| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Medicine, Section of Endocrinology, Tulane University Medical School (V.F.), New Orleans, Louisiana 70112; and the Department of Pathology, University of Arkansas for Medical Sciences and the John L. McClellan Memorial Veterans Hospital (S.C.G., L.M.F.), Little Rock, Arkansas
| Abstract |
|---|
|
|
|---|
| I. Introduction |
|---|
|
|
|---|
Homocystinuria is an inherited disorder characterized by severely elevated plasma H(e). Homocystinuric children are known to develop premature vascular disease involving all major blood vessels (14 15 ). McCully (16 ) first drew attention to a possible link between elevated plasma H(e) and vascular disease, making the seminal observation that extensive arterial thrombosis and atherosclerosis commonly occurs in children with homocystinuria. Boers et al. (17 ) highlighted the association between accelerated vascular disease and moderate elevation in plasma H(e), without the other manifestations of homocystinuria. Since then, there has been considerable interest in mild HH(e) as a risk factor for coronary artery disease, stroke, and peripheral vascular disease. The understanding of the different etiologies of HH(e) is changing because of the ability to discriminate between the types of mutations present in inherited disorders, the ability to distinguish between homozygous and heterozygous mutations, and recognition of factors that modify H(e) metabolism.
A mutation resulting in a thermolabile variant of the methylene tetrahydrofolate reductase (MTHFR) enzyme is common, occurring in one-third to one-half of alleles and varying slightly with the population studied. In the homozygous state, the thermolabile variant is found in about 8% of the population (1 2 18 19 ). Mild fasting hyperhomocysteinemia has only been reported in individuals homozygous for this polymorphism (2 ) and, as discussed below, occurs only with concomitantly low folate levels. Increased cardiovascular risk is not associated with the mutation per se. However, the risk of cardiovascular disease is increased in homozygotes with concomitantly low folate levels. A substantial proportion of patients with cardiovascular disease have post methionine load HH(e), suggesting a possible defect in cystathionine-ß-synthase (CBS) action. However, mutations of CBS are relatively rare, occurring in only approximately 9% of the population (2 ).
Thus, the high frequency (2530%) of postmethionine load HH(e) that occurs in patients with cardiovascular disease again suggests that most cases of HH(e) are due to nongenetic factors. Two possible such factors are nutritional status and hormonal changes. Motulsky (2 ) highlighted a potential interaction between nutritional and genetic factors, an example of a gene-environment interaction.
The purpose of this review is to update the practicing endocrinologist on methionine-homocysteine metabolism, H(e) measurements, genetics of HH(e), and mechanisms of vascular disease in HH(e). In particular, we will highlight the evidence of interactions between the endocrine system and H(e) metabolism.
| II. Methionine-Homocysteine Metabolism |
|---|
|
|
|---|
|
Dietary and metabolically derived methionine is conjugated by ATP to form SAM. SAM serves primarily as a methyl donor to a variety of acceptors, including guanidinoacetate, nucleic acids, neurotransmitters, phospholipids, and hormones (21 ). Creatine synthesis accounts for a major portion of SAM consumption. S-adenosylhomocysteine is the byproduct of these methyl transfer reactions and is hydrolyzed to form H(e), which then starts a new cycle of methyl group transfer. In one study, red cell SAM levels were found to be low in patients with coronary artery disease (22 ). However, red cell SAM may not adequately reflect levels in other tissues, such as the liver, which may be metabolically more active (21 ).
Studies in rodents have demonstrated that SAM is both an allosteric
inhibitor of MTHFR (23 ) and an activator of CBS (24 ). Selhub and Miller
(21 ) have proposed that the ability of SAM to act as an enzymatic
effector provides a means by which remethylation and transsulfuration
can be coordinated. When cellular SAM concentration is low, CBS will be
suppressed, resulting in increased remethylation of H(e) for methionine
synthesis. Conversely, when SAM concentration is high (as occurs after
a methionine load), inhibition of methionine synthase is accompanied by
diversion of H(e) through the transsulfuration pathway by stimulation
of CBS. Figure 2
illustrates the
hypothetical regulation of the metabolic pathways by SAM.
|
| III. Nomenclature and Methodology in the Measurement of Plasma H(e) |
|---|
|
|
|---|
Homocysteine is the reduced (sulfhydryl) form, and homocystine is the oxidized (disulfide) form of the homologs, cysteine and cystine. For the purpose of this review, both forms of "homocyst(e)ine" will be referred to as the H(e) and hyperhomocyst(e)inemia will be referred to as HH(e). There is confusion between the American and European literature in the abbreviations used: H(e) is abbreviated as Hcy or tHcy in the European literature and H(e) in the American literature. It may be important to achieve consensus in nomenclature and use of abbreviations in this field. In patients in whom H(e) levels are normal, about 7080% of the total H(e) is bound to protein by a disulfide linkage. With elevated H(e) levels, the percentage of H(e) in the sulfhydryl form represents an increasing percentage of the total H(e) concentration and can increase to 1025% of the total H(e) (26 ).
The total H(e) is measured as the free thiol, which is obtained by reduction. This is accomplished by treatment with reducing agents such as sodium borohydride, butylphosphine, or monobromobimane (25 26 27 28 ). Methods used to assay H(e) include gas chromatography with mass spectroscopy, HPLC with or without fluorescence detection, and HPLC with electrochemical detection (25 26 ). Methods with HPLC coupled to electron capture detectors do not require derivitization (29 ). There have been new developments for measuring H(e) that will allow more laboratories to measure these metabolites using immunological analyses, including an enzyme-linked immunoassay and an automated fluorescence polarization analyzer (IMX Abbott Diagnostics, Chicago, IL) (30 31 ). Quality controls for the standardization of plasma H(e) measurements are not widely available. Coefficients of variation (CV) for intraassay range between 2 and 8% and for interassay the CVs are between 2 and 10%. Studies on a small population for 4 weeks have shown that within-person variance for a 30-month period showed a high reliability coefficient, but the value was within the accepted range for the most commonly measured chemistry analytes (32 ).
