiologic studies with myocardial infarction or ischemic heart disease as an endpoint have shown positive associations with moderately elevated plasma tHcy.3,6,32"34,35 Two case-control studies36,37 and one prospective study4 have failed to show such a relationship. 79 Chapter 5 Prospective design Thus, most other epidemiologic studies support the hypothesis that moderately elevated tHcy may lead to coronary heart disease, either by promoting formation of atherosclerosis or by promoting thrombotic events. However, the design of the majority of the studies was retrospective or cross-sectional, and some study samples were rather small. Our study employs a prospective design with a clinical definition of coronary atherosclerosis, i.e. angina pectoris with strong evidence of severe coronary obstruction. However, it is one of the few studies that shows no clear positive association between elevated tHcy and coronary heart disease. The use of a prospective design reduces the possibility that tHcy levels in cases were altered by the disease, medication or changes in lifestyle associated with the disease. Samples of cases and controls were stored for the same amount of time, and were handled together and identically throughout thawing, aliquoting and sending, but any nondifferential change in plasma tHcy may have caused an underestimation of the effect. However, the effects of storage for years, nonfasting specimens, and repeated thawing are relatively small.20 Also, aspirin assignment and length of follow-up did not influence our results. Metabolites and vitamin status Homocysteine can be metabolized through two enzymatic pathways: transsulfuration and remethylation. 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 B6. Subsequently cystathionine is converted to cysteine, in another vitamin B6-dependent reaction. In remethylation, homocysteine receives a methyl group from 5-methyltetrahydrofolate (a vitamin B[2- dependent reaction) or from betaine to form methionine. In the reaction involving betaine, which is independent of folate and vitamin B12, dimefhylglycine is 15,16 formed. Consequently, inadequate intakes of folate, vitamin BI2, and vitamin B6 may cause elevation of plasma tHcy. None of the above mentioned metabolites, nor glycine, which takes part in various reactions of homocysteine metabolism, differed between cases and controls in this study. Also, the 19 cases and controls with hyperhomocysteinemia had virtually the same levels of these metabolites as seen in the total study sample. However, compared to 11 hyperhomocysteinemic controls, the eight cases with hyperhomocysteinemia had higher levels of cystathionine and dimefhylglycine, indicating that tissue deficiency of folate or vitamin B12 was probably more common among cases with hyperhomocysteinemia than among controls.16 80 Homocysteine and Angina Pectoris Accordingly, due to impaired remethylation, methionine levels were lower. Cases with hyperhomocysteinemia also showed reduced serine levels, as previously reported by another study in patients with hyperhomocysteinemia, probably because of increased requirement of serine in both routes of homocysteine catabolism.38 Combining this information with the fact that current multivitamin use was lower among cases, it seems plausible that a lower status of vitamin B12 and folate, and maybe vitamin B6, may have contributed to the slightly higher tHcy levels in cases. However, lack of plasma and dietary intake levels of the vitamins and of any measurement of inherited enzyme defects, makes it difficult to explain what the backgrounds of hyperhomocysteinemia were among cases and controls in this population. The fact that US physicians are probably better nourished than the general population, and real deficiencies of these vitamins are likely to be uncommon among this population, may explain why the association of tHcy with coronary artery disease was weaker in our study, compared to other studies. Pathophysiological mechanisms Homocysteine is thought to have thrombogenic as well as atherogenic properties, as supported by in vitro studies and experiments in animals.1 Some of the thrombogenic properties have recently been supported by studies in humans.10,39 During copper-catalyzed oxidation of homocysteine, hydrogen peroxide is formed, which may have detrimental effects on the vascular endothelium.40 Subsequently, this may cause a decreased function of tissue plasminogen activator, throm• bomodulin and protein C, and an increased factor VIIc activity, platelet count and platelet adhesiveness.7"10,41 Furthermore, endothelial cell damage may cause increased uptake of modified LDL cholesterol in the vascular wall.12 In addition, homocysteine may even promote lipid peroxidation and susceptibility to oxidation of LDL cholesterol,13 although this was not confirmed by a recently conducted study in humans.42 A recently proposed atherogenic mechanism involves inhibition of growth of endothelial cells and stimulation of proliferation of smooth muscle cells by homocysteine, finally leading to thickening of arterial walls.14 All these processes may lead to significant occlusions of blood vessels and eventually to thrombotic events. 81 Chapter 5 Conclusion In contrast to the relatively strong positive associations found in other epidemiologic investigations, our study does not support the hypothesis that moderately elevated tHcy is associated with coronary atherosclerotic disease. The fact that our study population consisted of a unique well-nourished population, diminishing case-control contrast with regard to vitamin status, may be in part responsible for our null finding. Thus, extrapolating findings to the population at large cannot be easily done. Considering the fact that we did show a positive association in a previous nested case-control study of myocardial infarction within the Physicians' Health Study, one could speculate that the thrombogenic part of causal mechanims relating moderately elevated plasma tHcy to vascular disease may be more important than the atherogenic part. Acknowledgments This study was supported by research grants (CA42182, HL26490, HL34595, CA34944, CA40360, and RROO163-31) from the National Institutes of Health, Bethesda, Maryland. The authors thank the Physicians' Health Study Steering Committee (C. Belanger, MA, J.E. Buring, ScD, N. Cook, ScD, K. Eberlein, MPH, S.Z. Goldhaber, MD., D. Gordon, MA, C.H. Hennekens, MD [chair], S. Mayrent, PhD, R. Peto, MD, B. Rosner, PhD, M.J. Stampfer, MD, F. LaMotte, MPH, W.C. Willett, MD; T. Balzkowski, PhD, and A. Vargosko, PhD - ex officio); the End Points Committee (H. Funkenstein, MD [deceased], S.Z. Goldhaber, MD, M.J. Stampfer, MD, and J.O. Taylor, MD [chair]); Stephanie Bechtel, Mary Ann O'Hanesian, Kim Eberlein, MPH, Mary Lou Lyons, RN, MSN, and Georgina Friedenberg, MPH, provided skilled assistance; Jing Ma, PhD, and Fran Grodstein, PhD, helped preparing the data. References 1. Rees MM, Rodgers GM. Homocysteinemia: association of a metabolic disorder with vascular disease and thrombosis. Thromb Res 1993;71:337-59. 2. Ueland PM, Refsum H, Brattström L. Plasma homocysteine and cardiovascular disease. In: Francis RB Jr, ed. Atherosclerotic Cardiovascular Disease, Hemostasis. and Endothelial Function. New York, NY: Marcel Dekker Inc, 1992:183-236. 82 Homocysteine and Angina Pectoris 3. Malinow MR. Homocysteine and arterial occlusive diseases: a mini review. Clin Chem 1994;40:173-6. 4. Alfthan G, Pekkanen J, Jauhiainen M, et al. Relation of serum homocysteine and lipoprotein(a) concentrations to atherosclerotic disease in a prospective Finnish population based study. Atherosclerosis 1994;106:9-19. 5. Stampfer MJ, Malinow MR, Willett WC, et al. A prospective study of plasma homocysteine and risk of myocardial infarction in US physicians. JAMA 1992;268:877-81. 6. Arnesen E, Refsum H, B(|>naa KH, Ueland PM, F(|)rde OH, Nordrehaug JE. Serum total homocysteine and coronary heart disease. Int J Epidemiol 1995;24:704-9. 7. Hajjar KA. Homocysteine-induced modulation of tissue plasminogen activator binding to its endothelial cell membrane receptor. J Clin Invest 1993;91:2873-9. 8. Lentz SR, Sadler JE. Inhibition of thrombomodulin surface expression and protein C activation by the thrombogenic agent homocysteine. J Clin Invest 1991;88:1906-14. 9. Rodgers GM, Conn MT. Homocysteine, an atherogenic stimulus, reduces protein C activation by arterial and venous endothelial cells. Blood 1990;75:895-901. 10. Schreiner PJ, Wu KK, Malinow MR, Nieto FJ, Heiss G. Plasma homocysteine and hemostasis: the ARIC study. Circulation 1994;89:924. 11. McCully KS, Carvalho AC. Homocysteine thiolactone, N-homocysteine thiolactonyl retinamide, and platelet aggregation. Res Commun Chem Pathol Pharmacol 1987;56:349-60. 12. Parthasarathy S. Oxidation of low density lipoproteins by thiol compounds leads to its recognition by the acetyl LDL receptor. Biochim Biophys Acta 1987;917:337-40. 13. Heinecke JW, Rosen H, Suzuki LA, Chait A. The role of sulfur-containing amino acids in superoxide production and modification of low density lipoprotein by arterial smooth muscle cells. J Biol Chem 1987;262:10098-103. 14. 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 USA 1994;91:63- 69-73. 15. 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. 16. Allen RH, Stabler SP, Savage DG, Lindenbaum J. Metabolic abnormalities in cobalamin (vitamin BI2) and folate deficiency. FASEB J 1993:7; 1344-53. 83 Chapter 5 17. Stampfer MJ, Buring JE, Willett W, Rosner B, Eberlein K, Hennekens CH. The 2x2 factorial design: its application to a randomized trial of aspirin and carotene in US physicians. Stat Med 1985;4:111-16. 18. Steering Committee of the Physicians' Health Study Research Group. Final report on the aspirin component of the ongoing Physicians' Health Study. N Engl J Med 1989;321:129-35. 19. Stampfer MJ, Sacks FM, Salvini S, Willett WC, Hennekens CH. A prospective study of cholesterol, apolipoproteins, and the risk of myocardial infarction. N Engl J Med 1991;325:373-81. 20. Ueland PM, Refsum H, Stabler SP, Malinow MR, Andersson A, Allen RH. Total homocysteine in plasma or serum: methods and clinical applications. Clin Chem 1993;39:1764-79. 21. Wilcken DEL, Wilcken B. The pathogenesis of coronary artery disease: a possible role for methionine metabolism. J Clin Invest 1976;57:1079-82. 22. Murphy-Chutorian DR, Wexman MP, Grieco AJ, et al. Methionine intolerance: a possible risk factor for coronary disease. J Am Coll Cardiol 1985;6:725-30. 23. Murphy-Chutorian D, Alderman EL. The case that hyperhomocysteinemia is a risk factor for coronary disease. Am J Cardiol 1994;73:705-7. 24. Kang SS, Wong PW, Cook HY, Norusis M, Messer JV. Protein-bound homocysteine. A possible risk factor for coronary artery disease. J Clin Invest 1986;77:1482-6. 25. Genest JJ Jr, McNamara JR, Salem DN, Wilson PW, Schaefer EJ, Malinow MR. Plasma homocysteine levels in men with premature coronary artery disease. J Am Coll Cardiol 1990;16:1114-9. 26. Ubbink JB, Vermaak WHJ, Bennett JM, Becker PJ, Van Staden DA, Bissbort S. The prevalence of homocysteinemia and hypercholesterolemia in angiographically defined coronary heart disease. Klin Wochenschr 1991;69:527-34. 27. Pancharuniti N, Lewis CA, Sauberlich HE, et al. Plasma homocysteine, folate, and vitamin B]2 concentrations and risk of early-onset coronary artery disease. Am J Clin Nutr 1994;59:940-8. 28. Von Eckardstein A, Malinow MR, Upson B, et al. Effects of age, lipoproteins, and hemostatic parameters on the role of homocysteinemia as a cardiovascular risk factor in men. Arterioscler Thromb 1994;14:460-4. 29. Selhub J, Jacques PF, Bostom AG, et al. Association between plasma homocysteine con• centrations and extracranial carotid-artery stenosis. New Engl J Med 1995;332:286-91. 84 Homocysteine and Angina Pectoris 30. Malinow MR, Nieto FJ, Szklo M, Chambless LE, Bond G. Carotid artery intimal-medial wall thickening and plasma homocysteine in asymptomatic adults. The Atherosclerosis Risk in Communities Study. Circulation 1993;87:1107-13. 31. Bots ML. Wall thickness of the carotid artery as an indicator of generalized atherosclerosis. The Rotterdam Study. Thesis, Roterdam: Haveka BV, Alblasserdam, 1993. 32. Israelsson B, Brattstrom LE, Hultberg BL. Homocysteine and myocardial infarction. Atherosclerosis 1988;71:227-33. 33. Malinow MR, Sexton G, Averbuch M, Grossman M, Wilson D, Upson B. Homocysteine in daily practice: levels in coronary artery disease. Coronary Artery Dis 1990;1:215-20. 34. Clarke R, Daly L, Robinson K, et al. Hyperhomocysteinemia: an independent risk factor for vascular disease. N Engl J Med 1991;324:1149-55. 35. Graham I. Interactions between homocysteinaemia and conventional risk factors in vascular disease. Eur Heart J 1994;15:530 (abstract). 36. Wilcken DE, Reddy SG, Gupta VJ. Homocysteinemia, ischemic heart disease, and the carrier state for homocystinuria. Metabolism 1983;32:363-70. 37. Boers GHJ, Smals AGH, Trijbels FJM, et al. Heterozygosity for homocystinuria in premature peripheral and cerebral occlusive arterial disease. N Engl J Med 1985;313:709-15. 38. Dudman NP, Tyrell PA, Wilcken DE. Homocysteinemia: depressed plasma serine levels. Metabolism 1987;36:198-201. 39. Bienvenu T, Ankri A, Chadefaux B, Montalescot G, Kamoun P. Elevated total plasma homocysteine, a risk factor for thrombosis. Relation to coagulation and fibrinolytic parameters. Thromb Res 1993;70:123-9. 40. Starkebaum G, Harlan JM. Endothelial cell injury due to copper-catalyzed hydrogen peroxide generation from homocysteine. J Clin Invest 1986;77:1370-6. 41. Graeber JE, Slott JH, Ulane RE, Schulman JD, Stuart MJ. Effect of homocysteine and homocystine on platelet and vascular arachidonic acid metabolism. Pediatr Res 1982;16:490-3. 42. Blom HJ, Kleinveld HA, Boers GHJ, et al. Lipid peroxidation and susceptibility of low- density lipoprotein to in vitro oxidation in hyperhomocysteinaemia. Eur J Clin Invest 1995;25:149-54. 85 A Prospective Study of 6 Plasma Total Homocysteine and Risk of Ischemic Stroke Abstract Background and purpose Several studies have reported elevated circulating total homocysteine (tHcy) levels in subjects with cerebral atherosclerosis. We assessed prospectively whether high plasma levels of tHcy affect risk of ischemic stroke and evaluated whether high blood pressure modifies any such effect. Methods The study sample was drawn from the Physicians' Health Study, a randomized, double-blind, placebo-controlled trial of aspirin and 8-carotene in 22 071 US male physicians. A total of 14 916 subjects, aged 40 to 84, with no prior history of stroke, transient ischemic attack or myocardial infarction provided blood samples at baseline and were followed for 5 years, with 99.7% morbidity and 100% mortality follow-up. Using a nested case-control design, we assayed tHcy in samples from 109 subjects who subsequently developed ischemic stroke and 427 control subjects. Results The mean plasma concentration of tHcy was slightly higher in cases (11.1 ± 4.0 [SD] umol/L) than in controls (10.6 ± 3.4 umol/L), but the difference was not statistically significant (P=0.12). The crude odds ratio of ischemic stroke for subjects in the upper 20% (> 12.7 umol/L) compared with those in the bottom 80% of tHcy levels was 1.4 (95% confidence interval, 0.8-2.2). The odds ratio was 1.2 (95% confidence interval, 0.7-2.0) after controlling for several risk factors and other potential confounders. In subgroup analyses, elevated tHcy levels appeared to be more strongly predictive of ischemic stroke in normotensives and in men 60 years or younger. Although not statistically significant, in these subgroups increases in risks of 100% and 70% respectively, were observed for men in the upper 20% of tHcy values. Conclusions In this study, the data were compatible with a small but nonsignificant association between elevated plasma tHcy and risk of ischemic stroke. However, since the sample size is small and the confidence intervals are wide, either no association or a moderate increase in risk cannot be excluded, particularly in subgroups otherwise at low risk, e.g., younger men and those with normal blood pressure. Published as: Verhoef P, Hennekens CH, Malinow MR, Kok FJ, Willett WC, Stampfer MJ. A prospective study of plasma homocyst(e)ine and risk of ischemic stroke. Stroke 1994;25:1924-30. 87 Chapter 6 Introduction Homocysteine is a thiol-containing amino acid derived from the metabolism of methionine. The term total homocysteine (tHcy) is generally used to refer to homocysteine, disulphides of homocysteine with itself or with cysteine, and protein- bounds forms.1 Elevated tHcy levels are observed in subjects with metabolic defects in several enzymes. These enzyme defects can be due to genetic disorders or deficiencies of folate, vitamin B6 and vitamin Bl2, essential cofactors in the metabolism of homocysteine.2 Several epidemiologic studies, including a prospective study,3 have shown that markedly elevated levels of tHcy, either in the fasting state or after challenge with an oral dose of methionine, are associated with symptomatic atherosclerotic disease.4,5 However, previous studies of occlusive cerebral disease were retrospective or cross-sectional, and therefore could not exclude the possibility that elevated plasma tHcy was a consequence of the disease or conditions related to it.6"13 In a previous study, we observed that men with high levels of tHcy were at a threefold excess risk of myocardial infarction.3 In the present analysis, we examined the relationship of tHcy with risk of ischemic stroke in apparently healthy men, aged 40 to 84, who were free from diagnosed stroke, transient ischemic attack, and myocardial infarction at baseline when they provided plasma samples. Prior studies suggested that the relation between tHcy levels and atherosclerotic disease might be influenced by the presence of hypertension. Thus, in patients with peripheral arterial occlusive diseases, hypertension was more prevalent in those with elevated tHcy.14 Araki et al.6 concluded that high plasma levels of tHcy in conjunction with hypertension could particularly increase risk for atherosclerotic cerebral infarction. Similarly, a study showed a stronger association of plasma tHcy and carotid artery intimal-wall thickening in hypertensives than in normotensives.15 To address this matter, we analyzed the association between tHcy level and ischemic stroke separately for men with or without high blood pressure. Methods Design and study population We conducted a nested, case-control study in the Physicians' Health Study in which blood samples were collected prospectively. The Physicians' Health Study is an ongoing randomized, double blind, placebo-controlled 2x2 factorial trial of 88 Homocysteine and Ischemic Stroke aspirin and B-carotene.16,17 A total of 22 071 US male physicians, aged 40-84 years in 1982, were enrolled. Men were excluded if they had a prior history of stroke, transient ischemic attack, myocardial infarction, cancer (except nonmelanoma skin cancer), current renal or liver disease, peptic ulcer or gout, contraindication to aspirin, or current use of aspirin, other platelet active agents, or vitamin A supplements. The aspirin component of the trial was terminated on January 25, 1988, principally because of a statistically extreme 44% reduction of first myocardial infarction among the aspirin group. At enrollment, study participants completed questionnaires concerning medical history, behavior, and use of medications and vitamin supplements. A limited food- frequency questionnaire, mainly focused on vitamin A and carotene, was also included. Before randomization, between August 1982 and December 1984, we sent kits for blood sampling to the participants who were instructed to have their blood drawn into EDTA Vacutainer tubes, to centrifuge them, and to return the plasma in polypropylene cryopreservation vials by prepaid overnight courier. The kit included a cold pack to keep the specimens cool (but not frozen) until receipt at Charming Laboratory the next morning, when they were aliquoted and stored at -80°C. During storage, no specimen thawed or warmed substantially. We received specimens from 14 916 (68%) of the randomized physicians, over 70% between September and November 1982. The study participants reported cerebrovascular and cardiovascular events and other acute illnesses annually. Persistent nonresponders to the questionnaires were telephoned. Deaths were usually reported by the families or postal authorities. The follow-up for those with nonfatal outcomes was 99.7% complete and for those with fatal outcomes was 100% complete. Strokes were confirmed by medical records if they were characterized by a typical neurological deficit, sudden or rapid in onset, lasting at least 24 hours, and attributable to a cerebrovascular event. The End Points Committee, comprising four physicians blinded to treatment status, confirmed reported events by review of all medical records, according to those criteria, established in the National Survey of Stroke.18 The diagnosis was confirmed by computed tomographic scans in 95% of patients with ischemic stroke. For fatal cases, we accepted diagnoses based on autopsy, or confirmation by records that the death was due to cerebrovascular disease (International Classification of Diseases [ICD] codes 430-438). Plasma tHcy was measured in samples of 109 subjects who were diagnosed with ischemic stroke (ICD code 434) and 124 control subjects. Ninety-seven cases and controls were individually matched. Control subjects were free from any type of cerebrovascular disease at the time of the case's diagnosis. They were randomly selected from participants who met the matching criteria of age (± 1 year), smoking 89 Chapter 6 status (current, past, or never smoker), and length of follow-up (by 6-month intervals). To increase the statistical power of the study, we also performed unmatched analyses, using data from the total group of 109 subjects with ischemic stroke. The twelve additional cases had been matched to controls that had insufficient plasma for analysis. We also augmented the control group to 427 subjects, by adding 303 controls from our recent study on tHcy and risk of myocardial infarction.3 In that study controls had been matched to cases with myocardial infarction, using the same criteria as described above. Blood samples of cases of ischemic stroke and myocardial infarction and of all the control subjects were treated identically and were analyzed at the same time. Laboratory methods Levels of tHcy were measured by Dr. M. Rene Malinow at the Oregon Regional Primate Center. tHcy - the sum of homocysteine, the homocysteinyl moieties of homocysteine, and cysteine-homocysteine mixed disulfide, whether free or protein- bound - was assayed by high pressure liquid chromatography and electrochemical detection, based on the method of Smolin and Schneider,19 as previously described20 with minor modifications.21 Samples from case-control pairs were handled together, blindly and identically throughout processing and analysis, and the position within pairs was varied randomly. Blind paired quality-control samples (n=26 pairs) were interspersed at random among the specimens. The quality-control samples were aliquots of a large, well-mixed plasma pool from healthy volunteers that were treated identically to the samples collected from the participants. The mean within-pair coefficient of variation in these paired quality control samples was 3.2 %. Plasma levels of total and high-density lipoprotein (HDL) cholesterol were measured in the laboratory of Dr. Frank Sacks.22 Statistical analysis We compared mean values and proportions of various cerebrovascular risk factors between cases and controls. For matched data a paired Student's t test was used