A. Methionine load test
The methionine load test (MLT) is essential in the comprehensive
assessment of HH(e) because heterozygotes for CBS deficiency have
abnormal methionine load test results in the setting of normal fasting
H(e) levels (16 ). Conclusive evidence that an abnormal MLT represents
an abnormality of the transulfuration pathway is lacking in humans.
However, this hypothesis is supported by the data of Dudman et
al. (33 ) demonstrating decreased CBS activity in most subjects
with an abnormal MLT. Bostom et al. (34 ) have emphasized the
importance of the MLT in diagnosing HH(e) in patients with vascular
disease. A large proportion of such patients have normal fasting plasma
H(e) with an abnormal MLT; and the rate of detection of HH(e) increases
significantly when both fasting and postload HH(e) levels are assessed
in a study population. The MLT is performed after an overnight fast;
blood samples are collected immediately and 28 h after a 100 mg/kg
methionine load. An abnormal load test results in a peak plasma H(e)
level more than 2 SDs above normal control levels.
| IV. Determinants of Plasma Homocysteine |
|---|
|
|
|---|
Positive correlations have also been found between plasma H(e) and uric acid and creatinine concentrations that may be related to the links between H(e) metabolism with those of creatinine and uric acid (35 ). Plasma albumin concentration also correlates with plasma H(e) and may reflect an increase in protein-bound H(e). The exact significance of protein binding of H(e) with respect to cardiovascular disease is unknown.
In the Hordaland H(e) study, elevated plasma H(e) was associated with male gender, increasing age, smoking, hypertension, elevated cholesterol, and lack of exercise (40 ). In a multivariate analysis, Malinow et al. (41 ) demonstrated that systolic blood pressure, plasma uric acid, and hematocrit were predictors of concentrations of plasma H(e) in men who did not have a history of atherosclerotic disease.
It is possible that H(e) in plasma is an "acute phase reactant," rising after vascular injury. Plasma H(e) concentrations rise acutely immediately after a stroke and then decrease over several weeks (42 ). In contrast, plasma H(e) concentrations tend to be lower immediately after a myocardial infarction (MI) than 6 weeks later (43 ). The reason for this discrepancy is not clear.
Several other disease states and medications also cause elevations in
plasma H(e). The recently described association between HH(e) and
diabetes mellitus is described in detail below. These associated
factors are outlined in Table 1
.
The role of genetic mutations, acute events, nutritional status, and
hormonal effects are discussed below.
|
B. Genetics of hyperhomocysteinemia
Numerous enzyme mutations associated with HH(e) have been
described including 17 CBS mutations (reaction 1) and 10 MTHFR point
mutations (reaction 3a). One of the mutations for MTHFR, a common
polymorphism that is present in one-third to one-half of alleles,
results in a thermolabile variant of the MTHFR enzymes (1 ). The enzyme
5-methyltetrahydrofolate-homocysteine methyltransferase (reaction 3b)
(methionine synthase) has been shown to contain one common
polymorphism, but no correlation has been found between H(e) levels and
genotype. Mutations of the cobalamin coenzyme synthesis enzymes (Cb C,
D, E, F, or G) (20 ) that impair the formation of the cosubstrate
methyl-B12 (reaction 3b) are also rarely involved in HH(e).
A comprehensive review of cobalamin coenzyme synthesis enzyme mutations
is available elsewhere (44 ). The characteristics of CBS and MTHFR
mutations are summarized in Table 2
.
Functional mutations are defined as those associated with increased
H(e) levels. Whether any of these mutations per se directly
increase the risk of arterial or venous occlusive disease remains an
area of debate. Most reported studies have not evaluated the genotype
of study subjects and have only correlated elevated H(e) levels with
risk of vascular occlusive disease. To determine whether genotype is
related to the risk of vascular disease, future studies need to
correlate genotype, vitamin status (folate, vitamin B6,
pyridoxal phosphate, vitamin B12), H(e) level, and vascular
events. These studies may eventually show that genotype does not
directly contribute to the risk of vascular disease. Rather, by
increasing vitamin requirements, genotype may indirectly affect the
risk of vascular disease.
|
The CBS gene has been assigned to the subtelomeric region of band 21q22 (46 ). CBS deficiency is inherited in an autosomal recessive pattern, resulting in homozygous (homocystinuria) and heterozygous (hyperhomocysteinemia) carriers (16 ). Sequencing of the cDNA for the CBS gene has to date identified 17 mutations (47 48 ). Tsai et al. (49 ) have characterized three of the more common mutations as either a G919A, a T833C, or a C341T transition. The first two mutations account for 50% of affected CBS alleles (50 ). Different populations demonstrate differing mutation frequencies, with the G919A transition occurring in 70% of an Irish cohort while the T833C transition occurred in 50% of a Dutch population (1 ). Tsai et al. (45 ) have estimated that the heterozygote frequency for a CBS point mutation is 1/20,000 to 1/200,000, and that 3040% of individuals with premature vascular disease are heterozygous for CBS point mutations.
An additional mutation described by Sebastio et al. (51 ) involves a 68-bp insertion in the coding region of exon 8 of the CBS gene. This insertion mutation, which creates an alternate splice site at the intron 7-exon 8 border, has only been reported in combination with the T833C missense mutation; the latter is located in cis 10 bp upstream of the insertion (46 49 50 51 52 53 ). In addition, the insertion sequence contains a premature stop codon; however, Tsai et al. (52 ) reported finding only normal size RNA, implying either that splicing did not introduce the premature stop codon, or that the truncated RNA was not detectable. This insertion results in a benign mutation because the T833C missense mutation is eliminated through alternate splicing at the intron 7-exon 8 border (within the insertion there is no substitution of the 833 nucleotide). The prevalence of this mutation varies with the population studied. Tsai et al. (52 ) reported a prevalence of 11.7% in a control population (heterozygotes); the double mutation has been reported to occur in 25.8% of Northern Italians.