for the continuous variables and a McNemar paired Chi2 test for proportions, whereas for unmatched data the unpaired Student's t test and a Pearson Chi2 test were used. Odds ratios (OR) and 95% confidence intervals (CI) were calculated for matched and unmatched data, using logistic regression. Conditional logistic regression (matched data) and multiple logistic regression (unmatched data) were used to control for several cerebrovascular risk factors, simultaneously. Levels of 90 Homocysteine and Ischemic Stroke tHcy above the 80th percentile among the control subjects were a priori defined as elevated levels, and those below the 80th percentile as normal. Use of tHcy concentrations above the 95th percentile as defining abnormal, as we did in our previous study3, would have led to unstable results in the matched analysis, due to small numbers. However, for the unmatched data we show ORs for elevated tHcy concentrations based on the 95th percentile cut-off point, to facilitate comparison with our earlier work. All P-values are two-tailed. Associations between tHcy levels and several risk factors for ischemic stroke were evaluated by calculating Spearman rank correlation coefficients. To evaluate a possible modifying role of high blood pressure, we performed separate risk analyses in strata of hypertensive and normotensive men, using unmatched data. High blood pressure was considered to be present in subjects with systolic blood pressure higher than 150 mm Hg, diastolic blood pressure higher than 90 mm Hg, or current use of anti-hypertensive drugs. This information was obtained through the enrollment questionnaire. Since elevated tHcy levels mainly have been reported in subjects with premature vascular disease, we performed additional risk analyses in subgroups of men 60 years or younger and men older than 60 years at the start of the study. To study whether the time of follow-up affected the association, we assessed the effect of elevated tHcy levels among men stratified by the median time since enrollment in the trial, based on time of diagnosis for cases. Results Table 6-1 shows the risk factors of 109 men with ischemic stroke and 427 controls. Almost half of the cases had high blood pressure at enrollment, compared to one fifth of the control subjects. The proportion with history of diabetes was also higher among cases. The cases had a higher mean blood pressure and Quetelet's index, as well as lower total and HDL cholesterol levels. The total cholesterol/HDL cholesterol ratio and alcohol consumption were slightly, but not significantly, higher in cases than in controls. Controls were significantly younger than cases, due to the fact that the majority of the control subjects had originally been matched to men with myocardial infarction, whose mean age was lower than that of men with ischemic stroke. Cigarette smoking status, a matching variable, was identical among cases and controls. We also compared the same risk factors among the 97 pairs of cases of ischemic stroke and matched controls. The mean levels and proportions of risk factors were similar to the ones observed in the unmatched groups. However, mean 91 Chapter 6 age and smoking habits did not differ, since these were matching factors (data not shown). Table 6-1. Potential risk factors among patients with ischemic stroke and control subjects in the Physicians' Health Study Patients Control P-value* (n=109) subjects (n=427) case-control difference Mean SD Mean SD Age (years) 61.6 9.1 59.2 8.9 MV+ Quetelet's index (kg/m2) 26.0 - 3.7 24.9 2.8 0.003 Systolic BP (mm Hg) 137 13 128 12 < 0.0001 Diastolic BP (mm Hg) 83 7 79 7 < 0.0001 Alcohol consumption (drinks/day) 0.56 0.59 0.54 0.47 0.81 Total cholesterol (mmol/L) 206 37 214 36 0.04 HDL cholesterol (mmol/L) 45.6 12.9 49.5 13.1 0.006 Ratio of total to HDL cholesterol 4.8 1.4 4.6 1.4 0.13 With characteristic (%): Cigarette smoking habits Never 39.5 40.8 MV+ Past 38.5 41.8 Current 22.0 17.4 History of diabetes 12.8 3.0 < 0.0001 High blood pressure 46.8 19.2 < 0.0001 HDL, high density lipoprotein; BP, blood pressure; MV, matching variable. * Unpaired t test for continuous variables, and chi-square test for proportions. + Differences in age and cigarette smoking between patients and control subjects were heavily influenced by matching for those variables, so tests for differences were not appropriate. Overall the mean level of tHcy was slightly, although not statistically significant, higher among cases with ischemic stroke (11.1 ± 4.0 umol/L) than for controls (10.6 ± 3.4 umol/L; JP=0.12). The mean tHcy level for the matched cases was virtually the same as for the matched controls (Table 6-2). The crude OR of ischemic stroke, comparing individuals with elevated levels (higher than 80th percentile) with those with normal levels was 1.4 (95% CI, 0.8- 92 Homocysteine and Ischemic Stroke 2.2). After adjustment for age, cigarette smoking habits, diabetes, high blood pressure, Quetelet's index, aspirin assignment, total cholesterol/HDL cholesterol ratio, and time since the last meal before the blood was drawn, the OR was 1.2 (95% CI, 0.7-2.0) (Table 6-2). Five cases (5%) had plasma levels higher than the 95th percentile of the control distribution (16.6 umol/L), compared with 21 (5%) control subjects. Contrasting the top 5% with the lower 90% of the control distribution, we observed crude and multivariate adjusted ORs of 1.0 (95% CI, 0.4- 2.7) and 0.8 (95% CI, 0.3-2.4), respectively (data not shown). In a conditional logistic regression analysis in 97 cases of ischemic stroke and 97 controls, matched for age and smoking status, both the crude and multivariate ORs revealed no effect of elevated tHcy (Table 6-2). Table 6-2. Levels of total homocysteine and risk of ischemic stroke Mean ± SD Cutoff (umol/L) P* points4" elevated Patients Controls OR* tHcy (n) (n) (95% CI) (umol/L) Unmatched analysis Patients (n=109) 11.1 ±4.0 > 12.7 28 86 1.2 0.12 Controls (n=427) 10.6 ± 3.4 < 12.7 81 341 (0.7-2.0) Matched analysis Patients (n=97) 11.1 ±4.2 > 12.7 24 24 1.1 0.86 Controls (n=97) 11.0 + 4.8 < 12.7 73 73 (0.5-2.8) tHcy, total homocysteine; OR, odds ratio; CI, confidence interval. * Using an unpaired t test and a paired t test for unmatched and matched data, respectively. * 80th percentile among 427 unmatched controls. * Adjusted for age, cigarette smoking status (either as matching factors or in multivariate logistic model), history of diabetes, high blood pressure, Quetelet's index, aspirin assignment, ratio of total cholesterol and HDL cholesterol, and hours since last meal. We evaluated the associations of plasma tHcy levels with several possible risk factors for ischemic stroke and other factors in controls (Table 6-3). The observed correlations were all small, although some of them were statistically significant. Blood pressure was directly associated with tHcy levels, as were alcohol 93 Chapter 6 consumption and total cholesterol level. Plasma tHcy levels correlated inversely, although not strongly, with multivitamin use (^=-0.17;P=0.0005). Most of the blood samples were not drawn from subjects in the fasting state. Among 351 controls we observed a positive association between tHcy and time since the last meal before the blood was drawn (r=0.14; P=0.007). The mean amount of time that had elapsed since the last meal was 5 hours and 52 minutes in cases and 5 hours and 21 minutes in controls (P=0.45) (data not shown). Table 6-3. Spearman correlation coefficients between plasma total homocysteine and risk factors in control subjects Correlation P-value Risk factor Age 0.01 0.77 Quetelet's index 0.09 0.08 Systolic blood pressure 0.12 0.02 Diastolic blood pressure 0.18 0.0005 High blood pressure (yes versus no) 0.18 0.0002 History of diabetes (yes versus no) 0.05 0.31 Alcohol consumption 0.12 0.02 Multivitamin use -0.17 0.0005 Total cholesterol 0.10 0.03 HDL cholesterol -0.06 0.25 Ratio of total to HDL cholesterol 0.11 0.02 HDL, high density lipoprotein. The numbers for the correlations ranged from 383 to 427 because not all subjects had all measurements. To determine whether an association between tHcy and risk of ischemic stroke was dependent on blood pressure, an interaction term was added to the multivariate model. The likelihood ratio test for the difference between the model with the interaction term and the model without it, showed an improvement of the fit of the model when the interaction term was included (P=0.02). We therefore separately analyzed subjects with normal and high blood pressure (Table 6-4). Among control 94 Homocysteine and Ischemic Stroke subjects the mean tHcy level was significantly higher in hypertensives than in normotensives, whereas among cases there was no such difference. Separate risk analyses for strata of hypertensive and normotensive men revealed that the impact of elevated tHcy level was limited to normotensive subjects only. In this group the multivariate adjusted OR was 2.0 (95% CI, 1.0-4.0) for tHcy concentrations above 12.7 umol/L. The corresponding OR in subjects with high blood pressure was 0.6 (95% CI, 0.3-1.5). Table 6-4. Levels of total homocysteine and risk of ischemic stroke in strata of normotensive and hypertensive subjects Odds ratios for elevated concentrations of tHcy* Normotensive subjects Hypertensive subjects < 12.7 > 12.7 < 12.7 > 12.7 umol/L umol/L umol/L umol/L tHcy level in patients* 11.1 ±4.4 11.3 ±3.5 tHcy level in controls* 10.3 ± 3.3 11.8 ± 3.6 No. of cases/controls 42/288 16/57 39/53 12/29 Crude OR 1 1.9 1 0.6 Adjusted OR* (95% CI) 1 2.0 (1.0-4.0) 1 0.6 (0.3-1.5) tHcy, total homocysteine; OR, odds ratio, CI, confidence interval. * Based on the tHcy distribution among 427 unmatched controls. + Mean ± SD (umol/L). * Adjusted for age, cigarette smoking status, history of diabetes, Quetelet's index, aspirin assignment, ratio of total cholesterol and HDL cholesterol, and hours since last meal. In an analysis restricted to men 60 years or younger at the start of the study (46 cases, 234 controls), the multivariate adjusted OR was 1.7 (95% CI, 0.7-3.8) for subjects with elevated tHcy levels as compared to those with normal values. In men older than 60 years (63 cases, 193 controls) the corresponding OR was 0.8 (95% CI, 0.4-1.8) (data not shown). We further assessed the effect of elevated tHcy levels, stratifying men by the median time since enrollment in the trial. The multivariate OR for elevated tHcy 95 Chapter 6 concentrations was 1.0 (96% CI, 0.5-2.4) when the analyses were restricted to cases diagnosed before the median time. It was 1.3 (95% CI, 0.6-2.8) when only cases diagnosed after that time were included in the analyses (data not shown). Discussion In this prospective, nested case-control study, we observed a 20% higher risk of ischemic stroke in men with tHcy values above the 80th percentile of the control distribution, compared to those with lower values. This association was much weaker than found in many previous studies on tHcy and cerebrovascular disease. More convincing support for a potential role of tHcy were the observed 100% and 70% increases in risk among normotensive men and men 60 years or younger, respectively. However, the findings in subgroups might be due to play of chance, especially considering the small number of subjects. We considered several issues of internal validity that might explain why we observed a weaker association between tHcy and occlusive cerebrovascular disease than most other studies did. Unlike previous investigations, our study employed a prospective design, reducing the possibility that tHcy levels were altered by the disease, medication or changes in lifestyle associated with the disease. However, in a recent study Malinow et al.15 studied the association between tHcy and thickening of the carotid artery intimal-medial wall, measured in asymptomatic individuals, a design less prone to bias than retrospective studies with symptomatic disease as an endpoint. The OR for having a thickened carotid wall was 3.15 (P=0.001) for subjects in the top quintile of plasma tHcy level (>10.5 umol/L) compared to those in the bottom quintile. Although the mean storage period in our study was about 8 years, since samples of cases and controls were stored for the same amount of time, and were handled together and identically throughout processing, we do not consider this an important source of bias. Moreover, the mean value observed in controls was similar to those in studies using fresh samples. Most of our subjects were not fasting when the blood was drawn. It could be possible that fasting levels reveal abnormalities in tHcy metabolism better, and thus use of non-fasting levels might have blurred the association. Furthermore, among controls we observed that subjects, who had their last meal longer ago, tended to have higher plasma concentrations of tHcy, a similar finding to that of Ubbink et al.23 However, since cases had their blood drawn slightly longer after the last meal than controls, this could only have caused their tHcy concentrations to be higher, and therefore this cannot have been responsible for the observed weak association. Finally, we 96 Homocysteine and Ischemic Stroke evaluated the likelihood of aspirin assignment being partly responsible for the observed weaker association. For subjects assigned to placebo the mean tHcy level was higher in cases than in controls, with a difference of 1.1 umol/L, as compared to virtually no difference in the aspirin group. The multivariate adjusted OR for subjects in the upper 20% compared with those in the bottom 80% of tHcy levels was 1.3 (95% CI, 0.6-2.8) in the placebo group and 1.22 (95% CI, 0.8-1.6) in the aspirin group. In our previous study of tHcy and myocardial infarction in the same population we showed a significant three-fold relative risk for men with abnormally high tHcy levels.3 Blood samples of MI cases and cases of ischemic stroke were treated in the exact same way and were sent and analyzed at the same time under the same conditions. However, in the present study the number of ischemic strokes is modest, and the 95% CI therefore rather wide. In contrast to our findings, several studies have reported significantly higher fasting tHcy levels in patients with cerebrovascular disease than in control subjects. A study of Araki et al.6 showed higher mean fasting levels of both free and total homocysteine among 45 patients with cerebral infarction than in 45 normotensive and 45 hypertensive control subjects of similar age and sex. Coull et al.12 reported a significantly higher mean fasting tHcy concentration in 41 patients with acute stroke than in 31 controls. Recently, Brattstrom et al.10 detected fasting hyperhomocysteinemia in 40% of 142 men and women with stroke, but only in four of 66 control subjects. In that study the cutoff-point for hyperhomocysteinemia among men was 17.7 umol/L. These results were consistent with previous findings of the same group of authors,8 but inconsistent with those of another of their studies, in which no significant difference in total fasting plasma tHcy was found between 17 patients with premature cerebral thrombosis and 46 controls.9 Mereau- Richard et al.13 found significantly higher fasting tHcy levels in 92 patients with cerebral vascular disease before the age of 50 years, compared with those in 25 controls. Malinow et al.15 recently reported that fasting plasma levels of tHcy were significantly higher in 287 subjects with thickened intimal-medial carotid walls than in 287 matched controls. Several studies have reported data on plasma levels of tHcy after methionine loading in relation to cerebrovascular risk.7'8,9,11 All studies, except one,9 showed a significantly higher proportion of subjects with hyperhomocysteinemia (usually defined as exceeding the mean post-load tHcy level of controls by more than 2 SD) among cases with cerebrovascular disease than among controls. In our study the mean age of cases at the time they were diagnosed with ischemic stroke was higher than in most of the other studies we mentioned above. 97 Chapter 6 tHcy levels have often been studied in patients with premature cerebrovascular disease, in most studies defined as first diagnosed before 50 or 55 years of age 7,8,9,1 i,i3 j-[owever) several studies found a positive association between tHcy and vascular disease in subjects of the same age as in our study6,10'12 or even in older subjects.10 We observed a stronger association between plasma tHcy and risk of ischemic stroke among younger than among older men. Considering that high blood pressure is a strong risk factor for ischemic stroke, one might expect to observe the highest relative risk associated with elevated levels of tHcy in normotensive men. Indeed, we observed a stronger relationship between tHcy and risk of ischemic stroke in the group of normotensive subjects. Findings of Malinow et al.,15 who reported that the association between plasma tHcy levels and risk for having a thickened carotid arterial wall was stronger among subjects with hypertension, are in contrast with our finding. The association between high blood pressure and tHcy levels that we observed in control subjects, confirms previous observations (Levenson et al., personal communication). We considered the fact that use of anti-hypertensive drugs could have been responsible for the raised plasma tHcy levels in hypertensives. However, tHcy levels in 55 (67%) of the 82 control subjects with high blood pressure who were currently using anti-hypertensive drugs were not different from those in 26 subjects who had never used these drugs (1 subject had used anti-hypertensive drugs in the past). This finding is in concordance with that of Araki et al.,6 who found that significantly higher tHcy levels in 45 normotensive as compared to 45 hypertensive subjects could not be attributed to use of anti-hypertensive drugs. Many mechanisms for the observed positive association between tHcy and vascular disease have been postulated. It has been suggested that the oxidation of homocysteine may result in formation of free radicals and hydrogen peroxide, promoting oxidation of low-density lipoprotein cholesterol24,25 and damaging of endothelial cells.26 Furthermore, findings in several in vitro studies suggest that homocysteine and its derivatives can affect blood coagulation factors, by increasing platelet thromboxane production,27 platelet aggregation,28 and factor V activity,29 or by decreasing protein C activation.30 Both the suggested atherogenic- and thrombogenic effects of homocysteine, its derivatives and related disulphides could account for a positive association between tHcy and risk of ischemic stroke. Increased plasma tHcy levels might be the consequence of low plasma levels of vitamins B6, Bl2, and folate, since these are important cofactors in homocysteine metabolism.31 We observed significant inverse associations for tHcy levels with 3 intakes of vitamin B6, B12 and folate. However, these were themselves highly 98 Homocysteine and Ischemic Stroke intercorrelated due to common sources of vitamin supplements and fortified foods. The correlations were most likely underestimates because the data on dietary intake were quite limited and incomplete. In the same population, plasma levels of vitamin B6 and folate were highly inversely correlated with tHcy levels (Chasan-Taber et al., submitted for publication). The nutritional state of physicians is presumably better than in the general population, which might explain why the difference between tHcy levels of cases and controls was smaller than in other studies. In a recent cross-sectional study of Selhub et al.32 among elderly men and women, the prevalence of high tHcy levels (>14 umol/L) was 29%, of which 67% could be attributed to inadequate plasma concentrations of one or more of the B vitamins involved in homocysteine metabolism. In summary, the present study adds only weak evidence to the hypothesis that elevated tHcy levels are an independent risk factor for ischemic stroke. In our population, the association seemed to be more pronounced among men who are at a low risk of having cerebrovascular disease, i.e. young and without high blood pressure. We suggest that further prospective studies of tHcy and risk of cerebrovascular disease be conducted, including in particular such persons at lower risk. Furthermore, research is needed to define which mechanisms can account for the observed positive relationship between tHcy levels and blood pressure. Sufficient data have accumulated to warrant clinical trials to evaluate the effect of lowering tHcy levels for the primary and secondary prevention of vascular diseases. Acknowledgements This study was supported by research grants (CA42182, HL26490, HL34595, CA34944, CA40360, and RR00163-31) from the National Institutes of Health, Bethesda, Md. The authors thank the Physicians' Health Study Steering Committee (C. Belanger, MA, J.E. Buring, ScD, N. Cook, ScD, K. Eberlein, MPH, S.Z. Goldhaber, MD., D. Gordon, MA, C.H. Hermekens, MD [chair], S. Mayrent, PhD, R. Peto, MD, B. Rosner, PhD, M.J. Stampfer, MD, F. LaMotte, MPH, W.C. Willett, MD; T.Balzkowski, PhD, and A. Vargosko, PhD - ex officio); the End Points Committee (H. Funkenstein, MD [deceased], S.Z. Goldhaber, MD, M.J. Stampfer, MD, and J.O. Taylor, MD [chair]); Stephanie Bechtel, Mary Ann O'Hanesian, Kim Eberlein, MPH, Mary Lou Lyons, RN, MSN, and Georgina Friedenberg, MPH, provided skilled assistance; Fran Grodstein, PhD, gave expert advice and guidance. 99 Chapter 6 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. Mudd SH, Levy HL, Skovby F. Disorders of transuffuration. In: Scriver CS, Beaudet AL, Sly WL, Valle D, eds. The Metabolic Basis of Inherited Disease. 6th ed. New York, NY:McGraw-Hill International Book Co; 1989:693-734. 3. Stampfer MJ, Malinow MR, Willett WC, et al. A prospective study of plasma homocysteine and risk of myocardial infarction in US physicians. JAMA 1992;268:877-81. 4. Malinow MR. Hyperhomocyst(e)inemia: a common and easily reversible risk factor for occlusive atherosclerosis. Circulation 1990;81:2004-6. 5. Robinson K, Clarke R, Graham I. Hyperhomocysteinaemia and vascular disease. J Irish Coll Phys Surg 1991;20:25-9. 6. Araki A, Sako Y, Fukushima Y, Matsumoto M, Asada T, Kita T. Plasma sulphydryl- containing amino acids in patients with cerebral infarction and in hypertensive patients. Atherosclerosis 1989;79:139-46. 7. Boers GHJ, Smals AGH, Trijbels FJM, et al. Heterozygosity for homocystinuria in premature peripheral and cerebral occlusive arterial disease. N Engl J Med 1985;313:709-15. 8. Brattstrom LE, Hardebo JE, Hultberg BL. Moderate homocysteinemia - a possible risk factor for arteriosclerotic cerebrovascular disease. Stroke 1984;15:1012-6. 9. Brattstrom L, Israelsson B, Norrving B, et al. Impaired homocysteine metabolism in early- onset cerebral and peripheral occlusive arterial disease. Effects of pyridoxine and folic acid treatment. Atherosclerosis 1990;81:51-60. 10. Brattstrom L, Lindgren A, Israelsson B, Malinow MR, Norrving B, Upson B. Hyperhomocysteinaemia in stroke: prevalence, cause, and relationships to type of stroke and stroke risk factors. Eur J Clin Invest 1992;22:214-21. 11. Clarke R, Daly L, Robinson K, et al. Hyperhomocysteinemia: an independent risk factor for vascular disease. New Engl J Med 1991;324:1149-55. 12. Coull BM, Malinow MR, Beamer N, Sexton G, Nordt F, De Garmo P. Elevated plasma homocysteine concentration as a possible independent risk factor for stroke. Stroke 1990;21:572-6. 13. Mereau-Richard C, Muller JP, Faivre E, Ardouin P, Rousseaux J. Total plasma homocysteine determination in subjects with premature cerebral vascular disease. Clin Chem 1991;37:126. 100 Homocysteine and Ischemic Stroke 14. Malinow MR, Kang SS, Taylor LM, et al. Prevalence of hyperhomocyst(e)inemia in patients with peripheral arterial occlusive disease. Circulation 1989;79:1180-8. 