Another novel point mutation has been described by Kluijtmans et al. (54 ) in a partially vitamin B6-responsive homocystinuric patient. This mutation was a G1330A transition and was unique in that it abrogated CBS responsiveness to SAM. Thus, in contradistinction to the other 17 identified point mutations that result in an altered protein that attenuates the catalytic activity of CBS, this mutation interferes with the regulatory domain of the CBS protein. (54 )
2. MTHFR deficiency. MTHFR catalyzes the reduction of
5,10-methylenetetrahydrofolate to 5-methyltetrahydrofolate (reaction
3a, Fig. 1
). The gene is located on chromosome 1p36.3; 10en different
mutations for the MTHFR gene have been identified from isolated cDNA
(55 ). Nine of these mutations result in thermostabile mutations. The
two most common of these mutations are a C559T transition,
which converts an arginine codon to a termination codon, and a
G482A transition, which converts an arginine to a glutamine
residue (55 ). Another mutation in the MTHFR gene, a C677T
transition, results in a thermolabile variant (56 57 58 ) of the enzyme.
This autosomal recessive mutation creates a HinfI
restriction site and substitutes an alanine for a valine residue in the
MTHFR protein (57 ). The allele frequency of this polymorphism varies
slightly with the population studied, but approximately 8% of studied
populations are homozygous for this autosomal recessive polymorphism.
The homozygous thermolabile MTHFR polymorphism is characterized by an
enzyme activity of about 30% of normal. Heat inactivation at 46 C
distinguishes the mutant from the normal MTHFR enzyme. The enzyme is
considered thermolabile when there is less than 20% residual activity
after heating to 46 C. HH(e) does not occur in individuals heterozygous
for this polymorphism and appears only to occur in homozygotes who are
concomitantly folate deficient (2 18 56 57 58 59 60 61 62 ). Some data suggest that
homozygotes for this mutation may actually have a higher folate
requirement (59 60 61 ).
3. Methionine synthase. Methionine synthase, localized to chromosome 1q42.343 (63 ), catalyzes the transfer of a methyl group from 5-methyltetrahydrofolate to H(e) via the intermediary methyl-B12. Methionine synthase contains one common polymorphism that results in a A2756G transition. This polymorphism does not correlate with H(e) level and does not appear to be a risk factor for vascular occlusive disease, nor neural tube defects (64 ). Other rare clinical conditions have reported functional methionine synthase deficiency. These have resulted from mutations in cobalamine coenzyme synthesis that has resulted in abnormal methyl B-12 production, a cosubstrate in the methionine synthase reaction (44 64 65 66 67 68 69 ). No discrete mutations of methionine synthase itself that are associated with hyperhomocysteinemia have been described.
4. Does MTHFR polymorphism increase the risk of MI? Data conflict over the risk of vascular occlusive disease and hyperhomocysteinemia in patients with the homozygous thermolabile polymorphism for MTHFR (1 ). In contrast, no increased risk for vascular occlusive disease or hyperhomocysteinemia is present in heterozygotes for this polymorphism. Kluijtmans et al. (1 ) screened 60 cardiovascular patients and 111 controls and found that 15% of the cardiovascular patients vs. 5% of controls were homozygous for the thermolabile MTHFR polymorphism (1 ). This translated into a 3-fold risk of premature cardiovascular disease for the homozygous polymorphism. Kluijtmans and co-workers did not evaluate for folate levels in their study. In contrast, Ma et al. (19 ) in the Physicians Health Study reported that the thermolabile MTHFR polymorphism in homozygotes was associated with hyperhomocysteinemia only if folate levels were concomitantly low (19 ). An increased risk for MI, solely on the basis of the homozygous polymorphism, was not found.
In another study by Legnani et al. (18 ), the presence of the homozygous polymorphism did not increase the risk of thrombosis over control patients (18 ). Christensen et al. (62 ) showed that the incidence of the homozygous polymorphism and a normal genotype was not different between patients with coronary artery disease and healthy controls, but patients with the homozygous polymorphism did have higher H(e) levels if the serum folate levels were below the median value. Jacques et al. (61 ) reported that homozygotes for the polymorphism had H(e) levels 24% greater than those with normal genotype if the serum folate levels were <15.4 nmol/liter. Malinow et al. (59 ) suggested that homozygotes with the polymorphism may have an increased folate requirement, and Ali et al. (60 ) reported that serum folate levels are lower in individuals homozygous for the MTHFR polymorphism. The recent large prospective study by Folsom et al. (12 ) provides additional evidence against the MTHFR mutation being associated with cardiovascular disease (see below for details).
Taken together, these studies suggest that homozygotes may have an increased folate requirement, and that in the presence of normal folate levels homozygotes are not at increased risk for hyperhomocysteinemia or vascular occlusive disease.
C. Nutritional
1. Folate. Plasma H(e) is thus a sensitive biomarker of folate
deficiency. Lewis et al. (70 ) demonstrated that in subjects
with plasma folate concentrations above 15 nmol/liter the H(e)
concentration is on a low, normal plateau. At lower levels of plasma
folate, the plasma H(e) concentration increases steadily (38 ). Rimm
et al. (71 ) reported in a large prospective study of women
that the 20% who had the highest consumption of folate (almost all of
whom consumed multivitamin tablets) had significantly less
cardiovascular disease than the lowest 20% (very few of whom consumed
multivitamins). However, there have been no randomized control trials
of the effect of folic acid alone on cardiovascular disease.
There is controversy about the exact amount of supplemental folic acid that is required to reduce plasma H(e). Shimakawa et al. (72 ) reported that people who use multivitamin supplements have significantly lower plasma H(e) concentrations than nonusers. The RDA for folate in the United States is 200 µg/day. It has been suggested that an intake of 400 µg of folic acid above the dietary level will prevent birth defects. Such an increased supplementation will also significantly decrease plasma H(e) concentrations in most of the population (32 ).
Recently, Malinow et al. (73 ) reported that breakfast cereal fortified with 400 µg of folic acid was adequate in lowering plasma H(e) by more than 100 µg. Increasing the dose of folate supplementation to 600 µg had very little additional effect.