15. Malinow MR, Nieto FJ, Szklo M, Chambless LE, Bond G. Carotid artery intimal-medial wall thickening and plasma homocysteine in asymptomatic adults. The Atherosclerosis Risk in Communities Study. Circulation 1993;87:1107-13. 16. Stampfer MJ, Buring JE, Willett W, Rosner B, Eberlein K, Hennekens CH. The 2x2 factorial design: its application to a randomized trial of aspirin and carotene in US physicians. Stat Med 1985;4:111-6. 17. Steering Committee of the Physicians' Health Study Research Group. Final report on the aspirin component of the ongoing Physicians' Health Study. N Engl J Med 1989;321:129-35. 18. Walker AE, Robins M, Weinfeld FD. The National Survey of Stroke: clinical findings. Stroke 1981;12(supplement I):I13-144. 19. Smolin LA, Schneider JA. Measurement of total plasma cysteamine using high- performance liquid chromatography with electrochemical detection. Anal Biochem 1988;168:374-9. 20. Malinow MR, Kang SS, Taylor LM, et al. Prevalence of hyperhomocyst(e)inemia in patients with peripheral arterial occlusive disease. Circulation 1989;79:1180-8. 21. Malinow MR, Sexton G, Averbuch M, Grossman M, Wilson D, Upson B. Homocysteine in daily practice: levels in coronary artery disease. Coronary Artery Dis 1990;1:215-20. 22. Stampfer MJ, Sacks FM, Salvini S, Willett WC, Hennekens CH. A prospective study of cholesterol, apolipoproteins, and the risk of myocardial infarction. N Engl J Med 1991;325:373-81. 23. Ubbink JB, Vermaak WJH, Van der Merwe A, Becker PJ. The effect of blood sample aging and food consumption on plasma total homocysteine levels. Clin Chim Acta 1992;207:119-28. 24. Parthasarathy S. Oxidation of low density lipoproteins by thiol compounds leads to its recognition by the acetyl LDL receptor. Biochim Biophys Acta 1987;917:337-340. 25. Heinecke JW, Rosen H, Suzuki LA, Chait A. The role of sulfur-containing amino acids in superoxide production and modification of low density lipoprotein by arterial smooth muscle cells. J Biol Chem 1987;262:10098-103. 26. Starkebaum G, Harlan JM. Endothelial cell injury due to copper-catalyzed hydrogen peroxide generation from homocysteine. J Clin Invest 1986;77:1370-6. 101 Chapter 6 27. Graeber JE, Slott JH, Ulane RE, Schulman JD, Stuart MJ. Effect of homocysteine and homocystine on platelet and vascular arachidonic acid metabolism. Pediatr Res 1982;16:490-3. 28. McCully KS, Carvalho AC. Homocysteine thiolactone, N-homocysteine thiolactonyl retinamide, and platelet aggregation. Res Commun Chem Pathol Pharmacol 1987;56:349-60. 29. Rodgers GM, Kane WH. Activation of endogenous factor V by homocysteine-induced vascular endothelial cell activator. J Clin Invest 1986;77:1909-16. 30. Rodgers GM, Conn MT. Homocysteine, an atherogenic stimulus, reduces protein C activation by arterial and venous endothelial cells. Blood 1990;75:895-901. 31. Ubbink JB, Vermaak WJH, Van der Merwe A, Becker PJ. Vitamin B-12, vitamin B-6, and folate nutritional status in men with hyperhomocysteinemia. Am J Clin Nutr 1993;57:47-53. 32. Seihub J, Jacques PF, Wilson PWF, Rush D, Rosenberg IH. Vitamin status and intake as primary determinants of homocysteinemia in an elderly population. JAMA 1993;270:2693-8 102 Homocysteine and Cardiovascular Disease: 7 Gender Differences and Effect of Menopause The European Concerted Action Project Abstract Background Elevated levels of plasma total homocysteine (tHcy) are associated with increased risk of vascular disease. Levels of tHcy are generally higher in men than in women, but after menopause levels may rise, possibly exceeding levels of men. It has been suggested that this may in part account for the marked increased risk of vascular disease in women after menopause. Methods A multi-center case-control study of 750 cases of vascular disease and 800 controls, of both sexes and under age of 60 years, was conducted. Plasma tHcy levels, fasting and after a methionine loading test, were measured. Results Men had higher geometric mean levels of fasting (15% [95% CI, 10% to 20%]) and post-load tHcy (8% [95% CI, 4% to 13%]) than women, after adjustment for age and center. However, expressed per dose of administered methionine, post-load tHcy was significandy lower in men (-15% [95% CI, -20% to -10%). There was a graded relationship between tHcy and risk of vascular disease. In men, adjusting for age, center, and conventional risk factors, the relative risks per 1 SD increase in tHcy were 1.5 (95% CI, 1.2 to 1.8) and 1.4 (95% CI, 1.2 to 1.7) for fasting and post-load tHcy, respectively. In women, these risks were 1.7 (95% CI, 1.1 to 2.6) and 1.8 (95% CI, 1.4 to 2.4). Elevation of tHcy was a strong risk factor for pre- and postmenopausal women. Levels of tHcy showed a positive association with age, for both sexes. After menopause, levels of post- load tHcy in women surpassed levels of men. Conclusions In this population, elevation of tHcy was an independent, graded risk factor for cardiovascular disease. Especially when measured after methionine loading, tHcy showed a stronger association with vascular disease in women than in men. There were indications that post-load tHcy may increase after menopause. However, the lower tHcy levels in premenopausal women, in comparison with men or postmenopausal women, do not protect them from developing vascular disease. Petra Verhoef, Ray Meleady, Leslie Daly, Ian Graham for the EC Concerted Action Project: Hyperhomocysteinemia and Vascular Disease.* Names of all persons that collaborated in this project are listed at the end of the thesis (appendix 1), including acknowledged persons and institutions. This chapter is based on a draft, which has not yet been approved by other project members. 103 Chapter 7 Introduction Epidemiologic studies have almost unanimously shown a positive association between elevations of plasma total homocysteine (tHcy) and risk of vascular disease.1,2 In most studies, women formed only a small part of the study population (or were not included), and estimations of vascular disease risk are mostly given for the sexes combined. A recent meta-analysis2 and a case-control study3 have shown a stronger effect of elevation of fasting tHcy on vascular disease risk in women than in men. Generally, tHcy levels are lower in women than in men, both in the fasting state and after methionine loading.4,5,6,7 One study,8 however, found tHcy levels after methionine loading not to differ between the sexes. Some studies,3,9,10,11 but not all,8 have reported higher tHcy levels in postmenopausal women than in premenopausal women, which may even exceed those in men of the same age. It has been suggested that this may in part account for the marked increased risk of vascular disease in women after menopause. Differences between the sexes and between pre- and postmenopausal women are possibly related to a tHcy-lowering effect of estrogens, as has been suggested by some studies,12,13,14 but not by others.15,16 Furthermore, the higher tHcy levels in men may be related to their larger muscle mass, since about 75% of tHcy is formed in conjunction with creatine-creatinine synthesis.17 We have conducted a large multi-center case-control study of 750 cases of vascular disease and 800 controls, of both sexes. In the present analyses, we had the following two objectives: 1. To compare tHcy levels of women and men, and elucidate whether in women there is evidence of increase in tHcy after menopause, relative to the possible age-related changes in men. 2. To compare associations of tHcy with vascular disease in women and men, and in premenopausal and postmenopausal women. Cofactors of homocysteine metabolism (pyridoxal 5'- phosphate [PLP], vitamin B12, and folate), and creatinine, which are important determinants of circulating tHcy levels, were studied as well. Methods Study population Investigators from 19 centers in 11 European countries recruited 1550 study participants (750 patients of vascular disease, 800 controls) of both sexes and under 60 years of age. For cases, both clinical and objective evidence of vascular disease were prerequisites. Centers tried to enroll new or recently diagnosed cases as much 104 Homocysteine, CVD, Gender and Menopause as possible. Control subjects were free of overt vascular disease. One half of controls came from industrial employee registers, one third from random population samples, and the remainder were hospital-based. Important exclusion criteria for all subjects were: evidence of non-afherosclerotic vascular disease, diabetes mellitus, renal, hepatic, and thyroid diseases, cardiomyopathy, pregnancy, anti-convulsant medication, and recent (< 3 months) major systemic illness, including myocardial infarction. Cases and controls were frequency matched for sex and age. Study measurements Information was obtained on demographic variables, smoking, family history of vascular disease, weight, height, drug and vitamin usage. Duplicate blood pressure measurements were taken before and after methionine loading, after 5 minutes rest with the subject seated. Women provided information on date of last menstruation, use of oral contraceptives and hormone replacement therapy. We defined women with a normal menstruation and not on hormone replacement therapy as premenopausal. Postmenopausal women were those who had not menstruated in the past two years. For women with missing date of last menstruation, we defined those who were younger than 45, and not on hormone replacement therapy, as premenopausal, and those older than 55, not using oral contraceptives, as postmenopausal. Women older than 45, with a missing date, but using oral contraceptives were defined as premenopausal. All other women were excluded (25 controls, 35 cases) for the analyses by strata of pre- and postmenopausal women. Blood samples for tHcy determination were taken fasting and six hours after administration of L-mefhionine (0.1 g/kg body weight), using a standardized method.18 Samples were placed immediately on ice, protected from light, centrifuged within one hour and stored at -70° C. Measurements of tHcy were performed centrally in Bergen, Norway, using an automated method with HPLC separation and fluorescence detection.19 Assays of serum lipids, serum creatinine, and levels of cofactors of homocysteine metabolism (serum vitamin B12, blood folate, plasma PLP) were performed centrally by MEVIELAB-AB in Soraker, Sweden. Vitamin B12 and folate estimations were by radioimmunoassay and PLP by enzymatic photometry with HPLC separation. Blood folate was expressed per haematocrit to obtain values for erythrocyte folate. Current smoking was defined as the use of any tobacco at the time of vascular diagnosis (cases) or at the day of methionine loading (controls). Subjects were diagnosed as hypertensive at a systolic blood pressure > 160 mmHg or 105 Chapter 7 diastolic blood pressure > 95 mmHg (measured at day of methionirie loading), or when they were using anti-hypertensive medication. Hypercholesterolemia was defined as serum cholesterol > 6.5 mmol/L or use of cholesterol-lowering drugs. Statistical methods In the present analyses, we pooled all vascular disease categories (coronary heart disease, cerebrovascular or peripheral vascular disease). Except for Table 7-1, all analyses in the paper were controlled for age and center. Five center strata were used, some consisting of one large center, and others of groups of small centers. Grouping was based on a ranking of mean fasting fHcy. Variables that showed a positive skewness (fHcy, vitamins) and creatinine were log-transformed. We show means plus standard deviations (SDs) and geometric means. In Table 7-1, case- control differences have been tested with Student's unpaired t test or Pearson's chi- square test. In all other comparisons, we calculated percentage differences based on the log-transformed variables. Using analysis of variance, we controlled for effects of age and center (and vitamins and creatinine in some comparisons). We looked at gender differences by comparing the groups of male and female controls. By means of logistic regression analysis, odds ratios (OR, as an estimate of relative risk [RR]) of vascular disease were calculated with 95% confidence intervals. We simultaneously adjusted for age, center, and conventional risk factors of vascular disease by use of multivariate logistic regression analysis. ORs were calculated for those with elevated fasting tHcy (defined as > 12.0 umol/L, the 80th percentile of controls) and for those with elevated post-load tHcy levels (defined as > 38.0 umol/L, 80th percentile of controls). Subjects with levels at or below the 80th percentile were considered as the reference group. To evaluate a possible graded association of plasma tHcy with vascular disease, we also computed the ORs per 1 SD increase of plasma tHcy. We wanted to study whether in women there was evidence of an increase in tHcy after menopause, as compared to a possible age-related change in men. For that purpose we used linear regression analysis to model the relationship between log-transformed tHcy (dependent variable), and age (both linear and quadratic terms), gender, menopausal status, and the interaction between age and menopausal status (independent variables). Menopausal status was set to zero for men. Differences in age and sex between center (5 strata) were accounted for by including four dummy variables in the model. We used data of control subjects only. Gender-specific expected tHcy values were calculated and pictured in a graph as a function of age, allowing tHcy of postmenopausal women to increase faster 106 Homocysteine, CVD, Gender and Menopause than that of premenopausal women. In additional analyses, we included vitamins and creatinine as independent variables as well, to adjust for possible age-related differences in these factors. For all tests a two-tailed 5% level of statistical significance was taken. Results Characteristics and vascular disease risk factors in women Table 7-1 shows results of 230 female control subjects and 206 female cases of cardiovascular disease. Cases were older, had higher mean total cholesterol levels, a lower ratio of HDL to total cholesterol, higher mean triglyceride level, higher mean blood pressure, and a higher mean body mass index. Concordantly, hypercholesterolemia and hypertension were more prevalent among cases. Furthermore, there were more current smokers among cases. Use of oral contraceptives, other hormone drugs', and vitamin supplements was less common among cases. For men, risk factors showed similar case-control differences. Table 7-1. Characteristics of female cases of cardiovascular disease and controls Controls All cases P* (n=230) (n=206) Age (years) 40.6 ± 10.4 45.2+ 9.7 <0.0001 Body mass index (kg/m2) 23.5 ± 3.6 25.4+ 4.8 0.0001 Serum cholesterol (mmol/L) 5.7 ± 1.3 6.3 + 1.4 <0.0001 Serum HDL/total cholesterol 0.29 + 0.11 0.22 + 0.08 0.0001 Serum triglycerides (mmol/L) 1.08 ± 0.59 1.50 + 0.79 0.0001 n(%) hypercholesterolemic 58 (25.2) 99 (48.1) <0.0001 Systolic blood pressure (mmHg) 121.5 + 13.5 130.4 + 21.9 0.0001 Diastolic blood pressure (mmHg) 77.6 + 9.2 81.3 + 10.8 0.0002 n(%) hypertension 14 (6.1) 81 (39.3) <0.0001 n(%) currently smoking any tobacco 61 (26.5) 111 (53.9) <0.0001 n(%) currently smoking cigarettes 59 (25.7) 105 (51.0) <0.0001 n(%) using oral contraceptives 28 (12.2) 14 (6.9) 0.06 n(%) on hormone replacement therapy 8 (3.5) 6 (2.9) 0.74 n(%) using vitamins with B6/BI2/folate 11 (4.8) 6 (2.9) 0.60 * Tested with Student's t test for continuous variables (mean ± SD is shown) and Pearson's chi- square test for frequencies. 107 Chapter 7 Gender differences The fact that in both sexes cases were older than controls, shows that frequency matching on age had not been successful (Table 7-2). In the last column of Table 7-2, we show differences in tHcy levels between male and female controls, adjusting for age and center. Fasting and post-load tHcy levels were higher in men than in women. The difference was larger for fasting than post-load tHcy. However, when expressing post-load tHcy level per kg of body weight (which corresponds with the dose of methionine), women had a significantly higher response than men. Expressed per unit of body mass index, male and female controls had similar levels of post-load tHcy (data not shown). Levels of erythrocyte folate and serum creatinine were significantly higher in men, whereas levels of serum vitamin B,2 and plasma PLP were similar for the sexes. Geometric mean fasting tHcy was 11% (95% CI, 6% to 16%) higher and post-load tHcy 5% (95% CI, 1% to 10%) higher in males than in females, adjusting for differences in levels of creatinine and vitamins between the sexes. In women however, post-load tHcy expressed per kg of body weight was 17% (95% CI, 11% to 23%) greater than in men, after adjustment for vitamins and creatinine (data not shown). Case-control differences in tHcy among men and women For both women and men, cases had significantly higher tHcy levels than controls (Table 7-2). The case-control difference in post-load tHcy (also expressed per kg body weight) was greater in women than in men, whereas case-control differences in fasting tHcy were similar for the sexes. In both women and men, cases had statistically significant lower plasma PLP levels, but similar levels of erythrocyte folate and vitamin B12. In females, the case-control difference in PLP was larger than for men. Menopause-related differences Table 7-3 shows tHcy levels of premenopausal and postmenopausal women. Postmenopausal control women had nonsignificantly higher plasma tHcy levels in the fasting and post-load state (including post-load tHcy expressed per kg body weight) than premenopausal control women. Furthermore, postmenopausal control women had significantly lower plasma PLP levels than premenopausal women. Adjustment for creatinine and vitamins slightly reduced the differences in tHcy levels (data not shown). 108 Table 7-2. Levels of plasma total homocysteine and vitamins among female and male cases of cardiovascular disease and controls FEMALES MALES ¥ Age (years) 40.6 ± 10.4 45.2 ± 9.7 0.0001 45.2 ± 9.8 47.9 ± 7.9 0.0001 0.0001 Serum creatinine**(umol/L) 60.4 59.9 -3 (-7 to 0) 71.6 72.8 1 (-1 to 3) 18 (15 to 21) 61.2 ± 10.2 60.9 ±11.4 72.6 ±11.8 74.5 ± 17.0 Fasting tHcy (umol/L) 8.6 10.2 14 (7 to 22) 10.2 11.7 12 (8 to 17) 15 (10 to 20) 9.0 ± 2.9 11.1 ± 5.9 10.7 ± 4.3 12.7 ± 7.1 Post-load tHcy (umol/L) 28.1 37.0 22 (14 to 30) 31.3 35.0 9 (6 to 13) 8 (4 to 13) 29.5 ± 9.3 40.6 ± 19.6 32.6 ± 10.4 37.1 ± 15.3 Post-load tHcy/body weight 0.45 0.56 16 (8 to 25) 0.40 0.45 11 (7 to 15) -15 (-20 to -10) (umol/L*kg) 0.47 ±0.15 0.63 ± 0.34 0.42 ± 0.14 0.49 ± 0.22 Erythrocyte folate**(nmol/L) 800.5 787.7 -3 (-12 to 6) 857.1 819.4 -6 (-11 to 1) 8 (1 to 15) 1 (-4 to 6) Serum vitamin B12 (pmol/L) 238.5 247.8 3 (-6 to 13) 233.4 232.0 1 (-5 to 8) Plasma PLP (nmol/L) 32.1 25.7 -24 (-33 to -15) 30.7 26.4 -14 (-18 to -9) 0 (-5 to 6) tHcy, plasma total homocysteine; PLP, pyridoxal 5'-phosphate. * Percentage difference (P-value for age difference): cases versus controls or males versus females. Confidence interval estimation is based on log- transformed variables. All differences were adjusted for age and center. However, real (nonadjusted) mean levels are shown. ** For creatinine and tHcy geometric means (upper figure) and mean ± SD are given, for vitamins only geometric means. o Table 7-3. Levels of plasma total homocysteine and vitamins among female cases of cardiovascular disease and controls, stratified by menopausal status PREMENOPAUSAL ? POSTMENOPAUSAL ? Difference* Controls Cases Case-control Controls Cases Case-control pre- and post difference menopausal n=155 n=107 difference* n=52 n=64 % (95% CI) % (95% CI) controls % (95% CI) Age (years) 35.3 ± 7.8 38.7 ± 8.2 0.008 53.3 ± 4.3 53.9 ± 4.1 0.43 - Serum creatinine**(umol/L) 59.2 59.4 -2 (-7 to 2) 63.3 62.0 -2 (-10 to 5) 7 (1 to 13) 60.1 ± 10.2 60.3 ± 10.7 64.2 ± 10.6 63.2 ± 13.3 Fasting tHcy (umol/L) 8.4 9.6 13 (3 to 23) 9.2 10.8 17 (3 to 32) 6 (-4 to 17) 8.8 ± 2.8 10.6 ± 6.1 9.6 ± 2.6 11.6 ±5.5 Post-load tHcy (umol/L) 27.1 35.6 23 (12 to 35) 29.6 39.2 24 (10 to 40) 7 (-3 to 19) 28.4 ± 9.0 39.6 ± 20.5 30.8 ± 9.2 42.0 ± 17.6 Post-load tHcy/body weight 0.44 0.55 18 (7 to 30) 0.47 0.58 14 (0to31) 9 (-2 to 20) (umol/L*kg) 0.46 ± 0.15 0.62 ± 0.33 0.49 ± 0.14 0.64 ± 0.34 Erythrocyte folate**(nmol/L) 796.6 763.4 -3 (-15 to 8) 780.9 799.2 -1 (-20 to 15) -8 (-24 to 7) Serum vitamin BI2 (pmol/L) 237.2 251.3 7 (-5 to 22) 238.2 245.4 -3 (-23 to 14) 13 (-5 to 34) Plasma PLP (nmol/L) 33.3 25.4 -31 (-44 to-19) 29.9 26.8 -9 (-27 to 6) -17 (-32 to -4) tHcy, plasma total homocysteine; PLP, pyridoxal 5'-phosphate. * Percentage difference (P-value for age difference): cases versus controls or post- versus premenopausal females. Confidence interval estimation is based on log-transformed variables. All differences were adjusted for age and center. However, real (nonadjusted) mean levels are shown. ** For creatinine and tHcy geometric means (upper figure) and mean ± SD are given, for vitamins only geometric means. Table 7-4. Relative risks of cardiovascular disease for subjects with elevated plasma tHcy (or per 1 SD increase in tHcy) among males and females (all and stratified by menopausal status) FEMALES Premenopausal Postmenopausal MALES All females females females n=1114 n=436 n=262 n=116 Fasting tHcy OR (95% CI) Elevated* age, centre adjusted** 2.0 (1.6 to 2.7) 2.3 (1.3 to 3.8) 2.0 (1.0 to 3.8) 2.7 (1.1 to 6.5) multivariate adjusted* 1.9 (1.4 to 2.5) 1.9 (1.1 to 3.5) 1.5 (0.7 to 3.2) 2.7 (0.9 to 7.6) Per 1 SD (6 umol/L) increase age, centre adjusted 1.6 (1.3 to 2.0) 2.0 (1.4 to 2.9) 2.0 (1.2 to 3.3) 3.0 (1.2 to 7.3) multivariate adjusted 1.5 (1.2 to 1.8) 1.7 (1.1 to 2.6) 1.5 (0.9 to 2.6) 1.9 (0.7 to 5.0) Post-load tHcy Elevated* age, centre adjusted .9 (1.4 to 2.5) 3.2 (2.0 to 5.1) 2.9 (1.6 to 5.4) 3.3 (1.3 to 8.0) multivariate adjusted .6 (1.2 to 2.2) 2.4 (1.4 to 4.1) 2.3 (1.1 to 4.6) 2.5 (0.9 to 7.3) Per 1 SD (14 umol/L) increase age, centre adjusted .5 (1.3 to 1.8) 2.1 (1.6 to 2.7) 2.1 (1.5 to 3.0) 2.8 (1.5 to 5.3) multivariate adjusted .4 (1.2 to 1.7) 1.8 (1.4 to 2.4) 1.8 (0.9 to 2.7) 2.1 (1.1 to 4.1) * Cutoff-points were 12 umol/L and 38 pmoI/L for fasting elevated and post-load elevated tHcy, respectively. For calculation of relative risks in premenopausal and postmenopausal females we used specific cutoffs for women: 11 pmol/L and 36 pmol/L. ** Using multivariate logistic regression analysis, adjusting for age and center. * Using multivariate logistic regression analysis, adjusting for age, center, smoking, hypertension and hypercholesterolemia. Chapter 7 Figure 7-1. Association of age with fasting and post-load levels of plasma total homocysteine in male and female controls. Ill Homocysteine, CVD, Gender and Menopause Case-control differences in fflcy among pre- and postmenopausal women Case-control differences in tHcy were very similar in pre- and postmenopausal women (Table 7-3). In both groups of women, the differences were strongest for tHcy changes in response to methionine loading. In premenopausal women, vascular disease patients had significantly lower plasma PLP levels than controls. Age-related changes in tHcy For fasting tHcy, age and gender were strongly associated with the tHcy level, but menopausal status or the interaction of age and menopausal status were not significant explanatory variables. Thus, fasting tHcy increased slowly with increasing age in women and men, with no additional effect of menopause (Figure 7-1, upper graph). For post-load tHcy, age, gender, menopause and the interaction term of age and menopause were found to be significantly associated with tHcy levels. In men (and premenopausal women), post-load tHcy appeared to slowly increase in the younger age-groups, stabilizing around age 45 to 50 years. In women however, the model showed evidence of an abrupt elevation in the post• menopausal age-groups (Figure 7-1, lower graph). For post-load tHcy expressed per kg body weight (which was generally higher for women and increased very slowly with age) there was a steep increase in post-menopausal age-groups as well (data not shown). Creatinine levels increased steadily with increasing age in men and women, whereas plasma PLP and serum vitamin B12 showed a decrease with increasing age in both sexes. Erythrocyte folate showed no clear association with age. There was no evidence of changes in vitamins and creatinine in women related to menopause (data not shown). When vitamins and creatinine were added to the linear regression model as independent variables, the effects on tHcy of age, gender, menopause, and the age-menopause interaction remained similar (data not shown). Association of tHcy with risk of vascular disease Adjusting for age, center and conventional risk factors for vascular disease, the RR of vascular disease for subjects with fasting elevated tHcy (defined as levels above the 80th percentile among control subjects) was about two-fold increased relative to subjects with lower levels (Table 7-4). It was similar for both sexes. However, the multivariately adjusted RR of vascular disease associated with post- load tHcy levels above the 80th percentile among controls, was higher in women 113 Chapter 7 than in men. For fasting, but especially for post-load tHcy, the increase in RR for each 1 SD increase in tHcy was higher in women than in men (Table 7-4). Table 7-4 also gives relative risks for elevated tHcy in subgroups of pre- and postmenopausal women. The data suggested that risks were slightly higher in postmenopausal women, relative to premenopausal women. However, the attenuating effects of confounding by smoking, hypertension and hypercholesterolemia were strong. In addition, possibly also because of small numbers, confidence intervals included the null value for most estimates. To study the possibility that the stronger association of especially post-load tHcy with vascular disease in women relative to men, was explained by selection of genetically predisposed female patients in the study population, we looked at family history of vascular disease. Among female cases with post-load tHcy above the 80fh percentile of controls, we found that 40% of them had a first degree relative with vascular disease, as compared to 30% among female cases without elevated post- load tHcy. Among male cases these percentages were 33 and 35, respectively. Thus, among female cases, post-load elevated tHcy may more frequently have had a genetic background. Association of PLP with risk of vascular disease Since the observation of lower levels of PLP among cases than controls was striking, we calculated RRs of vascular disease, associated with low PLP levels (< 20 nmol/L). This was done recently in another epidemiologic study.20 In men, the RR of vascular disease in those with low PLP levels, adjusting for age and center, was 2.3 (95% CI, 1.6 to 3.4). In women, the corresponding RR was 3.7 (95% CI, 2.0 to 7.0). After further adjustment for conventional risk factors the RR associated with low PLP levels was 2.4 (95% CI, 1.6 to 3.5) in men, and 2.7 (95% CI, 1.3 to 5.4) in women. When additionally including tHcy (either fasting or post-load tHcy) in the multivariate logistic regression model, RRs for subjects with low PLP remained virtually the same, indicating that PLP is inversely related to vascular disease, independently of tHcy. When looking at these associations in pre- and postmenopausal women, we observed a larger effect of low PLP on cardiovascular risk in premenopausal women than in postmenopausal women (data not shown). 114 Homocysteine, CVD, Gender and Menopause Discussion Main findings Elevated plasma tHcy was a strong, independent risk factor for vascular disease in both sexes. Especially post-load tHcy, and to a lesser extent fasting tHcy, showed a stronger association with vascular disease in women than in men. Effects were graded for both sexes. Levels of tHcy showed a positive association with age, for both sexes. After menopause, however, levels of post-load tHcy in women surpassed levels of men. Both in pre- and postmenopausal women, plasma tHcy showed a positive association with risk of vascular disease, with a suggestion of higher risk in postmenopausal women. Gender Female controls had lower fasting and post-load tHcy levels than male controls, even when adjusted for differences in age, creatinine and vitamins between the sexes. The observation that fasting tHcy is lower in women than men is confirmed by the majority of studies.3'4,6'8'11,21'22 With one exception,8 post-load tHcy levels have also been found to be higher in men than in women.22 A large cross- sectional study of elderly participants in the Framingham study5 found that the difference between the sexes was largely attributable to differences in vitamin status. The present study and another one21 do not confirm this. In the present study, although women had lower absolute tHcy levels after methionine loading than men, their response per unit dose of methionine (-per kg body weight) was greater, also after adjustment for creatinine and vitamins. Assuming that, for both men and women, absorption of methionine is similar and doses were high enough to saturate the transsulfuration pathway, women's higher response per dose of methionine may be related to reduced ability to either catabolize homocysteine through the transsulfuration pathway or to remethylate homocysteine to methionine. Blom et al." found evidence for a greater ability to catabolize tHcy in women: they had higher serum and urinary concentrations of metabolites of the transamination pathway (an alternative pathway of catabolizing methionine) than men. Of course, this may contribute to keeping tHcy levels low (which may be one of the causes of lower tHcy in women relative to men), but could also indicate that women utilize this alternative pathway to compensate for lower catabolism through the transsulfuration pathway (as suggested by higher post- load response per dose of methionine in our study). Another explanation of higher post-load response per dose of methionine in women may be lower tHcy clearance 115 Chapter 7 relative to men, although a previous study by Boers et al.9 did not find differences in clearance between men and women. An alternative explanation of the higher response per dose of methionine in women, may be that they received an overdose per kg of lean body mass (the metabolic active tissue). It may be advisable in future research to standardize the methionine dose to body mass index rather than to bodyweight, especially when looking at differences between the sexes. Age Overall, both there was a positive association between tHcy and age. Other references have shown conflicting results on age and tHcy, some showing no association, and others showing a positive association.6,22 In general, age-related increase of creatinine (most likely caused by reduced renal function with age) and decreases of PLP, vitamin B12, and folate levels may be partly responsible. In the present study, in line with other findings,21 age remained a strong explanatory variable for plasma tHcy after controlling for the age-related differences in these determinants. Decreased activity of enzymes involved in homocysteine metabolism with ageing, as has been suggested for cystathionine B-synthase23, may be one explanation for this observation. Alternatively, decreasing activity of the enzyme methionine synthase (which remethylates homocysteine to methionine) with increasing age may play a part.24 Menopause In contrast to other studies of fasting tHcy and menopause,3'10'25 our study showed no evidence of an additional menopausal effect in women, on top of the age-related increase. In contrast, in the postmenopausal women, there was a marked increase in post-load tHcy (also when expressed per kg body weight), exceeding levels in men, which was still evident after adjustment for serum vitamin and creatinine levels. This may very likely be related to hormonal changes, as no corresponding age-related changes were observed in men. Our finding is in line with several previous findings,9,10'24,25 but not with another one.8 Biological mechanisms The gender differences in fasting and post-load tHcy can most likely, in part or in whole, be explained by the larger creatine-creatinine synthesis in men.17'21 Adjustment for serum creatinine may not be sufficient to control for this different rate in homocysteine formation entirely. Hormonal differences, especially a tHcy- 116 Homocysteine, CVD, Gender and Menopause reducing effect of estrogens (see below) may be responsible as well, and effects of other, unmeasured factors cannot be excluded. The hypothesized tHcy-lowering effect of estrogens is suggested by the fact that fHcy concentrations have been found to be lower in pregnant women than in nonpregnant women.12'13 Also, tHcy was observed to be lower during the high hormonal (= high estrogen) phase than in the low hormonal phase, in women using oral contraceptives.14 This was not confirmed by others.15'16 A recent study achieved a significant 17% reduction of fasting tHcy with oral replacement therapy in postmenopausal women with initially high fasting tHcy levels. Unfortunately, their study design lacked a control group.26 Another study,27 showed significant decreases in plasma fasting tHcy in postmenopausal women with breast cancer treated with tamoxifen. These observations are consistent with the observation that women on hormone replacement therapy have reduced risk of developing cardiovascular disease.28 Relations of tHcy and PLP with vascular disease Our estimations of RRs per 1 SD (6 umol/L) of fasting tHcy are similar to those estimated by Boushey et al.2 per 5 umol/L in a meta-analysis. As in the present study, they found higher risk in women. The strong association between post-load tHcy and vascular disease among women in the present case-control study, may be explained by a selection of female cases who are genetically predisposed to developing cardiovascular disease at a young age. Thermolability of the enzyme 5,10-methylenetetrahydrofolate reductase, associated with reduced enzyme activity and elevation of tHcy levels (both fasting and after a methionine load) has been described.29 We found a higher percentage of subjects with a positive family history of vascular disease among women with elevated post-load tHcy (> 80th percentile of control subjects) relative to male cases with elevated post-load tHcy, which may support this hypothesis of a selection of genetically-predisposed women. Of course, this may point to other inherited factors, such as lipid metabolism or shared environmental factors. It may also be possible that a synergistic effect between tHcy and other risk factors for cardiovascular disease is stronger in women than in men, as has been suggested by results of the Hordaland homocysteine study for smoking4 and for hypercholesterolemia in a previous report of our European Concerted Action Project.30 RRs of vascular disease for fasting tHcy were somewhat higher in postmenopausal women than in premenopausal women. However, 95% confidence intervals included unity many times for multivariately adjusted RRs. Other studies 117 Chapter 7 have also observed the largest effects of tHcy for postmenopausal women.3'24 The higher risk in postmenopausal women may be related to possibly stronger interaction effects between tHcy and other risk factors. Due to small numbers, we have not studied this hypothesis. It is important to note that plasma PLP was significantly lower in cases compared to controls (except in postmenopausal women). PLP levels lower than 20 nmol/L were associated with an approximate two-fold increased risk, in both sexes. This observation was most evident in premenopausal women. A recently conducted study found subjects with PLP levels below 20 nmol/L to have a four-fold increased risk of coronary artery disease.20 Like in our study, the effect appeared to be largely independent of tHcy. Conclusion The incidence of vascular disease and levels of tHcy are both lower in women compared to men, and lower in premenopausal women than in postmenopausal women. However, our data have shown that it would be erroneous to think that women are protected from vascular disease before menopause, because of lower tHcy levels. Estrogen treatment may only be a way of tHcy lowering in postmenopausal women, when levels are initially high.26 However, tHcy is now generally considered a graded risk factor for vascular disease, also in the low- normal range. Thus, with respect to reduction of cardiovascular disease risk, postmenopausal women may possibly benefit more from B-vitamin supplements, which have been shown to effectively reduce plasma tHcy, over a wide range of values.31,32 Of course, this applies to premenopausal women and men as well. 20 According to our findings, and those of others, vitamin B6 should be included in such supplements. References 1. Arnesen E, Refsum H, B(|)naa KH, Ueland PM, F(()rde OH, Nordrehaug JE. Serum total homocysteine and coronary heart disease. Int J Epidemiol 1995;24:704-9. 2. Boushey CJ, Beresford SAA, Omenn GS, Motulsky AG. A quantitative assessment of plasma homocysteine as a risk factor for vascular disease. Probable benefits of increasing folic acid intakes. JAMA 1995;274:1049-57. 3. Kang SS, Wong PW, Cook HY, Norusis M, Messer JV. Protein-bound homocysteine. A possible risk factor for coronary artery disease. J Clin Invest 1986;77:1482-6. 118 Homocysteine, CVD, Gender and Menopause 4. Nygard O, Vollset SE, Refsum H, et al. Total plasma homocysteine and cardiovascular risk profile: The Hordaland homocysteine study. JAMA 1995;274:1526-33. 5. Selhub J, Jacques PF, Wilson PWF, Rush D, Rosenberg IH. Vitamin status and intake as primary determinants of homocysteinemia in an elderly population. JAMA 1993;270:2693-8. 6. Malinow MR. Homocysteine and arterial occlusive diseases (Frontiers in Medicine). J Intern Med 1994;236:603-17. 7. Jacobsen DW, Gatautis VJ, Green R, et al. Rapid HPLC determination of total homocysteine and other thiols in serum and plasma: sex differences and correlation with cobalamin and folate concentrations in healthy subjects. Clin Chem 1994;40:873-81. 8. Andersson A, Brattstrom L, Israelsson B, Isaksson A, Hamfelt A, Hultberg B. Plasma homocysteine before and after methionine loading with regard to age, gender and menopausal status. Eur J Clin Invest 1992;22:79-87. 9. Boers GH, Smals AG, Trijbels FJ, Leermakers Al, Kloppenborg PW. Unique efficiency of methionine metabolism in premenopausal women may protect against vascular disease in the reproductive years. J Clin Invest 1983;72:1971-6. 10. Brattstrom LE, Hultberg BL, Hardebo JE. Folic acid responsive post-menopausal homocysteinemia. Metabolism 1985;34:1073-7. 11. Blom HJ, Boers GH, Van den Elzen JP, Van Roessel JJ, Trijbels JM, Tangerman A. Differences between premenopausal women and young men in the transamination pathway of methionine catabolism, and the protection against vascular disease. Eur J Clin Invest 1988;18:633-8. 12. Kang SS, Wong PWK, Zhou J, Cook HY. Total homocysteine in plasma and amniotic fluid of pregnant women. Metabolism 1986;35:889-91. 13. Andersson A, Hultberg B, Brattstrom L, Isaksson A. Decreased serum homocysteine in pregnancy. Eur J Clin Chem Clin Biochem 1992;30:377-9. 14. Steegers-Theunissen RPM, Boers GHJ, Steegers EAP, Trijbels FJM, Thomas CMG, Eskes TKAB. Effects of sub-50 oral contraceptives on homocysteine metabolism: a preliminary study. Contraception 1992;45:129-39. 15. Beaumont V, Malinow MR, Sexton G, et al. Hyperhomocyst(e)inemia, anti-estrogen antibodies and other risk factors for thrombosis in women on oral contraceptives. Atherosclerosis 1992;94:147-52. 16. Brattstrom L, Israelsson B, Olsson A, Andersson A, Hultberg B. Plasma homocysteine in women on oral oestrogen-containing contraceptives and in men with oestrogen-treated prostatic carcinoma. Scand J Clin Lab Invest 1992;52:283-7. 119 Chapter 7 17. Mudd SH, Pool JR. Labile methyl balance for normal humans on various dietary regimens. Metabolism 1975;24:721-33. 18. Boers G. Refinement of the methionine loading test. In: Robinson K ed. Homocysteinemia and vascular disease. Proceedings of an EC COMAC-Epidemiology Expert Group Workshop. Luxembourg: Commission of the European Communities, 1990:61-6. 19. Refsum H, Ueland PM, Svardal AM. Fully automated fluorescence assay for determining total homocysteine in plasma. Clin Chem 1989:35:1921-7. 20. Robinson K, Mayer EL, Miller DP, et al. Hyperhomocysteinemia and low pyridoxal phosphate. Common and independent reversible risk factors for coronary artery disease. Circulation 1995;92:2825-30. 21. Brattstrom L, Lindgren A, Israelsson B, Andersson A, Hultberg B. Homocysteine and cysteine: determinants of plasma levels in middle-aged and elderly subjects. J Intern Med 1994;236:633-41. 22. Ueland PM, Refsum H, Brattstrom L. Plasma homocysteine and cardiovascular disease. In: Francis RB Jr, ed. Atherosclerotic Cardiovascular Disease, Hemostasis, and Endothelial Function. New York, NY: Marcel Dekker Inc; 1992:183-236. 23. Nordstrom M, Kjellstrom T. Age dependency of cystathionine beta-syntase activity in human fibroblasts in homocyst(e)inemia and atherosclerotic vascular disease. Atherosclerosis 1992;94:213-21. 24. Dudman NPB, Wilcken DEL, Wang J, Lynch JF, Macey D, Lundberg P. Disordered methionine/homocysteine metabolism in premature vascular disease. Its occurrence, cofactor therapy, and enzymology. Arterioscler Thromb 1993;13:1253-60. 25. Wouters MGAJ, Van der Mooren MJ, Moorrees MThEC, Blom HJ, Boers GHJ, Eskes TKAB. Homocysteine metabolism and menopausal status: a study in healthy women. Abstract 321. Abstract Book Seventh International Congress on the Menopause, Stockholm, Sweden, June 20-24, 1994. 26. Van der Mooren MJ, Wouters MGAJ, Blom HJ, Schellekens LA, Eskes TKAB, Rolland R. Hormone replacement therapy may reduce high serum homocysteine in post-menopausal women. Eur J Clin Invest 1994;24:733-6. 27. Anker G, Lonning PE, Ueland PM, Refsum H, Lien EA. Plasma levels of the atherogenic amino acid homocysteine in post-menopausal women with breast cancer treated with tamoxifen. Int J Cancer 1995;60:365-8. 28. Stampfer MJ, Colditz GA. Estrogen replacement therapy and coronary heart disease: a quantitative assessment of the epidemiologic evidence. Prev Med 1991;20:47-63. 29. Frosst P, Blom HJ, Milos R, et al. A candidate genetic risk factor for vascular disease: a common mutation in methylenetetrahydrofolate reductase. Nat Genet 1995;10:111-3. 120 Homocysteine, CVD, Gender and Menopause 30. Meleady R, Daly L, Graham I for the EC Concerted Action Project: Homocysteine and Vascular Disease. Homocysteine interacts with other risk factors to modify the risk of vascular disease. Eur Heart J 1995;16 (supplement August):227 (abstract). 31. Brattstrom LE, Israelsson B, Jeppsson JO, Hultberg BL. Folic acid—an innocuous means to reduce plasma homocysteine. Scand J Clin Lab Invest 1988;48:215-21. 32. Ubbink JB, Van der Merwe A, Vermaak WJH, Delport R. Hyperhomocysteinemia and the response to vitamin supplementation. Clinical Investig 1993;71:993-8. 121 8 General Discussion Introduction The main objective of the studies that have been presented in this thesis, was to provide additional epidemiologic evidence for the hypothesis that elevated plasma tHcy is an independent risk factor for cardiovascular disease. We addressed several disease endpoints and used data of studies in which tHcy was either measured before or after onset of cardiovascular disease. Also, we studied the association in a large sample of women. In the present chapter, we discuss our main findings and results of other epidemiologic studies, concentrating on three questions that still have not been answered fully by previous studies. First, is the association of tHcy with risk of cardiovascular disease restricted to those with levels above a certain cutoff-point or is the effect graded? This issue may provide information on what part of the population could benefit from tHcy lowering: only those with high levels, or also those with low-normal levels? Second, does fasting or post-load tHcy emerge as a stronger predictor of cardiovascular disease risk? An answer to this question could yield information on the relative impact of enzymatic impairment in each of the metabolic pathways (i.e. transsulfuration and remethylation) on risk of cardiovascular disease. Conceivably, this is also related to the vitamins, that have specific roles in each of the pathways. Related to this, we studied a genetic defect, leading to reduced remethylation, as a predisposing factor to high tHcy levels and coronary atherosclerosis. Third, how do the vitamins relate to tHcy and to risk of cardiovascular disease? One would like to know whether inadequate dietary intake or status of the B-vitamins are possible causes of elevated tHcy in cardiovascular patients and how strong a determinant each of the vitamins is. In this chapter, we first discuss the main findings of our investigations. Subsequently, the study results are discussed in the light of other epidemiologic studies. Then, we address some methodological considerations. Next, in the concluding remarks, directions for future research are given, followed by possible consequences for public health. 123 Chapter 8 Main findings Main results are discussed by chapter and summarized in Table 8-1. Table 8-1. Main findings of the thesis research End- Support for Graded or Fasting or Ch. point tHcy as a risk threshold? post-load B-vitamins? factor? tHcy? C.A. yes, also related graded both equally not lower in cases to extent of strong atherosclerosis C.A. MTHFR defect is synergism of more strongly low folate and related to fasting MTHFR defect tHcy M.I. yes graded folate inversely related to tHcy and MI A.P. no n.r. LS. only in normotensives all yes, also in graded both equally only vitamin B6 (premenopausal) strong, and post- inversely related load stronger in to CVD, women than men independently of tHcy Note: "-" = not studied; "n.r." = not relevant, i.e. when main effect was absent; C.A. = coronary atherosclerosis; M.I. = myocardial infarction; A.P. = angina pectoris; I.S. = ischemic stroke. In a Dutch case-control study, plasma tHcy levels, both fasting and after methionine loading, were positively associated with risk of coronary atherosclerosis, even after adjustment for confounding factors. However, some findings were not statistically significant. Elevation of tHcy appeared to be a (weak) graded risk factor. Levels of erythrocyte folate and plasma PLP were higher in cases than controls, whereas serum vitamin B]2 was only slightly lower in cases, suggesting 124 General Discussion that higher tHcy levels in cases were not a consequence of inadequate vitamin levels (Chapter 2). In the same study population, homozygosity for the 677C—>T mutation in the gene that codes for 5,10-methylenetetrahydrofolate reductase (MTHFR)1 was associated with raised plasma tHcy levels, especially in the fasting state. Homozygosity for the mutation was not associated with increased risk of coronary atherosclerosis. However, the combination of low erythrocyte folate status and the homozygous mutant genotype was associated with particularly high tHcy levels, possibly inferring an increased risk of coronary atherosclerosis (Chapter 3). The data of a case-control study conducted in the Boston area (Boston Area Health Study, BAHS), revealed that elevation of fasting tHcy was an independent, graded risk factor for myocardial infarction. Both dietary and plasma levels of folate and vitamin B6 (in plasma measured as PLP) were significantly lower in patients than controls. Vitamin B12 did not show a clear association with risk of myocardial infarction, although subclinical vitamin B12 deficiency was more common among cases than controls. Folate was inversely related to risk of myocardial infarction, largely explained by its inverse relationship with plasma fasting tHcy (Chapter 4). In a case-control study nested within the Physicians' Health Study (PHS), we studied the association of plasma tHcy with risk of angina pectoris. Angina pectoris was accompanied by strong evidence of severe coronary atherosclerosis in all cases. We did not find convincing evidence that elevated plasma fasting tHcy was a risk factor (Chapter 5). In another prospective study, also with data of the PHS, we observed virtually no association between elevated tHcy and risk of ischemic stroke. The data suggested that the relative risk was higher in a subgroup of subjects without hypertension, but numbers of subjects were small, and the finding borderline significant (Chapter 6). The study that concentrated on gender differences in the relationship of tHcy with cardiovascular disease (not differentiated by endpoint), revealed that elevation of tHcy was a strong graded risk factor in both women and men. Risk associated with elevation of post-load tHcy was higher in women. Elevated tHcy was a strong risk factor in premenopausal women as well, despite lower tHcy levels than in men or postmenopausal women. Only levels of vitamin B6 (measured as plasma PLP), but not of folate or vitamin Bl2, were lower in cases compared to controls. Low PLP was a strong risk factor for vascular disease, largely independent of tHcy (Chapter 7). 