Schorah et al. (74 ) carried out a randomized double-blind placebo-controlled study in 119 healthy volunteers, whose intake of fortified or supplemental folic acid was low (74 ). Volunteers were randomized to receive unfortified cereals, or cereals fortified with 200 µg of folic acid per portion, with or without other vitamins, for up to 24 weeks. There were no significant changes in plasma H(e) in those eating unfortified cereals. Folic acid fortification of cereals led to significant increases in serum folate (66%), and red cell folate (24%), and a decrease in plasma H(e) (10%) (74 ). There were no changes in vitamin B12 or cysteine. The H(e) decrease was primarily seen in those who initially had the highest plasma H(e) or the lowest serum folate. Thus, if H(e) is found to be a causative risk factor in occlusive vascular disease, food fortification with physiological levels of folic acid should have a significant impact on the prevalence of the disease in the general population.
de Bree et al. (75 ) evaluated possible inconsistencies between recommended, actual, and desired folate intake in European adult populations. They concluded that in Europe, mean dietary folate intake in adults is 291 µg/day (range 197326) for men and 247 µg/day (range 168320) for women. The recommended intakes vary between 200300 µg/day (men) and 170300 µg/day (womenwith higher recommended intakes during pregnancy). The mean dietary folate intake in Europe is in line with recommendations, but the desired dietary intake of more than 350 µg/day to prevent an increase in plasma H(e) levels is only reached by a small part of studied European populations (76 ).
A meta-analysis of randomized trials with folic acid (76 ) reported that supplementation of the typical Western diet with 0.55 mg folic acid along with 0.5 mg vitamin B12 would be expected to reduce plasma H(e) concentrations by one fourth to one third (e.g., from about 12 µmol/liter to 8 to 9 µmol/liter). Whether such a "population approach" to prevention of HH(e) and associated cardiovascular disease will be effective requires further investigation.
2. Vitamins B6 and B12. It is well recognized that pyridoxal-5'-phosphate, the active form of vitamin B6 pyridoxine, is an important cofactor in the conversion of H(e) to cystathionine. The enzyme involved in this reaction is CBS. Because vitamin B6 is not involved in remethylation, pyridoxine deficiency will only result in HH(e) when the transsulfuration pathway is activated such as after a methionine load. Thus, the positive MLT that was designed to detect heterozygous CBS defects may result from vitamin B6 deficiency. Rats fed vitamin B6-deficient diets for 4 weeks develop HH(e) after a methionine load (77 ). However, clinical studies have demonstrated a lack of a relationship between plasma H(e) and vitamin B6 status. In human subjects made B6 deficient by diet (78 ), fasting plasma H(e) levels remain normal. In patients with asthma treated with theophylline (a vitamin B6 antagonist), plasma H(e) was significantly higher after a methionine load when compared with controls (79 ). This abnormality was corrected by treatment with pyridoxine supplementation for 6 weeks. Thus, pyridoxine deficiency should be excluded in patients who have HH(e) after a methionine load a common abnormality in patients with premature cardiovascular disease.
The importance of vitamin B12 in the remethylation of H(e) to methionine is well recognized, and HH(e) is a feature of vitamin B12 deficiency (19 39 ). However, the relationship between vitamin B12 intake/plasma levels and HH(e)-related cardiovascular disease is less well defined. In the Framingham study, plasma H(e) exhibited a strong inverse association with plasma folate but a weaker association with plasma vitamin B12. Subjects in the lowest decile of plasma B12 had significantly higher plasma H(e) when compared with those in the highest decile (80 ). Homocysteine was also inversely associated with intakes of folate and vitamin B6, but not vitamin B12. Many of the case control studies of HH(e) in patients with vascular disease have excluded subjects with vitamin B12 deficiency. As discussed in the treatment section below, treatment of patients with HH(e) with vitamin B12 seems to have very little impact on plasma H(e).
D. Hormones and H(e) metabolism
Studies have been done to clarify the role of hormonal changes in
the regulation of H(e) metabolism and in causing HH(e). As indicated
above, genetic abnormalities do not fully explain the relatively high
prevalence of HH(e) in patients with vascular disease. This is
particularly true of post methionine HH(e), which is not often
associated with genetic mutations of the CBS gene. Dudman et
al. (81 ) suggested that depressed CBS activity in several of their
patients was associated with causes other than abnormal CBS protein. A
reasonable alternative explanation for the depressed CBS activity would
be that it was metabolically down-regulated. Such down-regulation of
CBS as well as other enzymes in H(e) metabolism may be mediated by
hormonal changes.
1. Estrogen. There is some data to suggest that estrogen has an effect on H(e) metabolism, although the mechanism involved is not clear. Plasma H(e) in pregnant women decreases to almost half that in nonpregnant women (82 83 ), suggesting a possible effect of estrogen, although other factors such as increased vitamin intake obviously play a role.
Estrogen may have an effect on the activity of some of the enzymes in H(e) metabolism (84 ), although further studies are needed to confirm this effect.
Several investigators have examined the effect of menopausal status on plasma H(e). Wouters et al. (85 ) measured fasting and postmethionine plasma H(e) concentrations in premenopausal and postmenopausal healthy women without a history of vascular disease. Fasting and postmethionine plasma H(e) was significantly higher in postmenopausal women as compared with premenopausal women. The difference appears to be too large to be explained as an effect of age alone [the increase in plasma H(e) over a decade of life is modest] and is more likely to be related to hormonal status. In premenopausal women, postmethionine plasma H(e) was negatively and significantly correlated to serum 17ß-estradiol. Andersson et al. (86 ) reported that levels of fasting and postload plasma H(e) in premenopausal women did not differ from those of men of similar age. In contrast, in postmenopausal women, the level of fasting plasma H(e) was actually lower than that of men of similar age. This was because, in that study, fasting values in men increased with age and was associated with a decrease in serum vitamin B12, folate, and pyridoxal 5-phosphate.
The rise in H(e) levels after menopause may partly explain the sharp rise in cardiovascular disease that occurs in this age group, and its attenuation by hormone replacement therapy (HRT). However, many cardiovascular risk factors improve with estrogen and further investigation is required to determine whether the lower incidence of vascular disease in premenopausal women and in women taking HRT may be related to the lower concentrations of plasma.