125 Chapter 8 Comparison with results of other epidemiologic studies Previous epidemiologic studies on elevated plasma tHcy and risk of cardiovascular disease have been reviewed thoroughly.2'3 Elevated tHcy is now generally considered a risk factor for cardiovascular disease, which appears to act independently of other risk factors. In each of the chapters, we have reviewed most relevant literature, therefore in this paragraph we will discuss whether our findings are generally concordant with those of other studies, concentrating on the issues outlined in the introduction of this chapter (see also Table 8-1). Support for a positive association Overall, we found support for the hypothesis that elevated tHcy is a risk factor for cardiovascular disease. However, the two prospective analyses in the PHS, with ischemic stroke and angina pectoris as disease endpoints (Chapters 5 and 6), showed the weakest effects. The presence or absence of hypertension appeared to modify the association between tHcy and ischemic stroke, which was also found by other investigators.4 The weaker effect in the prospective case-control studies, relative to the retrospective studies in the thesis, may be related to certain types of bias and the possibly well-nourished state of physicians, as we will discuss later. To date, five other prospective studies on tHcy and risk of cardiovascular disease have been conducted. Four of these used a nested case-control design,5,6'7'8 and one followed patients with cardiovascular disease to study disease progression according to plasma tHcy levels.9 A Finnish study by Alfthan et al.8 was the only one that showed no association between plasma tHcy and risk of cardiovascular disease. The other studies found positive associations. Our comparison of risk of cardiovascular disease related to elevated tHcy in men and women (Chapter 7) showed a stronger relationship in women than in men, similar to the results obtained by Boushey et al.2 Recently, a study by Robinson et al.,10 observed a stronger association in women as well. The possibility that premenopausal women are protected from developing cardiovascular disease, as suggested by outcomes of two studies,11'12 is contradicted by our data. Graded versus threshold effect - size of effect In general, the association of elevation of tHcy with risk of cardiovascular disease has turned out to be graded, in our studies (BAHS, European study, and to a lesser extent in the Dutch study), and in those of others.2'5'6,13 In a recent meta• analysis of Boushey et al.2, the summary ORs calculated for coronary artery disease 126 General Discussion per 5 umol/L increment in fasting plasma tHcy were 1.6 (95% CI, 1.4-1.7) for men and 1.8 (95% CI, 1.3-1.9) for women. These are much higher than the OR of 1.2 per 5 umol/L increase that was calculated in the Dutch study (Chapter 2). In the BAHS, an OR per 5 umol/L increase in fasting tHcy would have been 1.6 (95% CI, 1.0-2.7 [see estimates given in Chapter 4, multivariately adjusted]). In the European study (Chapter 7), the multivariately adjusted ORs per 5 umol/L increase in fasting tHcy would have been 1.4 (95% CI, 1.2-1.6) for men and 1.6 (95% CI, 1.1-2.2) for women. ORs that were estimated in other studies come close the risk estimations of the European study and BAHS: Arnesen et al.5 found an OR of 1.32 (95% CI, 1.05- 1.65) per 4 umol/L increase in tHcy and Pancharuniti et al.13 observed an OR of 1.4 (95% CI, 1.0-2.0) per quartile increase in tHcy. Stampfer et al.7 reported an OR of 1.18 (95% CI, 1.03-1.36) per 3 umol/L increase in tHcy, showing a smaller effect. In fact, that study showed evidence that was more in favor of a threshold effect. Fasting and post-load tHcy - association with vascular disease In the study of Boushey et al.2, the summary ORs based on studies that measured fasting tHcy were of similar magnitude as the summary ORs including both studies of fasting and post-load tHcy. This suggests that fasting and post-load tHcy are equally strongly related to risk of cardiovascular disease. A study of Mansoor et al.14 also found that differences in tHcy levels between patients with early-onset peripheral vascular disease and control subjects were similar for fasting and post-load levels. Generally, this is what we found in the Dutch study and the European study. However, among females in the latter study, we found that risk associated with elevated post-load tHcy (> 80th percentile of control subjects) was higher than risk associated with elevated fasting tHcy (Chapter 7). Similarly, in the Dutch study (Chapter 2), there was a weak indication that risk was somewhat higher for subjects with abnormal post-load tHcy (> 95th percentile of controls, possibly indicating abnormalities in the transsulfuration pathway) than for subjects with fasting abnormal tHcy levels. In that same study, we reported that by measuring fasting tHcy alone, about two thirds of subjects with abnormal post-load levels would have stayed undetected. Other studies also showed that subjects with an abnormal response to methionine loading are not necessarily those with an abnormal fasting tHcy level.15,16 Several of our findings point to the significance of impaired homocysteine remethylation in predicting cardiovascular disease risk. In the BAHS (Chapter 4), based on levels of other compounds involved in homocysteine metabolism, we found indications that impaired remethylation was the predominant cause of high 127 Chapter 8 tHcy levels in the patients. Furthermore, in Chapter 3 (Dutch study) we observed that homozygosity for the 677C->T mutation in MTHFR, resulting in defective homocysteine remethylation, led to high fasting tHcy levels (and to a lesser extent to high post-load levels). We did not find, however, that homozygosity for the MTHFR mutation was a risk factor for vascular disease, unlike other investigators did.17 In the PHS, however, homozygosity for the mutation was not related to increased risk of cardiovascular disease either (unpublished results). That and our observation raise the possibility that the mutation will only emerge as a risk factor in a source population with a lower folate status than possibly existed in the source populations from which subjects were sampled. B-vitamins and gene-vitamin interaction Some previous epidemiologic studies have found lower levels of B-vitamins in cases of cardiovascular disease, compared to healthy controls. However, several other studies did not find lower blood levels of the vitamins in cardiovascular disease patients, despite their higher tHcy levels (reviewed in reference 3). For example, results from the PHS showed that plasma levels of folate and PLP were only slightly, nonsignificantly lower in 333 patients of myocardial infarction than in 333 age-matched control subjects (in spite of higher fasting tHcy levels of cases).18 Similarly, Dalery et al.19 did not observe lower plasma levels of folate or vitamin B12 in cases of coronary artery disease than in controls, despite significantly higher fasting tHcy levels in cases. Comparably, a recent study by Robinson et al.10, did not show lower plasma folate in cases of coronary disease than in controls, even though cases had higher fasting tHcy levels. In both studies, however, plasma PLP levels were significantly lower in cases than in controls. In our study of Chapter 2 (Dutch study), we did not observe lower vitamin levels in cases relative to controls. In Chapter 4 (BAHS), we showed significantly lower plasma levels of folate and PLP in cases than controls. In Chapter 7 (European Study) we observed lower levels of PLP, but not of erythrocyte folate, in cases than in controls. In that study and to a lesser extent in the BAHS, in line with findings of Robinson et al.10, the inverse relationship of vitamin B6 with risk of cardiovascular disease, was largely independent of effects of tHcy, also when measured after loading. This inverse relationship between PLP and cardiovascular disease, may be related to other 20 mechanisms, e.g. a cholesterol-lowering effect of vitamin B6. . However, in the study of Robinson et al.10 and the BAHS there is still a possiblity that PLP affected risk of cardiovascular disease through post-load tHcy, which was not measured in those studies. 128 General Discussion The observation that cases do not have lower folate levels than controls, despite their higher tHcy levels, as seen in some of our studies and several other investigations, may indicate that cases have a higher demand for folate, due to a hereditary defect in tHcy metabolism (e.g. thermolabile MTHFR). Although this may be an explanation in other studies, in our Dutch study we did not find a higher prevalence of cases with the mutation in the homozygous form (Chapter 3). In the studies of the thesis, in which vitamins were measured, plasma tHcy correlated inversely with levels of the vitamins within groups of cases and controls. Fasting tHcy correlated inversely with folate (BAHS and Dutch study). This is in line with other studies, e.g. those of Jacobsen et al.21 and Selhub et al.,22 that have shown that the vitamins, especially folate, are strongly, inversely associated with plasma fasting tHcy. In the Dutch study, PLP was the only vitamin that correlated inversely with the increase above the fasting level after methionine loading, in line with findings of Miller et al.23 Methodological considerations Most investigations on tHcy and vascular disease, including most of the studies presented in this thesis, were case-control studies that measured tHcy after the disease-onset. A major limitation of this type of study design is that one cannot rule out the possibility that elevated levels of tHcy may be influenced by the disease or its treatment. Also, in case-control studies one is always concerned about the appropriateness of the control group. The above mentioned issues pose much less of a problem in the prospective studies, like the case-control studies nested within a cohort study (e.g. PHS). However, although prospective studies have some distinct advantages, they also have limitations, i.e. tHcy in frozen plasma samples could deteriorate or prolonged follow-up of subjects could lead to attenuation of the association of interest. These advantages and disadvantages of both types of study design will be discussed. Furthermore, we will consider the choice of source population (concentrating on type of controls), and the choice of endpoints, including chances of disease misclassification. Finally, error in the measurement of tHcy and vitamins, and aspects of confounding will be reviewed. Important potential sources of bias in case-control studies Acute event Recent substantial data suggest that an acute myocardial infarction or stroke temporarily alters levels of tHcy, which could affect the apparent association with risk of disease. Lindgren et al.24 found that plasma tHcy 129 Chapter 8 was lower in the acute phase (mean of 2 days after stroke onset) than in the convalescent phase (at a median interval of about one and a half years after the stroke) in 17 patients who were newly diagnosed with a first stroke. In 20 control subjects, tHcy levels were measured twice with an interval of about 2 to 3 years, and mean tHcy levels were not different. A recent study from Landgren et al.25 compared levels of plasma tHcy measured 24-36 hours after onset of acute myocardial infarction (baseline) and 6 weeks later. The authors found plasma concentrations of tHcy to increase significantly during 6 weeks' follow-up. In the case-control study of first myocardial infarction conducted in the Boston area (Chapter 4), we showed similar results. Mean plasma tHcy was significantly lower in the blood drawn two days after the event than eight weeks later. The most plausible explanation for these findings was an acute-phase reduction in tHcy, perhaps due to a decrease in plasma albumin (the main binding protein of tHcy). Dietary changes and medication Beside the likelihood of alteration of tHcy levels after an acute event, there are two other limitations of the case-control design: 1) medication prescribed for patients with cardiovascular disease may modify tHcy levels and; 2) patients may alter their diet as a reaction to their illness. The first has been shown for use of diuretics, which is associated with increased plasma tHcy levels. In Landgren's paper,25 however, tHcy changes during follow-up were not related to medications (nitroglycerine, streptokinase, beta blockers or acetylsalicylic acid). Among the dietary changes, the two most important are changes in vitamin supplement use (either B-vitamins or multivitamins), or in intake of foods rich in folate. In our Boston area case-control study, the plasma folate level (which in that study was the strongest determinant of plasma tHcy) had not changed in the 8 weeks following the acute event, suggesting that folate intake had not changed. In Lindgren's study,24 during the convalescent phase, concentrations of cobalamin, folate, had not changed either. Thus, changed dietary habits and effect of medication appear to be less of a problem than acute phase effects on plasma tHcy levels, although this depends on the studied population, of course. The upper part of Figure 8-1 depicts the possible sources of bias in case-control studies. These potential limitations found with case-control studies do not apply to cross-sectional studies, e.g. those that have carotid-artery wall-thickness as an endpoint. Among individuals who had no symptoms of cardiovascular disease, Malinow et al.26 and Selhub et al.27 showed significantly increased risks of carotid atherosclerosis among those with high tHcy levels. These studies also avoid the often difficult problem that can plague case-control studies, which is choosing appropriate and comparable controls. However, these cross-sectional studies do not address clinical endpoints. 130 General Discussion Case-control studies Acute phase Medication and effects dietary changes lower tHcy? alter tHcy? Prospective studies Effect attenuation when prolonged? tHcy deterioration at long storage? Disease onset Blood sampling tHcy measurement Figure 8-1. Types of bias that may occur in case-control studies studies and prospective studies. Overall, in the presented case-control studies in which tHcy was measured after the disease (Chapters 2, 3, 4, 7), blood samples were not drawn in the acute phase (e.g. in the case of myocardial infarction), neither were samples drawn close to an operation (e.g. cardiac catheterization). Usually there was at least a time period of 2 or 3 months between event or hospitalization and blood sampling. However, medication that was prescribed during those months, and changed dietary habits may have affected tHcy in cases (e.g. patients of myocardial infarction in the BAHS) or even in cases and controls (e.g. catheterization patients in the Dutch 131 Chapter 8 study). In the latter study, there were indications that use of acetylsalicylic acid or other anti-thrombotic drugs and changed dietary habits had lowered tHcy levels in the hospital-based controls. Since most cases were using those drugs, and reported alteration of dietary habits more frequently than the controls, these factors may have weakened the association of interest in that study. For the genotyping of the thermolabile mutation in MTHFR (Chapter 3), the above mentioned types of bias cannot have occurred, since this is an inherited factor. Important sources of bias in prospective studies important potential sources of bias in prospective studies, possibly explaining the lack of an association found by some studies (including the ones presented in the thesis), will be subsequently discussed (see also lower part of Figure 8-1). Prolonged storage In nested case-control studies with prospectively collected samples, the blood is often stored for an extended period. Alfthan et al.8 suggested that their null findings may have been due to a low prevalence in the Finnish population of genes predisposing to elevation of plasma tHcy. However, another possibility these investigators raised was that the samples may have deteriorated during the 14 years of storage at -20°C. Mereau-Richard et al.28 reported that the plasma concentration of tHcy decreased after prolonged storage, although they did not describe the storage conditions. However, plasma tHcy is known to be stable for at least one year at -20°C, and the distribution of values from stored samples are generally similar to those from assays on freshly obtained plasma.29 Hence, this seems an unlikely source of bias in prospective studies, especially in those that store plasma samples at even lower temperatures. Accordingly, the follow-up study of Perry et al.,6 in which serum samples were stored for 12.8 years at -20 ° C, showed a convincing positive graded association of serum tHcy with risk of stroke. Although tHcy determination is usually performed relatively soon after blood samples have been drawn in case-control studies, in the BAHS blood samples had been stored for almost 10 years. Again, this did not seem to have led to effect attenuation. The problem with prolonged storage cannot have occurred in a recently published prospective study on serum folate and risk of ischemic stroke.30 In that study, serum folate determinations were performed in blood samples obtained from study participants during the baseline examination. Investigators found a weak, nonsignificant inverse association between serum folate and risk of ischemic stroke. Although many factors could have been responsible for this weak effect, one of the possibilities is an effect attenuation because of extended follow-up (13 years in that study), as will be discussed below. 132 General Discussion Prolonged follow-up When the follow-up interval after baseline sampling becomes long, the relation between tHcy levels and cardiovascular events occurring many years afterwards may become attenuated, due to changes in tHcy during the follow-up period that were not accounted for (see lower part of figure 8-1). This may provide another possible explanation for the lack of effect in the Finnish study,8 in which the follow-up period was 14 years. Evidence for the attenuation of the predictive capability for a single tHcy determination was present in the PHS in which the relative risk of myocardial infarction for men with elevated tHcy was reduced to 1.3 (95% CI, 0.5 - 1.3) when an additional group of men who developed myocardial infarction 6-9 years after baseline was included (the initial analyses were based on a 5 year follow-up period).18 However this problem does not seem to have occurred in the British Regional Heart Study,6 after almost 13 years of follow- up. Conversely, some data seem to indicate that plasma tHcy is fairly stable in individuals over time. In the PHS,18 a second blood sample was drawn, ten years after the first, from 49 Boston-area participants who had remained free from diagnosed cardiovascular disease. The Spearman correlation coefficient between tHcy measured in the samples drawn 10 years apart was 0.68 (P = 0.0001), showing that the within-person variation, even over a long period of time, is relatively small. In 76 men, Israelsson et al.31 measured tHcy in fresh plasma and plasma that had been stored at -20 ° C for a mean time of 10.9 ± 2.5 (SD) years. The mean plasma tHcy level was higher in the fresh samples than in the stored samples, which may be related to increasing age in the subjects. However, values in fresh and stored samples correlated significantly (r = 0.58, P < 0.001). This correlation was apparent in subjects that suffered a vascular event and in subjects that remained healthy. Similarly, as mentioned earlier in this chapter, Lindgren et al.24 showed virtually no change in median tHcy levels during a 2-3 year time period among 20 control subjects. However, in other populations, changes in diet and vitamin supplement usage during follow-up might vary, altering tHcy levels and possibly leading to an attenuation of the relative risk estimates. Thus, with respect to the prospective studies discussed in the thesis (Chapters 5 and 6), there are indications that the within-person variability over the years is small. Also, literature provides information that tHcy is stable for years when stored at -20° C or at lower temperatures. For both endpoints discussed in this thesis we also used a follow-up period of five years. However, if tHcy is less strongly related to an atherosclerotic endpoint (e.g. angina pectoris), than to a thrombotic endpoint (e.g. myocardial infarction), as we speculated in Chapter 5, then an association may disappear to a large extent, even at 5 years of follow-up. 133 Chapter 8 Choice of study population We will discuss the advantages and disadvantages of different source populations used in the studies presented in the thesis, and related to this the selection of an appropriate control group. Source population In the thesis, various source populations were chosen: a general population, i.e. the BAHS, a hospital-based population, i.e. the catheterization patients, and a specific segment of the general population, i.e. the PHS. We will not discuss in detail the European multi-center case-control study, since various types of source populations (and control groups) were used by the different centers. It can be hypothesized that associations studied in a specific, well-nourished, population segment like the PHS will be less strong than in a study population derived from the general population, for vitamin status will most likely be better in physicians. For example, among control subjects, use of vitamin supplements was twice as frequent in the PHS than in the European case-control study (see Chapters 5 and 7). Besides the biases discussed before, this may in part be responsible for the relatively weak (for ischemic stroke) and virtually absent (for angina pectoris) associations in the reported investigations based on the PHS. Of course, choosing a well-nourished population does not cause a problem per se when mutations in enzymes involved in homocysteine metabolism are studied (although not performed in the PHS investigations in this thesis). However, considering that a low folate status may be a prerequisite for homozygous deficient subjects to develop high tHcy levels, like our study in Chapter 3 suggests, the choice of a well-nourished population may also lead to a null-finding in that occasion. In fact, in the PHS, homozygosity for thermolabile MTHFR was not associated with increased risk of myocardial infarction (unpublished results). Control group In a cohort study where cases and controls are not selected on the basis of a certain exposure, like in the PHS, one does not have the problem of selecting an appropriate control group, since an internal comparison group will be used. However, in the case-control studies (Chapters 2, 3, 4, 7), the selection of a proper control group is of major importance. In the Dutch study (Chapter 2), we used both angiographic controls and population-based controls. The angiographic controls were selected from subjects that had undergone cardiac catheterization in the same hospital as the cases, but whose angiograms showed no substantial narrowing. Consequently, they shared the same selective processes by which the cases were identified and one could expect them to have the same accuracy in reporting information. However, the outcome of the catheterization and the factors 134 General Discussion that lead to the decision to perform catheterization (i.e. myocardial infarction in the cases or known valve defect in controls) may still have led to different types or strengths of biases (e.g. effect of medication or changed lifestyle) in both groups. This is even more likely to have happened since cases had been under cardiologic treatment for a longer time period. It is difficult to judge how this may have influenced the associations. The second control group used in this case-control study, a sample from the general population, consisted of subjects living close to the hospital. Had these subjects developed the disease (e.g. be diagnosed with angina pectoris) they would have been hospitalized for catheterization in that same hospital, and in that aspect they form a good reflection of the source population of cases. However, preferably we should have selected population-based controls from the same area cases were living in. In the BAHS (Chapter 4), controls were selected from the general population, from subjects residing in the same area as the subjects that were hospitalized with a first myocardial infarction, thus both groups were derived from the same source population. Of course, selection processes for the groups were different. Recall bias (e.g. with regard to dietary intake) may have occurred in this population. However, in cases and controls we observed similar correlations between reported vitamin intake and plasma levels, thus recall bias was probably not a major issue. Choice of disease endpoint Since we studied various manifestations of coronary heart disease, we will focus on that endpoint in this part of the discussion. Generally, a myocardial infarction is the result of the development, of atherosclerosis and an acute process, i.e. clot formation. Homocysteine may be involved in both processes.32 By taking angiographically defined coronary atherosclerosis as the disease endpoint in Chapter 2, we intended to sort out whether elevated tHcy levels could lead to coronary artery disease via its effects on coronary atherosclerosis. This goal could be additionally reached by grading the disease, i.e. by investigating a possible dose- response relationship between plasma tHcy and number of occluded arteries. Inevitably, a major part (about 50%) of the cases had a history of myocardial infarction prior to the catheterization, which made it difficult to draw conclusions for the processes of atherogenesis and thrombus formation, separately. Nevertheless, the mean plasma tHcy level was similar for cases with and without a history of myocardial infarction, and there was a trend of increasing plasma tHcy with increasing number of occluded vessels. Thus, our data confirm an atherogenic action of elevated tHcy levels, independently from'its possible fhrombogenic effects. 135 Chapter 8 With regard to preventing misclassification of coronary atherosclerosis, one could debate over which control subjects are preferable: those from the general population without clinical symptoms, or an angiographic control group?33 Use of a control group free from clinical disease may dilute the degree of association, since many of the subjects without clinical disease may have substantial coronary narrowing. In that respect, angiographic controls may be preferable. In the study of Chapter 2, we claim to have obtained a substantial contrast between cases and angiographic controls, since the majority of cases had major occlusions in at least two vessels, whereas most angiographic controls had no substantial narrowing in any of the vessels. Considering the severity of coronary occlusion in cases, there must have been a fair contrast between cases and population-based controls as well. The endpoint of angina pectoris in the PHS (Chapter 5) may pose more of a problem with regard to disease misclassification. Although in the majority of patients there was additional strong evidence of severe coronary occlusion, there was no objective information on the amount of stenosis in controls. In a group of middle-aged men, subclinical coronary artery stenosis may be common, and this may have weakened the effect. For myocardial infarction as disease endpoint (Chapter 4), chances of misclassification were small: diagnostic criteria included enzyme changes and electrocardiographic abnormalities. Exposure measurement When discussing the possible sources of bias in case-control studies and prospective studies, we have already referred to the possibility that tHcy levels may be affected during the acute phase of stroke or myocardial infarction, or that tHcy may deteriorate at prolonged storage. These phenomena would lead to exposure measurement error. In general, when there is little exposure measurement error, the measured exposure comes close to the "true" exposure of interest. Measurement error may be caused by high intra-individual variation (i.e. in tHcy and vitamin levels) and imprecision of the determinations. Incorrect handling of blood samples may lead to imprecise assessment of tHcy exposure status of an individual. Upon storage at room temperature of whole blood for longer than 4 hours, tHcy in plasma may increase with 35%.29 This increase can be avoided when blood is put on ice immediately after collection, or when plasma or serum is prepared within one hour after collection,29 as has been performed in the studies of Chapters 2, 3 (Dutch study) and 7 (European study). In prospective studies that were riot originally designed to study the relationship 136 General Discussion between tHcy and risk of cardiovascular disease, this may be a source of measurement error. Different methods of tHcy determinations were used within the studies in the thesis. The different methods correlate well.22 The coefficients of variation for the tHcy determinations vary between 3% and 7%, and this is not considered a major source of measurement error. Imprecise assessment of tHcy levels in an individual may also occur when blood is not sampled from subjects in the fasting state.34'35 The fact that blood was not drawn fasting in the reported prospective studies, may have led to a fair amount of measurement error in those studies. In a similar way, the conscientiousness with which study subjects follow guidelines regarding food consumption during the 6 hours following the methionine loading (as in the Dutch study, where participants were allowed to leave the research unit), may determine precision of post-load tHcy measurement. The within-person variability in tHcy relative to the between-person variability would also determine the magnitude of exposure measurement error. We do not know of an epidemiologic study that measured tHcy at various moments in time, to obtain a value closer to the "true" tHcy levels. The within-person variability has not been measured often, but appears to be small.18'24 Most of the remarks made for tHcy, account also for measurements of the vitamins in plasma, serum, or whole blood. For example, after blood sampling, precautions have to be taken to protect folate against oxidative destruction prior to assay (by adding a reducing agent such as ascorbate).36 Plasma PLP is probably stable at long storage, but for folate this is unsure.37 The sensitivity of folate to oxidation may lead to a substantial amount of measurement error, maybe explaining why folate was not found to be lower in cases than in controls in many studies. Finally, when biomarkers of dietary intake of folate and vitamin B6 are measured in studies, the long-term exposure measurements (e.g. erythrocyte folate, which reflects body stores of folate) may be preferable over short-term exposure measurements (e.g. plasma folate, which measures recent intake), for they may refer to a more relevant time-period (especially in case-control studies). In general, it is hard to draw conclusions on the effect of the above described causes of measurement error, even if nondifferential. Effect attenuation will probably occur in most occasions, but a strengthening of the association between tHcy and risk of cardiovascular disease is not unlikely either.38 However, considering the possibly low intra-individual variation and low coefficient of variation, a single measurement of tHcy may not be a problem when a sufficient number of subjects is studied and when strict guidelines for blood sampling and 137 Chapter 8 handling are applied. Confounders Gender and age may be important confounding factors in the association of interest, since men usually have higher tHcy levels, both fasting and after methionine loading,21,22,39 and tHcy has been shown to increase with advancing age.39,40 Notably, gender and age are related to risk of vascular disease as well. The PHS is restricted to males, thus confounding by gender was not a problem. Cases and controls were individually matched on age and smoking habits, and specific analysis for matched data were used. In the Dutch study, by applying frequency matching, we aimed at obtaining similar numbers of males and females, and similar age distributions in the three study groups. We hoped to be able to perform analyses stratified by gender and possibly by age strata, with equal ratios of cases and controls within the strata, but we did not succeed. To adjust for confounding by age and gender, we controlled for these factors in multivariate analyses. Similarly, in the BAHS, where we used unmatched groups of cases and controls (sampled from a larger study in which cases and controls were individually matched on age and gender), we controlled for these factors by multivariate analyses as well. In the European study, the aim was to frequency match cases and controls on age and gender, but matching was not successful. We studied associations for men and women separately, adjusting for age in multivariate analyses. Many studies have found that elevated tHcy is a risk factor for cardiovascular disease, independently of known risk factors, i.e. total cholesterol, high blood pressure, smoking, diabetes mellitus, and body mass index (reviewed in reference 2). In the studies presented in the thesis, we controlled for most of these factors in multivariate analyses, and found effects (if present in crude estimations) to remain, although some got weaker (e.g. European study, especially in women). There are some indications that elevated tHcy may be related to high blood pressure, high cholesterol levels, and smoking.39 Whether elevated tHcy performs its action through one of these factors (or the other way around), needs to be further studied. At this moment, however, there appears to be enough evidence to conclude that elevated tHcy acts as a risk factor for cardiovascular disease, independently of established risk factors. Impaired kidney function, reflected by elevation of serum creatinine levels, may be an additional confounder, since it is associated with increased tHcy levels41,42 and may be related to cardiovascular disease as well. We have not been able to control for creatinine levels in all studies. However, associations between 138 General Discussion elevated tHcy and risk of cardiovascular disease remained after adjustment for this factor (see Chapters 2 and 7). Concluding remarks Epidemiologic evidence Our findings, in line with those of other epidemiologic studies, indicated that elevated tHcy is an independent risk factor for cardiovascular disease. The effect appeared to be graded, indicating that also subjects with previously considered "normal" tHcy levels may be at increased risk of developing cardiovascular disease. There seemed to be no threshold phenomenon. We showed that elevated tHcy also was a strong risk factor for cardiovascular disease in women, even when premenopausal. Early studies focused on enzymatic deficiencies in the transsulfuration pathway, (i.e. heterozygosity for cystathionine 6-synfhase deficiency, responsible for extreme elevations of post-load tHcy). Recent studies, including our studies, have pointed to the importance of impaired remethylation of homocysteine, leading to moderately elevated fasting tHcy levels, in predicting risk of cardiovascular disease. With respect to the role of the vitamins, folate was found to be an important determinant of plasma (fasting) tHcy, even more so in subjects with a homozygous genetic defect leading to impaired homocysteine remethylation (thermolabile MTHFR). Low plasma PLP and low intake of vitamin B6 were associated with increased cardiovascular risk, largely independently of tHcy. Methodologic issues In this thesis, case-control studies in which tHcy was measured after the disease tended to find stronger effects than the studies in which blood samples were drawn prospectively. This could be a consequence of the fact that some of the case- control studies selected subjects who manifested cardiovascular disease at a relatively early age (e.g. Dutch study and European study). Also, the discussed effect dilution which may occur at long storage of samples or prolonged follow-up of study subjects, may play a role in this. Furthermore, the possibly better vitamin status in physicians may be responsible. However, the findings of the prospective studies conducted by others,5,6,7 are generally concordant with those from case- control studies and cross-sectional studies. They support an independent association between elevated tHcy levels and risk of cardiovascular disease, and exclude the possibility that elevated tHcy is a consequence rather than a cause of the disease. 139 Chapter 8 When designing case-control studies, the acute-phase effects, resulting in temporary reduction of tHcy, should be considered carefully. Since there may be a temporary decrease of plasma PLP as well (see BAHS), inverse relationships between PLP and risk of cardiovascular disease, observed in case-control studies, should be interpreted with caution. The issue of dietary changes after disease onset may be resolved by assessing any important changes in diet that occurred after the event, and excluding such subjects from the main analysis. A similar approach can be taken for medications that could have seriously affected tHcy levels. To better interpret results from prospective studies, more research is needed on the stability of tHcy during prolonged storage and the variability within subjects over time. This wifhin-person variability may differ between populations and for different time periods, depending on changes in factors that influence tHcy levels. Since we do not know these factors with certainty, it seems preferable in general to have large cohort or nested case-control studies with a relatively short follow-up period (e.g., up to five years), rather than relatively small cohort studies with a long follow-up period. Recommendations for future research Despite the large number of studies completed, the quantification of risk remains relatively imprecise. Our findings varied between 20% and 60% increase in risk per each 5 umol/L rise in fasting tHcy. The summary ORs of the meta-analysis of Boushey et al.2 showed an increase in risk of coronary artery disease of about 70% for each 5 umol/L increase in fasting tHcy. Whether there is a tHcy level below which there is no increased risk, remains to be studied. In particular, it will be important to assess, by studying different populations, the risk associated with tHcy in relation to varying dietary and genetic factors. The interaction between the thermolabihty of MTHFR and low folate, as shown in Chapter 3, needs special attention. Whether subjects with elevated post-load tHcy are a specific group, as suggested by our studies and those of others,15,16 and whether these subjects are at a higher risk for cardiovascular disease, needs further attention as well. The methionine loading test is, however, a time consuming measurement in epidemiologic research. The usefulness of a shortened test, e.g. of two hours, is promising,43 but needs further testing. Also, more research is needed on the biological mechanisms through which elevated tHcy causes vascular disease, since considerable controversy about the mechanisms remains. However, the results of various studies that have focused on 140 General Discussion that issue give plausible mechanisms, and support a causal role for elevated tHcy in the processes of atherosclerosis and possibly thrombosis.3'32 Most epidemiologic studies had atherosclerotic vascular disease as an endpoint,2 supporting the atherogenic action of elevated tHcy. However, some epidemiologic studies have indicated that elevated tHcy is strongly related to thrombosis, as well.44,45 Although further observational studies may add important new information, the available data are more than sufficient to warrant immediate initiation of large- scale intervention trials of tHcy lowering, studying possible reduction of risk of cardiovascular disease, as has been stressed before.46 There are indications for a beneficial effect, as for example vitamin supplementation may diminish the endothelial dysfunction in vascular patients with elevated tHcy.47 So far, no studies have investigated the effect of tHcy lowering, by increased vitamin intake, on the occurrence of "hard" cardiovascular endpoints. Public health implications Our data, and those of others have indicated that folate is most strongly related to plasma tHcy, compared to the other two vitamins involved in homocysteine metabolism. Additionally, high folate intake was found to be related to decreased risk of cardiovascular disease (Chapter 4). Although we found that vitamin B6 was inversely related to risk of cardiovascular disease (Chapter 7 and to a lesser extent Chapter 4), this association was largely independent of plasma tHcy. Therefore, we will focus on folate in this final part of the discussion. Folate can most successfully lower tHcy levels,'1'12'25'48"51 even in subjects with normal folate levels. The lowest dose of folate supplementation reported so far was 650 ug/day, with which tHcy levels were reduced by about 50% in subjects with elevated tHcy.50 In the study of Selhub et al.22 and our study of Chapter 4 (BAHS), it was observed that tHcy reached its nadir at folate intakes higher than approximately 400 ug/d. Apparently, this may be the minimal intake level we should aim at, in order to prevent elevation of tHcy. An extra intake of 400 ug/d of folate, either obtained from food, supplements or fortified foods does not seem to have adverse side 36 effects. There is a possibility that vitamin B12 deficiency would be masked, however. When cobalamin would be included in the supplements or added to the fortified foods, a possible vitamin B12 deficiency would be corrected. Selhub et al.22 estimated that the intake of 400 ug/d of folate was only attained by 30-40% of the participants in the Framingham study. From the NHANES II study, a percentage of 88% was estimated for US adults.52 For the 141 Chapter 8 population in the Netherlands, there is very little information available on the intake of folate. The Dutch Nutrients Table does not include folate levels in food, thus intake was not estimated in the last National Nutrition Survey. However, considering that green leafy vegetables and fruits are major sources of folate intake, and that the consumption of these food items declined between 1987 and 1992,53 a substantial proportion of the Dutch population may be ingesting too little folate as well. Therefore, increasing folate intake seems an important way of lowering tHcy in populations, possibly reducing cardiovascular disease incidence. Boushey et al.2 assessed that if the US population were to eat two or three more servings of fruits and vegetables per day (at a 80% effective education), dietary folate intake would increase by 100 ug/d, thus leading to an increase of approximately 50 ug/d bioavailable folate. They estimated that this would lead to a reduction of about 4% of cardiovascular disease deaths per year in the US, through effects of tHcy reduction. They furthermore estimated that if one half of the population were to take supplements, adding about 400 ug/d to the normal intake, a similar percentage of deaths could be prevented. Food fortification at 350 ug/100 g would lead to a reduction of about 8% of deaths due to cardiovascular disease per year. Large scale prevention trials are warranted to demonstrate the efficacy of tHcy lowering, and the minimal amount of extra folate intake that is required. However, at this moment, based on the available epidemiologic evidence, it is at least recommendable to increase consumption of fruits and vegetables in the general population. Green leafy vegetables, beets, broccoli, and cabbage are good sources of folate, whereas oranges and bananas are fair sources. In September 1995, in the Netherlands, the Fruit and Vegetables Council started a campaign to increase the intake of fruits and vegetables (advising to consume a daily rmiiimum of two servings of fruit and two servings of vegetables). According to estimations of Boushey at al.,2 80% success of the campaign could roughly increase mean daily intake of folate by 100 ug, reducing the total number of all cardiovascular deaths by about 4% per year, (assuming that we can extrapolate the US llndings to the Dutch situation, and that the effect is only through tHcy lowering). Results from prevention trials, which will hopefully become available soon, will indicate whether additional measures, i.e. food fortification or supplementation (e.g. in middle-aged and elderly subpopulations, susceptible to low folate intake) are justified. They may be necessary, since obtaining sufficient folate from dietary sources alone may be difficult. In general, if dosed correctly, an increase of folate consumption at a population level may also prevent a proportion of neural tube defects,54 and possibly of colon cancer as well.55 142 General Discussion References 1. Frosst P, Blom HJ, Milos R, et al. A candidate genetic risk factor for vascular disease: a common mutation in methylenetetrahydrofolate reductase. Nat Genet 1995;10:111-3. 2. Boushey CJ, Beresford SAA, Omenn GS, Motulsky AG. A quantitative assessment of plasma homocysteine as a risk factor for vascular disease. Probable benefits of increasing folic acid intakes. JAMA 1995;274:1049-57. 