In a prospective study, van der Mooren et al. (87 ) measured fasting serum H(e) during HRT in postmenopausal women. The mean serum H(e) decreased by approximately 11% with a greater decrease (17%) in those women who had a high H(e) level before treatment with very little change in those who had a low level. There are well recognized mechanisms to explain the beneficial effect of HRT on cardiovascular risk, including effects on lipids and fibrinolysis. However, it is possible that a lowering of plasma H(e) also contributes to this benefit.
Oral contraceptive agents, in contrast to HRT, do not affect biochemical folate indices and H(e) concentrations in young women (88 ).
Tamoxifen, an estrogen antagonist with partial agonist activity, decreased plasma H(e) by a mean of 30% after 912 months treatment in postmenopausal women with breast cancer (89 ). These changes were independent of the tumor burden. These data, in combination with the effect of estrogen, suggest that there is an estrogen receptor-mediated H(e)-lowering effect. In addition, there may be an indirect effect related to a modest elevation in plasma folate concentrations with tamoxifen (90 ).
2. Testosterone. Zmuda et al. (91 ) studied the effect of supraphysiological doses of testosterone on fasting H(e) in normal male weight lifters. Plasma H(e) levels were not significantly altered where testosterone was given alone or together with testolactone. It is likely, therefore, that short-term, high-dose testosterone administration does not affect fasting plasma H(e) levels in normal men. However, the study of Zmuda et al. has limitations in that only fasting plasma H(e) was measured and, therefore, the authors may have underestimated an effect of testosterone on postmethionine H(e).
Since testosterone has a negligible effect on plasma H(e) concentrations, an alternative explanation is needed for the difference in plasma H(e) concentrations between men and women. Another possible explanation for this gender difference is that creatine-creatinine production is directly coupled to S-adenosylhomocysteine generation from SAM (89 ). Lean body mass, muscle mass, amino acid turnover, and creatine-creatinine production tend to be higher in men than in women, all of which may explain in part the higher plasma H(e) in men. Plasma H(e) concentrations correlate directly with serum creatinine concentrations in men and women (90 ). In fact, Brattstrom et al. (92 ) demonstrated that the gender difference in plasma H(e) disappeared when men and women were matched for serum creatinine concentrations.
In male-to-female transsexuals, plasma H(e) decreased significantly while it increased in female-to-male transsexuals (93 ). Thus, the difference in sex steroid milieu between men and women may be an important factor in determining the sex difference between men and women in plasma H(e) levels. Interpretation of these data, however, is complex because of the major changes in sex hormone status occurring in transsexuals. In this context, however, it may be relevant that when inbred adult male rats were treated with estrogen, the plasma H(e) concentration fell by approximately 30%, indicating a direct effect of estrogen on lowering plasma H(e) (93 ). Plasma H(e) concentrations are also lowered in male rats treated with cortisol, estradiol, or a combination of both (94 ).
3. Thyroid hormones. An elevated plasma H(e) in hypothyroidism has previously been reported (95 ). In a recent study, Nedrebo et al. (96 ) demonstrated that plasma H(e) was significantly higher in patients with hypothyroidism (96 ). In contrast, plasma H(e) in hyperthyroid patients did not differ significantly from that of controls (96 ). Thus, HH(e) may exacerbate the increased risk of cardiovascular disease in hypothyroidism that is traditionally attributed to lipid changes.
| V. Homocysteine and Diabetes Mellitus (DM) |
|---|
|
|
|---|
Munshi et al. (98 ) reported that HH(e) after a methionine load occurred in approximately 40% of patients with type 2 DM who had macrovascular disease, but was normal in patients with insulin-dependent diabetes (type 1 DM). Patients with overt nephropathy had been excluded from that study, and only patients under the age of 60 were studied. In contrast, Araki et al. (101 ) found that fasting plasma H(e) was elevated in some patients with type 2 DM, and this abnormality was corrected with vitamin B12 injections (101 ). It is possible that these investigators had identified, by chance, patients with concomitant diabetes and B12 deficiency. In a population-based study of HH(e) and the risk of cardiovascular disease, 631 patients were stratified according to age, sex, and glucose tolerance (100 ). The authors also investigated the combined effect of HH(e) and DM with regard to cardiovascular disease. Fasting HH(e) was seen in 25.8% of individuals. After adjustment for age, sex, hypertension, hypercholesterolemia, diabetes, and smoking, the odds ratios (ORs; 95% confidence intervals) per 5 µmol/liter increment in H(e) were 1.44 (1.101.87) for peripheral arterial, 1.25 (1.031.51) for coronary artery, 1.24 (0.971.58) for cerebrovascular, and 1.39 (1.151.68) for any cardiovascular disease. After stratification by glucose tolerance category and adjustment for the classic risk factors and serum creatinine, the ORs per 5 µmol/liter increment in H(e) for any cardiovascular disease were 1.38 (1.031.85) in normal glucose tolerance, 1.55 (1.012.38) in impaired glucose tolerance, and 2.33 (1.114.90) in non-insulin-dependent diabetes mellitus (P = 0.07 for interaction)(100 ). Thus, the magnitude of the association between HH(e) and cardiovascular disease is stronger (1.6-fold) for patients with type 2 diabetes than in nondiabetic subjects. Subjects in that study were 50- to 75 yr old, and these findings may relate to an interaction of age, insulin resistance, and diabetes on both H(e) metabolism and cardiovascular disease.
Type 2 DM and HH(e) are both associated with increased lipid peroxidation (oxidative stress) (102 103 ). In a study to determine whether the coexistence of elevated H(e) levels stimulate oxidative stress further than that caused by diabetes alone, plasma concentrations of thiobarbituric acid reactive substances (TBARS), an index of lipid peroxidation, were measured in patients with type 2 DM. Plasma TBARS concentrations were elevated in diabetics with vascular disease. The additional presence of hyperhomocysteinemia was not associated with a further increase in plasma TBARS concentrations (104 ). Thus, diabetes maximally stimulates oxidative stress, and any further acceleration of vascular disease in patients who have coexistent hyperhomocysteinemia is mediated through mechanisms other than lipid peroxidation.