3. Ueland PM, Refsum H, Brattstrom L. Plasma homocysteine and cardiovascular disease. In: Francis RB Jr, ed. Atherosclerotic Cardiovascular Disease, Hemostasis, and Endothelial Function. New York, NY: Marcel Dekker Inc; 1992:183-236. 4. Araki A, Yoshiyasu S, Fukushima Y, Matsumoto M, Asada T, Kita T. Plasma sulphydryl- containing amino acids in patients with cerebral infarction and in hypertensive patients. Atherosclerosis 1989;79:139-46. 5. Arnesen E, Refsum H, Bíjmaa KH, Ueland PM, F(|)rde OH, Nordrehaug JE. Serum total homocysteine and coronary heart disease. Int J Epidemiol 1995;24:704-9. 6. Perry IJ, Refsum H, Morris RW, Ebrahim SB, Ueland PM, Shaper AG. Prospective study of serum total homocysteine concentration and risk of stroke in middle-aged British men. Lancet 1995;346:1395-8. 7. Stampfer MJ, Malinow MR, Willett WC, et al. A prospective study of plasma homocysteine and risk of myocardial infarction in US physicians. JAMA 1992;268:877-81. 8. Alfthan G, Pekkanen J, Jauhiainen M, et al. Relation of serum homocysteine and lipoprotein(a) concentrations to atherosclerotic disease in a prospective Finnish population based study. Atherosclerosis 1994;106:9-19. 9. Taylor LM Jr, DeFrang RD, Harris EJ Jr, Porter JM. The association of elevated plasma homocysteine with progression of symptomatic peripheral arterial disease. J Vase Surg 1991;13:128-36. 10. Robinson K, Mayer EL, Miller DP, et al. Hyperhomocysteinemia and low pyridoxal phosphate. Common and independent reversible risk factors for coronary artery disease. Circulation 1995;92:2825-30. 11. Boers GH, Smals AG, Trijbels FJ, Leermakers AI, Kloppenborg PW. Unique efficiency of methionine metabolism in premenopausal women may protect against vascular disease in the reproductive years. J Clin Invest 1983;72:1971-6. 12. Brattstrom LE, Hultberg BL, Hardebo JE. Folic acid responsive post-menopausal homocysteinemia. Metabolism 1985;34:1073-7. 143 Chapter 8 13. Pancharuniti N, Lewis CA, Sauberlich HE, et al. Plasma homocysteine, folate, and vitamin BI2 concentrations and risk of early-onset coronary artery disease. Am J Clin Nutr 1994;59:940-8. 14. Mansoor MA, Bergmark C, Svardal AM, L(|)nning PE, Ueland PM. Redox status and protein binding of plasma homocysteine and other aminothiols in patients with early-onset peripheral vascular disease. Arterioscler Thromb Vase Biol 1995;15:232-40. 15. Bostom AG, Jacques PF, Nadeau MR, Williams RR, Ellison RC, Selhub J. Post- methionine load hyperhomocysteinemia in persons with normal fasting total plasma homocysteine: initial results from The NHLBI Family Heart Study. Atherosclerosis 1995;116:147-51. 16. Daly L, Meleady R, Graham I. Fasting or post-methionine load homocysteine: which should be measured in relation to vascular risk? Irish J Med Sci 1995;164(S15):6 (abstract). 17. Kluijtmans LAJ, Van den Heuvel LPWJ, Boers GHJ, et al. Molecular genetic analysis in mild hyperhomocysteinemia: a common mutation in the methylenetetrahydrofolate reductase gene is a genetic risk factor for cardiovascular disease. Am J Hum Genet 1996 (in press). 18. Chasan-Taber L, Selhub J, Rosenberg IH, et al. A prospective study of folate and vitamin B6 and risk of myocardial infarction in US physicians. J Am Coll Nutr 1996 (in press). 19. Dalery K, Lussier-Cacan S, Selhub J, Davignon J, Latour Y, Genest J. Homocysteine and coronary artery disease in French Canadian subjects: relation with vitamins BI2, B6, pyridoxal phosphate, and folate. Am J Cardiol 1995;75:1107-11. 20. Willett WC. Does low vitamin B6 intake increase the risk of coronary heart disease? In: Reynolds RD, Leklem JE eds. Vitamin B6: Its Role in Health and Disease. New York, NY: Alan R. Liss Inc;1985:337-46. 21. Jacobsen DW, Gatautis VJ, Green R, et al. Rapid HPLC determination of total homocysteine and other thiols in serum and plasma: sex differences and correlation with cobalamin and folate concentrations in healthy subjects. Clin Chem 1994;40:873-81. 22. Selhub J, Jacques PF, Wilson PWF, Rush D, Rosenberg IH. Vitamin status and intake as primary determinants of homocysteinemia in an elderly population. JAMA 1993;270:2693-8. 23. Miller JW, Nadeau MR, Smith D, Selhub J. Vitamin B-6 deficiency vs. folate deficiency: comparison of responses to methionine loading in rats. Am J Clin Nutr 1994;59:1033-9. 24. Lindgren A, Brattstrom L, Norrving B, Hultberg B, Andersson A, Johansson BP. Plasma homocysteine in the acute and convalescent phases after stroke. Stroke 1995;26:795-800. 144 General Discussion 25. Landgren F, Israelsson B, Lindgren A, Hultberg B, Andersson A, Brattstrôra L. Plasma homocysteine in acute myocardial infarction: homocysteine-lowering effect of folic acid. J Intern Med 1995;237:381-8. 26. Malinow MR, Nieto FJ, Szklo M, Chambless LE, Bond G. Carotid artery intimal-medial wall thickening and plasma homocysteine in asymptomatic adults. The Atherosclerosis Risk in Communities Study. Circulation 1993;87:1107-13. 27. Selhub J, Jacques PF, Bostom AG, et al. Association between plasma homocysteine con• centrations and extracranial carotid-artery stenosis. N Engl J Med 1995;332:286-91. 28. Mereau-Richard C, Muller JP, Faivre E, Ardouin P, Rousseaux J. Total plasma homocysteine determination in subjects with premature cerebral vascular disease. Clin Chem 1991;37:126. 29. Ueland PM, Refsum H, Stabler SP, Malinow MR, Andersson A, Allen RH. Total homocysteine in plasma or serum: methods and clinical applications. Clin Chem 1993;39:1764-79. 30. Giles WH, Kittner SJ, Anda RF, Croft JB, Casper ML. Serum folate and risk for ischemic stroke. First national health and nutrition examination survey epidemiologic follow-up study. Stroke 1995;26:1166-70. 31. Israelsson B, Brattstrôm L, Refsum H. Homocysteine in frozen blood samples. A short cut to establish hyperhomocysteinemia as a risk factor for coronary atherosclerosis? Scand J Clin Lab Invest 1993;3:465-9. 32. Rees MM, Rodgers GM. Homocysteinemia: association of a metabolic disorder with vascular disease and thrombosis. Thromb Res 1993;71:337-59. 33. Pearson TA, Derby CA. Invited commentary: should artériographie case-control studies be used to identify causes of atherosclerotic coronary artery disease? Am J Epidemiol 1991;134:123-8. 34. Ubbink JB, Vermaak WJH, Van der Merwe A, Becker PJ. The effect of blood sample aging and food consumption on plasma total homocysteine levels. Clin Chim Acta 1992;207:119-28. 35. Guttormsen AB, Schneede J, Fiskerstrand T, Ueland PM, Refsum HM. Plasma concentrations of homocysteine and other aminothiol compounds are related to food intake in healthy human subjects. J Nutr 1994;124:1934-41. 36. Herbert V, Colman N, Jacob E. Folic acid and vitamin B12. In: Goodhart RS, Shils ME. Modern Nutrition in Health and Disease. Philadelphia: Lea and Febiger; 1980:229-59. 37. Willett WC. Nutritional Epidemiology. New York, NY: Oxford University Press, 1989. 145 Chapter 8 38. Armstrong BK, White E, Saracci R. Exposure measurement error and its effect. In: Armstrong BK, White E, Saracci R (eds). Principles of exposure measurement in epidemiology. Oxford: Oxford University Press, 1992:49-77. 39. Nygärd O, Vollset SE, Refsum H, et al. Total plasma homocysteine and cardiovascular risk profile: The Hordaland homocysteine study. JAMA 1995;274:1526-33. 40. Malinow MR. Homocysteine and arterial occlusive diseases (Frontiers in Medicine). J Intern Med 1994;236:603-17. 41. Hultberg B, Andersson A, Arnadottir M. Reduced, free and total fractions of homocysteine and other thiol compounds in plasma from patients with renal failure. Nephron 1995;70:62-7. 42. Bostom AG, Shemin D, Lapane KL, et al. Hyperhomocysteinemia and traditional cardiovascular disease risk factors in end-stage renal disease patients on dialysis: a case- control study. Atherosclerosis 1995;114:93-103. 43. Bostom AG, Roubenoff R, Dellaripa P, et al. Validation of abbreviated oral methionine- loading test. Clin Chem 1995;41:948-9. 44. Den Heijer M, Blom HJ, Gerrits WBJ, et al. Is hyperhomocysteinemia a risk factor for recurrent venous thrombosis? Lancet 1995;35:882-5. 45. Falcon CR, Cattaneo M, Panzeri D, Martineiii I, Mannucio Mannucci P. High prevalence of hyperhomocyst(e)memia in patients with juvenile thrombosis. Arterioscler Thromb 1994;14:1080-3. 46. Stampfer MJ, Willett WC. Homocysteine and marginal vitamin deficiency. The importance of adequate vitamin intake. JAMA 1993;270:2726-7. 47. Van den Berg M, Boers GHJ, Franken DG, et al. Hyperhomocysteinaemia and endothelial dysfunction in young patients with peripheral arterial occlusive disease. Eur J Clin Invest 1995;25:176-81. 48. Brattström LE, Israelsson B, Jeppsson JO, Hultberg BL. Folic acid—an innocuous means to reduce plasma homocysteine. Scand J Clin Lab Invest 1988;48:215-21. 49. Dudman NPB, Wilcken DEL, Wang J, Lynch JF, Macey D, Lundberg P. Disordered methionine/homocysteine metabolism in premature vascular disease. Its occurrence, cofactor therapy, and enzymology. Arterioscler Thromb 1993;13:1253-60. 50. Ubbink JB, Vermaak WJH, Van der Merwe A, Becker PJ, Delport R, Potgieter HC. Vitamin requirements for the treatment of hyperhomocysteinemia in humans. J Nutr 1994;124:1927-33. 51. Franken DG, Boers GHJ, Blom HJ, Trijbels FJM, Kloppenborg PWC. Treatment of mild hyperhomocysteinemia in vascular disease patients. Arterioscler Thromb 1994;14:465-70. 146 General Discussion 52. Subar AF, Block G, James LD. Folate intake and food sources in the US population. Am J Clin Nutr 1989;50:508-16. 53. Fruit and Vegetables Council. Consumption of vegetables, fruits, nuts, and fruit juices in the Netherlands. National Nutrition Survey 1992. The Hague, Fruit and Vegetables Council, 1994 (in Dutch). 54. Steegers-Theunissen RP. Folate metabolism and neural tube defects: a review. Eur J Obstet Gynecol Reprod Biol 1995;61:39-48. 55. Kearney J, Giovannucci E, Rimm EB, et al. Diet, alcohol, and smoking and the occurrence of hyperplastic polyps of the colon. Cancer Causes Control 1995;6:45-56. 147 Summary Cardiovascular disease constitutes a major public health problem in the Netherlands and other Western countries. The established risk factors for cardiovascular disease, such as hypercholesterolemia, cigarette smoking, and hypertension fail to predict the incidence of cardiovascular disease fully. Elevated plasma homocysteine has attracted growing interest as a "new" risk factor for cardiovascular disease. Homocysteine is an amino acid, which is formed from the essential amino acid methionine. Intracellularly, homocysteine is either transsulfurated to cysteine via two vitamin B6-dependent reactions or is remethylated to methionine. In most cells, the remefhylation pathway depends both on folate and vitamin BI2. Defects in homocysteine metabolism may lead to elevation of plasma total homocysteine (tHcy). Accumulation of tHcy can possibly promote the formation of atherosclerotic plaques, or affect processes of blood coagulation. Various epidemiologic studies have already shown that elevated plasma tHcy is more often found in patients with atherosclerotic disease, in the coronary, cerebrovascular, and peripheral vessels. Some studies have shown positive associations with thrombotic disease. The studies that are presented in this thesis, aimed to find additional epidemiologic evidence for the hypothesis that elevated plasma tHcy is an independent risk factor for cardiovascular disease. We addressed various disease endpoints, with data of prospective as well as retrospective studies. The emphasis was on three questions. First, is a possible elevation of cardiovascular disease risk restricted to those with tHcy levels above a certain cutoff-point or is there a graded effect? Second, does defective remefhylation (reflected by elevation of fasting tHcy) or defective tHcy breakdown through the transsulfuration pathway (reflected by elevation of tHcy in response to methionine loading) emerge as a stronger predictor of cardiovascular disease risk? Third, how do blood levels and intake of the B-vitamins relate to tHcy and to risk of cardiovascular disease? In Chapters 2 and 3, data of a Dutch case-control study with coronary atherosclerosis as disease endpoint were presented. Study subjects were selected from men and women aged 25-65 years, who underwent coronary angiography in the Rotterdam Zuiderziekenhuis hospital, between 1992 and 1994. Cases were defined as those having > 90% occlusion in one and > 40% occlusion in an additional coronary artery. Notably, the majority of the cases had > 70% occlusion 149 Summary in a second vessel. Coronary controls were defined as those having a maximum of 50% occlusion in only one coronary artery. Most of these controls had no substantial narrowing in all three coronary vessels. In addition, a control group sampled from the general population was studied. In Chapter 2, we showed that in patients with severe coronary atherosclerosis (n=131) geometric mean fasting tHcy was 9% higher (P=0.01) and post-load tHcy was 7% higher (P=0.04) than in the combined control groups (n=189), after adjustment for age and gender differences between the groups. Fasting and post- load tHcy increased with increasing number of occluded coronary arteries (P < 0.05, tests for linear trend). The frequency distribution of fasting as well as post- load tHcy among cases was shifted towards the right across the full range of values, compared to the distribution in the combined control groups. After correction for confounding factors, the relative risk (RR) associated with a rise of 1 standard deviation [SD] of tHcy (5 umol/L for fasting and 12 umol/L for post-load tHcy) was 1.2 (borderline significant). Geometric mean levels of the B-vitamins involved in homocysteine metabolism were not lower in cases than in the combined control groups. We concluded that elevation of tHcy is an independent, graded risk factor for coronary atherosclerosis, and that other factors than inadequate vitamin levels have to be responsible for the higher tHcy levels in cases of this population. In Chapter 3, we studied whether homozygosity for a genetic defect in homocysteine metabolism would predispose to severe coronary atherosclerosis. A C-VT substitution at nucleotide 677 of the 5,10-mefhylenetetrahydrofolate reductase (MTHFR) locus results in thermolability of the enzyme, subsequently leading to reduced remethylation of homocysteine to methionine. The frequency of homozygosity for the mutation (denoted with +/+) was 10.0% in cases, not significantly different from the frequencies observed in the combined control groups (9.2%) or the separate control groups. Compared to homozygous normal subjects, (+/+) subjects had 36% higher geometric mean fasting tHcy and 25% higher post- load tHcy, whereas subjects heterozygous for the mutation had intermediate levels (P=0.001 for both tests for linear trend). Subjects with the (+/+) genotype and low erythrocyte folate (<790 nmol/L, the population median level) had 76% (95% confidence interval [CI], 27% to 144%) higher geometric mean levels of fasting tHcy, and 30% (95% CI, -2% to 70%) higher post-load tHcy levels than (+/+) subjects with erythrocyte folate levels above the median. Thus, our data showed that homozygotes for the 677C—»T mutation in MTHFR with low folate status are particularly susceptible to high tHcy levels, and may thereby have increased risks of coronary artery disease. In Chapter 4, data are shown of a case-control study in 130 Boston area 150 Summary patients hospitalized with a first myocardial infarction and 118 population-based controls, of both sexes and less than 76 years of age (Boston Area Health Study). Adjusting for age and gender, geometric mean fasting tHcy was 11% higher in cases than controls (P=0.006). Adjusting for age, gender and other potential confounders, the RR for a 3 umol/L (1 SD) increase in fasting tHcy, was 1.35 (95% CI, 1.00 to 1.82). Comparing mean levels of several homocysteine metabolites between cases and controls, it appeared that impairment of homocysteine remethylation, dependent of folate and vitamin B12, rather than vitamin B6- dependent transsulfuration, was the predominant cause of high tHcy levels in cases. Concordantly, dietary intake and plasma levels of folate were inversely related to risk of myocardial infarction and to plasma tHcy. When plotting folate intake against plasma tHcy levels, we observed that tHcy reached its nadir at an intake of 400 ug/d. To summarize, elevation of fasting tHcy was an independent, graded risk factor for myocardial infarction and folate was its most important determinant. In Chapters 5 and 6, we analyzed data of the Physicians' Health Study (PHS). The PHS is a randomized, double-blind, placebo-controlled trial of aspirin and 8-carotene in 22 071 US male physicians. A total of 14 916 subjects, aged 40 to 84 years, with no prior history of cardiovascular disease provided blood samples at baseline. We used data of a 5-year follow-up period. In the study of Chapter 5, levels of tHcy and several compounds involved in homocysteine metabolism were measured in prospectively collected plasma samples of 218 men who developed angina pectoris during the follow-up period and 218 apparently healthy control men, matched for age and smoking. Angina pectoris was accompanied by strong evidence of severe coronary atherosclerosis in all cases. The mean plasma level of tHcy was only slightly higher in cases of angina pectoris than in controls (P=0.33). There were no significant differences even for tHcy levels at the right end of the distribution. Mean plasma levels of cysteine, cystathionine, methionine, dimethylglycine, serine and glycine did not differ significantly between cases and control subjects either. Plasma tHcy showed virtually no association with risk of angina pectoris. The lack of effect may be related to possibly better nutritional status of physicians, e.g. relative to the general population. However, in a previous publication on data of the PHS, a statistically significant positive association between tHcy and risk of myocardial infarction was observed. Thus, we speculated that any effect of elevated tHcy is more likely to be thrombogenic than atherogenic. In Chapter 6, we compared tHcy in plasma samples obtained from 109 subjects who developed ischemic stroke during the follow-up period and 427 control subjects. The mean plasma concentration was slightly higher in cases than in 151 Summary controls (P=0.13). The crude RR of ischemic stroke for subjects with tHcy levels above the 80th percentile of control subjects (> 12.7 umol/L), relative to those with lower levels, was 1.4 (95% CI, 0.8 to 2.2). The RR was 1.2 (95% CI, 0.7 to 2.0) after controlling for several risk factors and other potential confounders. In subgroup analyses, elevated tHcy levels appeared to be more strongly predictive of ischemic stroke in normotensive men and in men younger than 60 years. Our findings were inconclusive, and either compatible with no association between elevated tHcy and risk of ischemic stroke, or a moderate increase in risk, in subgroups otherwise at low risk. In Chapter 7, we concentrated on gender differences in the relationship of tHcy with cardiovascular disease (not differentiated by endpoint). Data of a European, multi-center case-control study of 750 cases of vascular disease and 800 controls, of both sexes and under age of 60 years, were used. In women, adjusting for age, center, and conventional risk factors, the RRs of cardiovascular disease per 1 SD increase in tHcy were 1.7 (95% CI, 1.1 to 2.6) and 1.8 (95% CI, 1.4 to 2.4) for fasting and post-load levels, respectively. RRs did not substantially differ between pre- and postmenopausal women. In men, RRs were somewhat smaller than in women. Levels of plasma pyridoxal 5'-phosphate, but not of folate or vitamin B12, were significantly lower in cases compared to controls. Low PLP was a strong risk factor for vascular disease, largely independent of tHcy. In conclusion, we showed that elevation of tHcy is also a strong risk factor in women, both before and after menopause. In Chapter 8, study results were discussed in the light of other epidemiologic studies. We addressed several methodologic issues. Overall, in line with other findings, our studies supported the hypothesis that elevated tHcy is an independent risk factor for cardiovascular disease. In general, the effect was of similar strength for fasting and post-methionine loading tHcy. Results favored a graded relationship between elevation of tHcy and risk of cardiovascular disease, with no threshold effect. Furthermore, some of our studies pointed to the significance of impaired homocysteine remethylation in predicting cardiovascular disease risk. Accordingly, folate was found to be an important determinant of plasma (fasting) tHcy, even more so in subjects with a homozygous genetic defect in homocysteine metabolism (thermolabile MTHFR), which is associated with impaired remethylation. Dietary folate intake of a large segment of the US general population and possibly also of the Dutch general population is below 400 ug/day, the intake level at which tHcy reached its nadir in the BAHS, and in some other studies. Thus, tHcy will be generally increased in many populations. Several studies have already shown that elevated tHcy can be normalized by supplementation with folate, even at a dose of 152 Summary 650 ug/day. Therefore, increasing folate intake seems an important way of tHcy lowering in populations, possibly reducing cardiovascular disease incidence. Intervention trials of tHcy lowering with folate, studying possible reduction of cardiovascular disease, should indicate whether the intended effect can be reached and at what minimal extra intake of folate. Based on the current epidemiologic evidence, it seems advisable to increase folate intake of the general population, e.g. by increasing consumption of fruits and vegetables. The outcomes of intervention trials will indicate, whether additional measures, such as fortification of food or supplementation are justified as well. 153 Samenvatting Hart- en vaatziekten vormen een belangrijk probleem voor de volksgezondheid in Nederland en andere westerse landen. De bekende risicofactoren voor hart- en vaatziekten, zoals hypercholesterolemie, roken, en hoge bloeddruk kunnen de incidentie van hart- en vaatziekten niet volledig verklaren. Een verhoogde plasmaconcentratie van homocysteïne trekt steeds meer de aandacht als een mogelijk "nieuwe" risicofactor voor hart- en vaatziekten. Homocysteïne is een aminozuur dat gevormd wordt uit het essentiële aminozuur methionine. In de cel kan het homocysteïne getranssulfureerd worden tot cysteïne via twee vitamine B6- afhankelijk reactiestappen. Ook kan het homocysteïne weer geremefhyleerd worden tot methionine. Deze remethylering is in de meeste lichaamscellen afhankelijk van foliumzuur en vitamine B]2. Verminderde activiteit van de enzymen, die betrokken zijn in deze beide routes van het homocysteïnemetabolisme, kunnen leiden tot verhoogde plasmaconcentraties van totaal homocysteïne (tHcy). Hoge concentraties van tHcy kunnen mogelijk het proces van afherosclerotische plaquevorming bevorderen, of leiden tot een verhoogde bloedstolling. Verschillende epidemiologische onderzoeken hebben reeds gevonden dat een verhoging van het plasmagehalte van tHcy vaker voorkomt bij patiënten met afherosclerotische aandoeningen, in de coronaire, cerebrovasculaire of perifere vaten. Sommige studies hebben een positief verband uitgewezen met trombose. De onderzoeken die in dit proefschrift zijn beschreven hadden als doel aanvullende epidemiologische onderbouwing te geven aan de hypothese dat een verhoogd tHcy-gehalte een onafhankelijke risicofactor is voor hart- en vaatziekten. Verschillende ziekte-eindpunten zijn bestudeerd, met gegevens van zowel prospectieve als retrospectieve onderzoeken. De nadruk lag daarbij op drie vragen. Ten eerste, is het risico op hart- en vaatziekten alleen verhoogd bij personen met een tHcy-niveau boven een bepaald afkappunt of is er een continu verband? Ten tweede, is verminderde remethylering van het homocysteïne (weerspiegeld door verhoogde nuchtere tHcy-waarden) of verminderde afbraak via de transsulfureringsroute (weerspiegeld door verhoogde tHcy-gehalten in respons op een orale belasting met methionine) een sterkere voorspeller van het risico op hart• en vaatziekten? Ten derde, hoe is de relatie tussen de bloedspiegels en inneming van de B-vitamines enerzijds en tHcy-waarden anderzijds, en tussen de B-vitamines en het risico op hart- en vaatziekten? 155 Samenvatting In de Hoofdstukken 2 en 3 worden gegevens gepresenteerd van een patiënt- controle onderzoek met atherosclerose van de coronaire arteriën (coronairsclerose) als eindpunt. Deelnemers zijn geselecteerd uit alle mannen en vrouwen in de leeftijd van 25-65 jaar, die in de periode 1992-1994 een hartcatlieterisatie ondergingen in het Zuiderziekenhuis in Rotterdam. De personen met > 90% occlusie in één coronaire arterie en > 40% occlusie in een tweede zijn beschouwd als patiënt. De meerderheid van hen had > 70% occlusie in een tweede coronaire arterie. Eén vergelijkingsgroep werd gevormd door personen met een maximum van 50% occlusie in slechts één coronaire arterie. De meesten van hen hadden echter geen significante vernauwingen in alle drie de coronaire arteriën. Een tweede controlegroep bestond uit een steekproef van de algemene populatie. In Hoofdstuk 2 vonden we dat geometrische gemiddelden van nuchter tHcy en van tHcy na methioninebelasting respectievelijk 9% (P=0.01) en 7% (P=0.04) hoger waren bij patiënten met ernstige coronairsclerose (n=131) dan in de gecombineerde controlegroepen (n=189), na correctie voor verschillen in leeftijd en geslacht tussen de groepen. De concentratie van tHcy, nuchter en na belasting, steeg bij toename van het aantal geoccludeerde coronaire arteriën (P < 0.05, voor alle testen voor lineaire trend). De frequentieverdeling van fHcy-waarden was bij de patiënten naar rechts verschoven ten opzichte van de controlegroepen. Na correctie voor verstorende variabelen, was het relatieve risico (RR), geassocieerd met een toename van 1 standaarddeviatie (SD) van tHcy (5 umol/L voor nuchter tHcy en 12 umol/L voor tHcy na belasting) 1.2 (borderline significant). Geometrische gemiddelden van de B-vitamines waren niet lager in de groep van patiënten, vergeleken met de gecombineerde controlegroepen. Wij concludeerden dat het risico op ernstige coronairsclerose continu toeneemt met stijgende concentratie van tHcy, en dat andere factoren dan inadequate niveaus van B-vitamines de oorzaak zijn geweest voor de verhoogde tHcy waarden bij patiënten. In Hoofdstuk 3 bestudeerden we of homozygotie voor een genetisch defect in het homocysteïnemetabolisme predisponeert tot ernstige coronairsclerose. Een C—>T substitutie op nucleotide 677 van het 5,10-methyleentetrahydrofolaat reductase (MTHFR) locus leidt tot thermolabiliteit van het enzym, hetgeen verminderde remethylering van homocysteïne tot gevolg heeft. De frequentie van homozygotie voor de mutatie (aangegeven met +/+) was 10.0% in de patiënten en verschilde niet significant van de frequentie in de gecombineerde controlegroepen (9.2%), of de aparte controlegroepen. Vergeleken met de groep van homozygoot-normale personen, waren geometrische gemiddelden van nuchter tHcy en van tHcy na methioninebelasting respectievelijk 36% en 25% hoger in de (+/+) groep, terwijl de groep met de mutatie in de heterozygote vorm een tussenhggende tHcy-waarde had 156 Samenvatting (P=0.001 voor beide lineaire trendtesten). Personen met het (+/+) genotype en een laag erytrocyt-foliumzuurgehalte (< 790 nmol/L, de populatiemediaan) hadden een 76% (95% betrouwbaarheidsinterval [BI], 27% tot 144%) hoger geometrisch gemiddelde van nuchter tHcy en een 30% (95% BI, -2% tot 70%) hoger tHcy na methioninebelasting, vergeleken met (+/+) personen met een foliumzuurgehalte hoger dan de mediaan. Onze bevindingen gaven dus aan, dat personen met de 677C—>T MTHFR mutatie in de homozygote vorm, in samenhang met een lage foUumzuurstatus, bijzonder gevoelig zijn voor verhoging van het tHcy gehalte, hetgeen mogelijk een verhoogd risico op hart-en vaatziekten tot gevolg heeft. In Hoofdstuk 4 werden resultaten besproken van een patiënt-controle onderzoek uitgevoerd in Boston en omgeving. Het betrof 130 patiënten met een eerste hartinfarct en 118 populatiecontroles, zowel mannen als vrouwen, jonger dan 76 jaar ("Boston Area Health Study"). Na correctie voor leeftijds- en geslachtsverschillen, was het nuchter tHcy 11% hoger in patiënten dan in controlepersonen (P=0.006). Het RR voor een toename van het tHcy met 3 umol/L (± 1 SD) was 1.35 (95% BI, 1.00 tot 1.82), na correctie voor leeftijd, geslacht, en andere mogelijk verstorende variabelen. Bij een vergelijking van gemiddelde waarden van verschillende andere metabolieten, bleek verminderde remefhylering van het homocysteïne, en niet zozeer verminderde afbraak via de transsulfureringsroute, de voornaamste reden voor tHcy-verhoging in de patiënten te zijn. In overeenstemming daarmee, waren inneming en plasmaconcentraties van foliumzuur invers gerelateerd aan het risico op een hartinfarct en aan plasma-tHcy. Wanneer inneming van foliumzuur in een grafiek werd uitgezet tegen het plasma- tHcy, namen we waar dat foliumzuur niet verder geassocieerd was met tHcy bij een inneming hoger dan 400 ug/dag. In dit onderzoek bleek verhoging van tHcy dus een onafhankelijke, continue risicofactor voor een hartinfarct, met fohumzuur als belangrijkste determinant. In de Hoofdstukken 5 en 6 hebben we gegevens geanalyseerd van de "Physicians' Health Study" (PHS), een gerandomiseerd, dubbel-blind, placebo- gecontroleerd interventie-onderzoek naar effecten van aspirine en B-caroteen. Bij aanvang van de studie waren bloedmonsters beschikbaar van 14.916 mannelijke artsen, 40-84 jaar oud, zonder een medische geschiedenis van hart- en vaatziekten. We gebruikten gegevens van de eerste 5 jaar follow-up. In het onderzoek van Hoofdstuk 5 werden concentraties gemeten van tHcy en enkele verbindingen betrokken in het metabolisme van homocysteïne in baseline• plasma van 218 mannen, bij wie angina pectoris werd vastgesteld gedurende de follow-up, en van 218 ogenschijnlijk gezonde mannen (controlegroep), gematcht voor leeftijd en rookgewoonten. Angina pectoris ging bij alle patiënten gepaard met 157 Samenvatting objectief bewijs voor ernstige coronairsclerose. Het gemiddelde nuchter plasma tHcy was slechts ietwat hoger in patiënten met angina pectoris dan bij de controlepersonen (P=0.33). Er waren evenmin significante verschillen aan het rechter einde van de frequentieverdeling van tHcy. Gemiddelde plasmaconcentraties van cysteïne, cystathionine, methionine, dimefhylglycine, serine en glycine verschilden evenmin significant tussen patiënten en controles. Plasma-tHcy vertoonde nauwelijks een associatie met het risico op angina pectoris. De afwezigheid van een effect zou een gevolg kunnen zijn van de mogelijk relatief goede voedingstoestand van artsen. Echter, in een vorige publikatie met PHS- gegevens werd een positief verband gevonden tussen tHcy en het risico op een hartinfarct. Derhalve speculeerden wij dat een effect van verhoogd tHcy mogelijk meer trombogeen is dan atherogeen. In Hoofdstuk 6 vergeleken we tHcy-concentraties van 109 mannen die gedurende de follow-up een herseninfarct kregen met die van 427 controlepersonen. De gemiddelde plasmaconcentratie van tHcy was iets hoger bij patiënten dan controles (P=0.13). Het ruwe RR voor een herseninfarct voor personen met tHcy- waarden boven de tachtigste percentielwaarde van controles (> 12.7 umol/L), vergeleken met personen met lagere waarden, was 1.4 (95% BI, 0.8 tot 2.2). Het RR was 1.2 (95% BI, 0.7 tot 2.0), na correctie voor verschillende risicofactoren en andere mogelijk verstorende variabelen. In subgroepanalyses bleek een verhoogd tHcy gehalte een sterkere voorspeller te zijn van het risico op een herseninfarct bij normotensieve mannen en bij mannen jonger dan 60 jaar. Onze bevindingen waren niet overtuigend, en ofwel verenigbaar met geen associatie tussen verhoogd tHcy en risico op een herseninfarct, danwel verenigbaar met een licht verhoogd risico in subgroepen die anderszins een lage kans hebben op een herseninfarct. In Hoofdstuk 7 richtten wij ons op verschillen tussen mannen en vrouwen, wat betreft het verband tussen tHcy en risico op hart- en vaatziekte (niet uitgesplitst naar eindpunt). We gebruikten gegevens van een grootschalig onderzoek bij 750 patiënten met hart- en vaatziekten en 800 controlepersonen, verzameld door een aantal onderzoeksgroepen in Europa. Mannen en vrouwen, jonger dan 60 jaar, vormden de onderzoekspopulatie. Voor vrouwen was het RR voor hart- en vaatziekte per 1 SD toename in tHcy 1.7 (95% BI, 1.1 tot 2.6) voor nuchter tHcy, en 1.8 (95% BI, 1.4 tot 2.4) voor tHcy na de methioninebelasting. RRs verschilden niet wezenlijk tussen pre- en postmenopauzale vrouwen. Voor mannen waren de RRs iets lager dan voor vrouwen. Concentraties van plasma pyridoxal 5-fosfaat (PLP, de biologisch actieve vorm van vitamine B6), maar niet van erytrocyt foliumzuur of serum vitamine B]2, waren significant lager bij patiënten dan bij controles. Een laag PLP gehalte was geassocieerd met een verhoogd risico op hart- 158 Samenvatting en vaatziekte, grotendeels onafhankelijk van tHcy. Concluderend stelden wij dat een verhoging van tHcy ook een sterke risicofactor is bij vrouwen, zowel voor als na de menopauze. In Hoofdstuk 8 bespraken we de resultaten van ons onderzoek in het licht van andere epidemiologische bevindingen. Verschillende methodologische aspecten werden belicht. In het algemeen ondersteunden onze bevindingen, net als resultaten van anderen, de hypothese dat een verhoogde tHcy-concentratie een onafhankelijke risicofactor is voor hart- en vaatziekten. Het verband was ongeveer even sterk voor nuchter gemeten tHcy en tHcy gemeten na een methioninebelasting. De resultaten wezen op een dosis-respons relatie, waarbij het risico continu steeg bij toenemend tHcy. Enkele van onze bevindingen benadrukten dat verminderde remethylering van homocysteïne van betekenis is bij de associatie tussen verhoogd tHcy en hart- en vaatziekten. In overeenstemming daarmee, bleek foliumzuur een belangrijke determinant van het (nuchter) plasma tHcy, met name in personen met homozygotie voor thermolabiel MTHFR. De inneming van fohumzuur in de Verenigde Staten en waarschijnlijk ook in Nederland is bij een groot deel van de bevolking lager dan dan 400 ug/dag, het innemingsniveau waarbij tHcy genormaliseerd wordt. Derhalve zullen verhoogde tHcy-waarden veelvuldig voorkomen in populaties. Verschillende studies hebben reeds aangetoond dat verhoogde tHcy-concentraties verlaagd kunnen worden met foliumzuursuppletie, zelfs bij een dosis van 650 ug/dag. Een verhoging van de foliumzuurinneming lijkt dus een belangrijk middel om het tHcy van de bevolking te verlagen, en mogelijk daarmee de incidentie van hart- en vaatziekten terug te dringen. Interventiestudies met foliumzuur, waarbij hart- en vaatziekten het eindpunt vormen, zullen moeten uitwijzen of het gewenste effect inderdaad bereikt kan worden en bij welke extra inneming van foliumzuur. Op grond van de huidige epidemiologische gegevens lijkt het noodzakelijk om foliumzuurinneming van de algemene bevolking te doen toenemen, bijvoorbeeld door verhoging van groente- en fruitconsumptie. De gegevens van de interventie-onderzoeken zullen aangeven of het wenselijk is om additionele maatregelen te nemen, zoals verrijking van voedingsmiddelen of supplementgebruik. 159 Appendix 1 HOMOCYSTEINE AND VASCULAR DISEASE: THE EUROPEAN CONCERTED ACTION PROTECT Ian Graham, F.R.C.P.I., Leslie Daly, Ph.D., Hon. M.F.P.H.M., Helga Refsum, M.D., Killian Robinson, M.R.C.P., M.D., Lars Brattstrôm, M.D., Ph.D., Per Ueland, M.D., Roberto Palma-Reis, M.D., Ph.D., Godfried Boers, M.D., Richard Sheanan, M.R.C.P.I., Bo Israelsson, M.D., Cuno Uiterwaal, M.D., Ph.D., Raymond Meleady, M.R.C.P.I., Dorothy McMaster, Ph.D., Petra Verhoef M.Sc, Jacqueline Witteman, Ph.D., Paolo Rubba, M.D., Hélène Bellet, M.D., Jan Wautrecht, M.D., Harold de Valk, M.D., Armando Sales Luis, M.D., Ph.D., Françoise Parrot-Roulaud, M.D., Kok Soon Tan, M.R.C.P., Isabella Higgins, Danielle Garçon, Ph.D., Maria José Medrano, M.D., Ph.D., Mirande Candito Ph.D., Alun Evans, M.R.C.P., F.F.P.H.M.I., Generoso Andria, M.D. From the Department of Cardiology, Adelaide Hospital, Trinity College, Dublin and the Department of Epidemiology, Royal College of Surgeons in Ireland (I.G.,R.M.,I.H.); the Department of Public Health Medicine and Epidemiology, University College Dublin, Ireland (L.D.); the Department of Clinical Biology, Divison of Pharmacology, University of Bergen, Norway (H.R., P.U.); the Department of Cardiology, Cleveland Clinic Foundation, Ohio, USA (K.R.); the Department of Medicine, County Hospital Kalmar, Sweden (L.B.); Servicio de Medicina, Hospital de S. Francisco Xavier, Lisbon, Portugal (R.P.R., A.S.L.); the Department of Endocrinology, Katholieke University Nijmegen, Netherlands (G.B.); Division of Cardiology, University of Texas Medical Branch at Galveston, Texas, USA (R.S.); the Department of Medicine, Malmo General Hospital, Malmo, Sweden (B.I.); the Department of Epidemiology and Biostatistics, Erasmus University Medical School, Rotterdam, The Netherlands (C.U., J.W.); the Department of Medicine, The Queens University of Belfast, Northern Ireland (D.McM, A.E.); the Department of Public Health and Epidemiology, Agricultural University, Wageningen, Netherlands (P.V.); Facolta di Medicina e Chirurgia, Universita degli Studi di Napoli, Federico fi (P.R., G.A.); Laboratoire de Medicine Experimentale, Institut de Biologie, Montpellier, France (H.B.); Service de Pathologie Vasculaire, Clinique Médicale, University Libre de Bruxelles, Belgium (J.W.); the Department of Internal Medicine, University Hospital, Utrecht, Netherlands (H. de V.); Department Chromatographie, Hôpital Pellegrin, Bordeaux, France (F.P-R.); the Department of Cardiology, Toa Payoh Hospital Singapore (K.S.T.); Laboratoire de Biochimie, Faculté de Pharmacie, Marseille, France (D.G.); Instituto de Salud "Carlos III", Centro Nacional de Epidemiologia, Madrid, Spain (M.J.M.); Laboratoire de Biochimie, Hôpital Pasteur, Nice, France (M.C.). 161 Appendix 1 ACKNOWLEDGMENTS We would like to acknowledge with gratitude the intellectual and professional expertsie of the following individuals in the planning and execution of this concerted action project: Paul Boissieras, Geoffrey Bourke, Seamus Cahalane, Francois Cambien, Robert Clark, Wulf Endres, Brian Fowler, Raffaella de Franchis, Michael Gibney, Anders Green, Arne Hamfelt (and other staff of Mimelab-AB, Soraker, Sweden), Noel Hickey*, Frans Kok, Marcel Kornitzer, Jan Kraus, Vincent Maher, David McConnell, Joseph McPartlin, Giovanni di Minnio, Anne Molloy, Luis Cayolla da Motta, Eileen Naughten, Luke O'Donnel, Jesus de Pedro Cuesta, Luca Raineri, Susanna Sans, John Scott, Emer Shelley, Cairns Smith, Suzanne Storey, Pietro Strisciuglo, Leslie Taylor, Noeleen Walsh, Ursula White, Steven Whitehead, John Yarnell. Thanks are also due to Dr. P.H. Rolland, Dr. P. Ambrosi, Dr. A. Barlatier and Prof. R. Luccioni. Special thanks are due to Mrs. Betty Turner for her organisational skills. * Since this programme began Prof. Noel Hickey has died. 162 Dankwoord Vele personen hebben mij op allerlei manieren geholpen bij het voltooien van dit proefschrift. In de eerste plaats wil ik mijn promotoren Prof. Frans Kok en Prof. Meir Stampfer bedanken. Frans, in 1988 deed ik een afstudeervak Epidemiologie bij jou en mede door jouw enthousiasme werd mijn belangstelling voor dit vakgebied gewekt. Ik was dus ook zeer blij dat ik als onderzoeker in opleiding onder jouw hoede bij de vakgroep kon komen werken en heb dat in de afgelopen jaren met veel plezier gedaan. Je hebt het project op de voet gevolgd en ik kon altijd bij je terecht voor een goed advies. Bedankt ook voor je belangstelling en het vertrouwen dat je in mij hebt gesteld, waardoor de afgelopen vier jaren wetenschappelijk ten volle zijn benut. Meir, I have learned a lot from you. Thank you very much for giving me the opportunity to "do homocysteine" in so many interesting studies. Working with you was a great pleasure and I am happy I can continue doing that! Het onderzoek in Rotterdam had nooit tot stand kunnen komen zonder de enorme inzet van Annelies Legters. Annelies, je hebt ontzettend goed werk gedaan! Ongelooflijk, al die catheterisatierapporten, statussen, bloedjes, telefoontjes, zakjes metMonine en tomaten (en nog veel meer!) die je de revue hebt zien passeren.... Heel erg bedankt! Verder wil ik graag cardioloog Dick Kruyssen hartelijk bedanken voor zijn hulp bij het opzetten en uitvoeren van dat onderzoek. Ook veel dank aan de cardiologen, arts-assistenten en secretaressen van het Zuiderziekenhuis en alle andere ziekenhuizen van Rotterdam en omstreken die een bijdrage hebben geleverd aan de werving van de onderzoeksdeelnemers. Annette Bak van Rocari wil ik bedanken voor haar hulp bij het werven van de controlepersonen uit de Rotterdamse bevolking. De medewerkers van Sticares, Mariët Penning en Ria van Vliet ben ik ook erkenteüjk voor hun hulp bij het onderzoek. Ook hartelijke dank aan de overige leden van het projectteam voor het meedenken over de onderzoeksopzet, het geven van commentaar op manuscripten en verder advies: Evert Schouten, Rick Grobbee en Jacqueline Witteman. Ook Geert van Poppel en Lucy van de Vijver wil ik bedanken voor hun bijdrage aan het onderzoek. Natuurhjk ben ik ook alle deelnemers aan het onderzoek heel erg dankbaar voor hun medewerking: zonder u was het niet gelukt! 163 Dankwoord Many thanks to all the colleagues of the Channing Laboratory in Boston for their assistance in the research projects and the fun time during my (many!) visits: Kathie Schneider, Fran Grodstein, Stefanie Parker, Jing Ma, and many other people that were always willing to help me. Prof. Walter Willett, thank you for your advice during these projects and the confidence you have had in me. Special thanks are also due to Charles Hennekens, Michael Gaziano, René Malinow, Robert Allen, and all other co-authors of the "Boston papers". The inspiring atmosphere at the workshops of the EC project Hyperhomocysteinemia and Vascular Disease have made my thesis work extra special. I would like to express my gratitude to Prof. Ian Graham for giving me the chance to participate in this project and his hospitality at the workshops in Ireland. Many thanks also to Betty Turner for her help with many things regarding the project. Leslie Daly and Ray Meleady are thanked for their scientific advice. Helga Refsum, I am very grateful for all the support you gave me during the years that your laboratory analyzed our blood samples. Also, you and Per Ueland gave very helpful advice when I was preparing the papers. Thanks very much to both of you! Also thanks to many other members of the project, that made the workshops interesting and pleasant. Van de Katholieke Universiteit in Nijmegen wil ik ook een aantal mensen bedanken: Frits Boers, hartelijk dank voor je wetenschappelijke adviezen, belangstelling en de gezelligheid in Ierland. Henk Blom en Leo Kluijtmans, bedankt voor jullie wetenschappelijke inzet en de fijne samenwerking. De Nederlandse organisatie voor Wetenschappelijk Onderzoek (NWO) wil ik bedanken voor de studiebeurzen die zij mij ter beschikking heeft gesteld voor de werkbezoeken aan Boston en congresbezoeken. I would like to thank the organizations and companies acknowledged at the beginning of this thesis for their contribution to the printing costs of this book. Van alle collega's van de vakgroep is er één heel bijzonder geweest: Jantine Schuit. Tina, ontzettende parariimf, heel erg bedankt voor al je hulp en je vriendschap tijdens de afgelopen jaren. In vele opzichten ben je onmisbaar geweest, bijvoorbeeld als mijn trouwe zaakwaarnemer als ik weer eens in Boston zat. Heel veel succes bij het afronden van jouw proefschrift! Alle andere collega's van de vakgroep, maar vooral een aantal lieve lui op de zolder van "John Snow" wil ik bedanken voor hun steun in zware tijden, gezelligheid en hulp bij het een of ander. Louise, dankjewel voor al je steun, vooral in de laatste paar maanden, toen je je ontpopte als reserve-paranimf! Dirk, Siegfried, Ada, Jos, en Jessica zijn steeds bij van alles en nog wat behulpzaam geweest: bedankt! Jan Burema, hartelijk bedankt voor je statistische adviezen! 164 Dankwoord Familie en vrienden hebben mij ook steeds bijgestaan bij het onderzoek en het schrijven van dit proefschrift. Ellen, m'n andere paranimf, jij hebt het hele gebeuren stap voor stap meegemaakt en me heerlijk veel pep-talk gegeven. Dankjewel! Van de andere vrienden wil ik vooral Free, Carolijn, Corine, Marjolein, Ron, Hans en Peter bedanken voor al hun opbeurende woorden, belangstelling, gezellige dagjes en ander liefs. Huisgenoten van al die jaren, maar vooral Barbara en Nicole, wil ik bedanken voor hun geduldige oren, gezelligheid en lekkere hapjes die mij op de been hielden. Hans, je hebt een geweldig mooie omslag gemaakt voor "m'n levenswerk", juist toen je het zo druk had (op het werk én thuis). Heel lief, dankjewel! Lieve papa en mama, heel erg bedankt voor jullie belangstelling in mijn werk en alles wat jullie voor me hebben gedaan! Af en toe dachten jullie dat het nooit zou gebeuren, maar nu zijn jullie dus toch trotse ouders van nóg een doctor! Lieve zus Anne, je was me net voor, dus heb ik mooi van je kunnen afkijken hoe het allemaal moet (al valt het bijna niet te evenaren!). Bedankt voor alles! Dankjewel tante Ulla en oom Cor, jullie Heve steun heeft ook bijgedragen aan het slagen van mijn promotie-onderzoek. Muchas gracias a mama Chefín, por su interés y apoyo cariñoso. Lieve Enrique, thank you for your sweet support, those hundreds of e-mails, all the things you did to help me finish this book. And most of all: to help me keep in mind that many things in Iife are much more important than a thesis. Finishing this work means starting a new life with you, which I look forward to very much! 165 Curriculum Vitae Petra Verhoef was born in Rotterdam, the Netherlands, on September 18th, 1966. In 1984 she completed secondary school (ongedeeld VWO, Emmauscollege Rotterdam), and started her studies in Human Nutrition at the Agricultural University in Wageningen. Her majors were in epidemiology and physiology. Furthermore, she studied malnutrition in Mexican children, working six months at the Instituto Mexicano de Tecnología del Agua. In November 1990, she obtained her M.Sc.-degree in Human Nutrition. From February until December 1991, she was appointed as a teaching associate at the Department of Epidemiology and Public Health, Wageningen Agricultural University. At the same department, in January 1992, she started the Ph.D.-project described in this thesis. In the summer of that year, she attended the Annual New England Epidemiology Summer Program in Boston, U.S.A. In 1993 she was registered as a M.Sc. in epidemiology by the Netherlands Epidemiological Society. In 1993 and in 1994 she worked for several months as a research fellow at the Harvard School of Public Health, Department of Epidemiology, Boston, U.S.A. In the last two years of her Ph.D.-studies, she was a member of the Ph.D.-board of the graduate school Food Technology, Agrobiotechnology, Nutrition and Health Sciences (VLAG). In the spring of 1996 she will start working as a post-doctoral fellow at the Harvard School of Public Health, Departments of Epidemiology and Nutrition, Boston, U.S.A. 167