Other studies of patients with type 1 DM have confirmed that plasma H(e) levels are normal early in the course of the disease. However, Hultberg et al. (105 ) reported that basal plasma H(e) concentrations were higher in patients who developed nephropathy and had an elevated plasma creatinine. These workers recently demonstrated that diabetic patients with the lowest age at onset and poorest metabolic control were most prone to a rapid increase in plasma H(e) (106 ). They concluded that this increase in plasma H(e) could at least partially be explained by marginal deficiency of blood folate concentrations.
Hofmann et al. (107 ) reported that plasma H(e) levels, both fasting and after a methionine load, were elevated in patients with type 1 DM who had microalbuminuria and were higher still in patients who had overt proteinuria. The patients with type 1 DM and HH(e) also had higher plasma thrombomodulin (TM) levels (indicating endothelial cell damage) and a higher prevalence of late diabetic complications including macrovascular disease.
These workers also reported a significant relationship between plasma
H(e) concentrations and urinary albumin excretion rate (Fig. 3
), as well as a significant relationship
between plasma H(e) and plasma TM (Fig. 4
). Thus, HH(e) represents an additional
cardiovascular risk factor in patients with microalbuminuria, perhaps
contributing to the enhanced risk of cardiovascular disease in this
subpopulation of people with diabetes.
|
|
|
Three recent reports have confirmed the association between plasma H(e) and albumin excretion rate in patients with DM. Chico et al. (110 ) measured fasting plasma H(e) in 165 diabetic patients and control subjects. Patients with type 2 DM had higher plasma H(e) than controls, whereas patients with type 1 DM did not. Univariate correlations and multiple regression analysis showed albumin excretion rate to be strongly related to plasma H(e) (110 ). In addition, patients with type 2 DM with hypertension had higher plasma H(e) than patients without hypertension (110 ). This finding may have been related to the fact that the patients with hypertension had more severe biochemical markers of nephropathy, with a higher albumin excretion rate.
Lanfredini et al. (111 ) recently studied the relationship between homocyst(e)inemia and microalbuminuria in non-insulin-dependent diabetes mellitus (NIDDM) patients. There was a significant correlation between urinary albumin excretion and fasting and postmethionine load plasma H(e) in NIDDM patients. Microalbuminuric NIDDM patients had higher fasting plasma H(e) than normoalbuminuric patients. These investigators also reported that patients with NIDDM and HH(e) had higher diastolic and mean arterial blood pressure than those with a normal plasma H(e). There was a significant correlation between plasma folate and mean arterial pressure (112 ).
Hoogeven et al. (113 ) recently reported a relationship between serum H(e) level, protein intake, and risk of microalbuminuria. In a population-based study of 680 subjects stratified according to age, sex, and glucose tolerance, serum total H(e) was positively associated with the presence of microalbuminuria independent of other risk factors including diabetes, hypertension, protein intake, and renal function. For each 5 µmol/liter increase in serum H(e), the risk of microalbuminuria being present increased by about 30% (114 ). The authors suggested that hyperhomocysteinemia may partly explain the link between microalbuminuria and the increased risk of cardiovascular disease. However, it is important to recognize that microalbuminuria is associated with several other cardiovascular risk factors, all of which could contribute to increased cardiovascular disease. Nevertheless, in a recent study, Stehouwer et al. (114 ) demonstrated that plasma H(e) predicts mortality in patients with NIDDM, with or without albuminuria. Further investigation is required to determine the exact role that HH(e) plays in worsening cardiovascular disease in patients with microalbuminuria and overt proteinuria.
A. Hyperhomocysteinemia, renal failure, and diabetic nephropathy
The effect of renal disease on H(e) metabolism has been
comprehensively reviewed (115 ). This subject is of considerable
importance to endocrinologists and nephrologists who treat patients
with diabetes, as they are at high risk of developing renal failure.
Studies to determine whether the treatment of HH(e) in patients with
renal insufficiency will reduce morbidity and mortality are thus of
considerable importance to physicians treating patients with diabetes.
Recognition of the importance of renal impairment and proteinuria in determining plasma H(e) is vital to the clinician in interpretation of laboratory results. While some of this elevation may be due to decreased clearance of H(e), other mechanisms may also be responsible. Intact kidneys have considerable H(e)-metabolizing capacity (116 ), impairment of which may be an important determinant of the marked HH(e) frequently observed in end-stage renal disease. High-dose multiple vitamin treatment has been shown to lower plasma H(e) in dialysis patients, although levels remain elevated in many patients (117 ). Hyperhomocysteinemia has also been demonstrated after successful renal transplantation (118 ) and may be exacerbated by cyclosporin (119 ).
B. Effect of glucose and insulin on H(e) metabolism
It is well recognized that insulin has important effects on
protein and amino acid metabolism. However, its effect on H(e)
metabolism has not been well studied.
During a hyperinsulinemic euglycemic clamp, the plasma H(e) response to
acute hyperinsulinemia was heterogenous (120 ). Plasma H(e) levels fell
by approximately 20% in normal subjects but did not do so in
insulin-resistant patients with type 2 DM (Fig. 5
) (121 ). These data
suggest that resistance to the effects of insulin on glucose disposal
may be associated with resistance to the suppressive effect of insulin
on H(e) levels in patients with type 2 DM. Such a resistance to
insulins effect on H(e) may contribute to the increased
cardiovascular disease associated with the insulin resistance syndrome
and type 2 DM (121 ).
It is well recognized that insulin decreases plasma methionine, the methionine being incorporated into newly synthesized protein (122 ). Plasma amino acid concentrations, including methionine, fall significantly after an oral glucose load and the subsequent rise in endogenous plasma insulin (123 ). However, in patients with diabetes, this fall in amino acids does not occur, indicating a possible resistance to insulins effect on amino acids in diabetics (123 ).
The insulin-induced fall in plasma methionine concentrations may be mediated through increased tissue uptake of methionine or metabolism via the transsulfuration pathway resulting in increased levels of H(e). Intracellular and plasma H(e) levels may then rise, particularly if a defect in CBS action exists. Further study is required to determine the effect of insulin on SAM, the key determinant of the relative activity of the transmethylation and transsulfuration pathways (see above).
A recent study has demonstrated an inverse relationship between plasma H(e) and insulin sensitivity in women with preeclampsia (124 ). However, because of the multiple abnormalities in metabolism and endothelial function present in preeclampsia, it is impossible to determine whether a cause-effect relationship exists between the two variables. Further work is needed in other insulin-resistant states to determine whether insulin resistance causes HH(e).
To examine the effects of hyperinsulinemia induced by a high-fat
sucrose (HFS) diet on H(e) metabolism, we measured hepatic mRNA and
activity of two key enzymes involved in this metabolic pathway: MTHFR
and CBS, in an insulin-resistant rat model (125 ). Fischer rats made
insulin resistant by a HFS diet were examined at 6 months and 2 yr of
age and compared with control rats fed a low-fat, complex-carbohydrate
(LFCC) diet. At the end of 6 months, the HFS-fed rats were heavier than
the LFCC rats and had hyperinsulinemia. The plasma H(e) concentrations
were elevated in the HFS-fed rats (10.77 ± 0.9 vs.
6.89 ± 0.34 µmol/liter; P < 0.01). Hepatic CBS
mRNA and enzyme activity was significantly lower in the HFS group
compared with control. In contrast, hepatic MTHFR enzyme activity and
mRNA levels were significantly elevated in the HFS group compared with
control. There were significant positive correlations between plasma
H(e) and fasting plasma, body weight, and MTHFR activity. There were
significant negative correlations between plasma insulin and CBS
activity and between CBS and MTHFR activities. The latter inverse
relationship supports the hypothesis of Selhub and Miller (21 ) (Fig. 2
)
and suggests that insulins effect on H(e) metabolism may be mediated
through SAM. In conclusion, HFS feeding leads to
hyperinsulinemia, which may be associated with hyperhomocysteinemia,
secondary to down-regulation of CBS message and activity, and may thus
contribute to the accelerated macrovascular disease associated with
type 2 diabetes. However, we wish to emphasize that the above study was
carried out in rodents, and the conclusions may not be applicable to
humans.
The effect of drug treatment of diabetes on H(e) metabolism has not been well studied. Evidence suggests that insulin and sulfonylurea treatment do not alter plasma H(e) (101 ). Hoogeveen et al. (126 ) reported that metformin does not increase plasma H(e). However, in a clinical trial of metformin to assess its lipid-lowering effects in nondiabetic patients with coronary artery disease, there was a moderate but significant increase (7.2% at 12 weeks and 13.8% at 40 weeks) in plasma H(e) with metformin treatment. This was associated with a significant fall in serum B12 levels (127 ).
In summary, HH(e) is not uncommon in patients with diabetes and may
play a role in the accelerated cardiovascular disease in these
patients. The mechanism for this increased prevalence of HH(e) is not
clear but data suggest a role for insulin in regulation of plasma H(e)
levels and that insulin resistance may lead to HH(e). Tables 3
and 4
summarize current knowledge of the
possible interaction between diabetes and HH(e).
|
|
| VI. Hyperhomocysteinemia and Cholesterol Metabolism |
|---|
|
|
|---|
In 482 patients already at high risk for atherosclerotic vascular disease by virtue of hyperlipidemia, 3.7% had high plasma H(e) (128 ). In hyperlipidemic patients, the relative risk of atherosclerotic events for the 80th percentile of plasma H(e) was 2.8-fold greater than that seen for the 20th percentile. Furthermore, it was possible to reduce H(e) concentrations in this hyperlipidemic population with vitamins, suggesting a possible therapeutic approach for multiple risk factor intervention. Plasma H(e) has been shown to be associated with a parental history of cardiovascular disease in children with familial hypercholesterolemia (129 ).
An elevated high-density lipoprotein cholesterol (HDL cholesterol) is well accepted as a protective factor against atherosclerosis. However, a high HDL cholesterol is not necessarily protective in the setting of an elevated plasma H(e) (130 ).
Considerable in vitro data exist to suggest that H(e) may interact with an elevated cholesterol by increasing oxidation of LDL. However, this area remains controversial. It has been suggested that H(e) inhibits glutathione peroxidase activity in vitro and leads to a reduction in steady state mRNA levels for the intracellular isoform in endothelial cells (131 ). Glutathione peroxidase is a member of the antioxidant enzyme family that catalyzes the reduction of lipid peroxides (132 ).
However, in vitro data are not supported by clinical in vivo studies, perhaps related to imprecisions in measuring oxidant stress in vivo, with current technology. As discussed above in the setting of diabetics with vascular disease, plasma TBARS are not elevated above those levels caused by diabetes alone. These data are supported by two other studies. In one, LDL isolated from two patients with homocystinuria showed a similar extent of copper-catalyzed oxidation as LDL from a group of healthy control subjects (133 ). Cordoba-Porras et al. (134 ) investigated the existence of oxidized LDL and the susceptibility to oxidation of lipoproteins in six patients with homocystinuria. The proportion of electronegative LDL and concentration of TBARS did not differ between patients and controls (134 ). Thus, more studies are needed on lipoprotein susceptibility to oxidation in patients with HH(e).
| VII. Hyperhomocysteinemia in Premature Vascular Disease |
|---|
|
|
|---|
A. Epidemiological and prospective studies
In the first prospective epidemiological study of H(e) levels as a
cardiovascular risk factor, 14,916 male physicians with no prior
vascular disease provided plasma samples at baseline and were followed
for 5 yr (6 ). Plasma H(e) levels in 271 physicians who developed a MI
were significantly higher than in paired controls. The relative risk
for MI in those physicians in the 95th percentile for H(e) levels was 3
times greater than those in the 10th percentile, even after adjustment
for other known risk factors for vascular disease (95% confidence
interval, 1.38.8) (6 ).
In 1998, Wald et al. (3 ) reported the British United Provident Association study, a prospective nested case-control study of 21,520 men between the ages of 35 and 64. Homocysteine levels were assayed from stored samples in 229 subjects who died of ischemic heart disease, and 1,126 age-matched controls. Homocysteine levels were higher in the group that died of ischemic heart disease than in the controls. For men with serum H(e) levels in the highest quartile, the increased risk was 2.9 (after adjustment for other factors) than for men in the lowest quartile (3 ).
In a prospective, nested case-control study within the British Regional Heart Study cohort, Perry et al. (135 ) examined the association between serum total H(e) concentration and stroke. Serum was saved from 5,661 men, aged 4059 yr, randomly selected from general practices. During follow-up of up to 12 yr, there were 141 incident cases of stroke among men with no history of stroke at screening. Serum H(e) was measured in 107 cases and 118 control men (matched for age group and town, without a history of stroke at screening, who did not develop a stroke or MI during follow-up). Serum H(e) concentrations were significantly higher in cases than controls. There was a graded increase in the relative risk of stroke in the second, third, and fourth quarters of the serum H(e) distribution (odds ratios 1.3, 1.9, 2.8; trend, P = 0.005) relative to the first. Adjustment for age group, town, social class, body mass index, hypertensive status, cigarette smoking, forced expiratory volume, packed-cell volume, alcohol intake, diabetes, HDL cholesterol, and serum creatinine did not attenuate the association (135 ). These findings add to the evidence that H(e) is a strong and independent risk factor for stroke.
B. Studies in patients with established vascular disease
Several studies have attempted to establish the prevalence of
HH(e) in patients with premature and accelerated vascular disease.
Interest in this field was triggered by the first report by Boers
et al. (17 ) who found elevated plasma H(e) after a
methionine load in 28% of patients with peripheral vascular and
cerebrovascular disease (17 ). Even after adjustment for other risk
factors, plasma H(e) has been found to be significantly higher in
patients with peripheral vascular disease compared with healthy
individuals (136 ). Elevations in peak H(e) after a methionine load
occur in 2842% of patients with vascular disease but rarely, if
ever, in normal subjects (8 34 98 ). It would appear that the MLT
delineates the "at risk" population better than basal levels (34 ).
In a meta analysis of 27 studies of H(e) in atherosclerotic vascular disease, Boushey et al. (7 ) concluded that elevations of H(e) were an independent risk factor for arteriosclerosis. In addition, approximately 10% of the populations coronary artery disease risk appears attributable to H(e). The odds ratio for development of coronary artery disease from increased plasma H(e) at a level of 5 µmol/liter above normal is 1.6 for men and 1.8 for women.
Fermo et al. (137 ) studied patients below the age of 45 with both venous thrombosis and arterial occlusive disease and found moderate HH(e) in 13.1% and 19.2% of patients, respectively. The prevalence of HH(e) was almost twice as high after a methionine load than when based upon fasting levels. Other studies have confirmed the high prevalence of HH(e) in early onset and recurrent venous thrombosis (10 ).
Compelling recent evidence comes from the multicenter study done in nine European countries (11 ). In this study, 750 cases of vascular disease were compared with 800 controls of both sexes younger than 60 yr of age. The relative risk for vascular disease in the highest quintile of plasma H(e) was 2.2 when compared with the lower four quintiles. Methionine loading identified an additional 27% of at risk cases. A dose response effect was noted between total plasma H(e) and risk. The risk was similar to and independent of other risk factors including smoking and hyperlipidemia. An elevated plasma H(e) had a multiplicative effect on risk in smokers and subjects with hypertension. Furthermore, subjects taking vitamins appeared to have a substantially lower risk of vascular disease when compared with nonusers of vitamin supplements. This difference was attributed to lower plasma H(e) levels.
A recent study has established the predictive value of plasma H(e) levels in patients with established coronary artery disease (138 ) in 587 patients with angiographically confirmed coronary artery disease (many of whom underwent bypass surgery and angioplasty). A strong, graded relationship between plasma H(e) levels and overall mortality was found over a 4-yr period. Less than 4% of patients with a plasma H(e) below 9 µmol/liter died, as compared with nearly 25% of those with a plasma H(e) greater than 15 µmol/liter. The relation of H(e) levels to mortality remains strong after adjustment for other potential confounding variables. It is important to emphasize that plasma H(e) levels are a continuous variable in the population studied, and there is no threshold above which the risk for mortality rises.
In another recent case-controlled study an elevated fasting and postload plasma H(e) showed a positive association with risk of severe coronary atherosclerosis (139 ). This association existed over a wide range of plasma H(e), without a clear cut-off point below which there was no increased risk.
Hyperhomocysteinemia appears to have its strongest association with carotid artery disease (140 ). This finding was also highlighted in a cross-sectional analysis of 1,041 elderly subjects in the Framingham Heart Study. A 2-fold increase in the incidence of carotid disease was seen in patients with the highest plasma H(e) concentrations when compared with those with the lowest concentrations (5 ). Furthermore, plasma concentrations of folate and pyridoxal-5'-phosphate were also inversely associated with carotid artery stenosis. The Atherosclerosis Risk in Communities Study (ARIC) has reported higher plasma H(e) in subjects with carotid intimal-media thickening (an early marker of atherosclerosis) when compared with matched controls who did not have such thickening (141 ). Atherosclerotic disease in the aorta as assessed by transesophageal echocardiography also correlates with plasma H(e) (142 ).
Bots et al. (143 ) examined the relationship of H(e) to MI and stroke among older subjects in a nested case-control study of a subset of participants in the Rotterdam Study. One hundred four patients with a MI and 120 with a stroke were compared with 533 control subjects drawn from the study base, who were free of MI and stroke. Nonfasting total H(e) levels were measured. The risk of stroke and MI increased directly with total H(e) (143 ). The linear coefficient suggested a risk increase by 67% for every 1-µmol/liter increase in total H(e). The odds ratios for subjects in the highest quintile of total H(e) level (>18.6 µmol/liter) compared with those with lower H(e) levels were 2.43 (95% confidence interval, 1.115.35) for MI and 2.53 (95% confidence interval, 1.195.35) for stroke (143 ). Associations were more pronounced among those with hypertension (143 ). The study, based on a relatively short follow-up period, provides evidence that among elderly subjects an elevated H(e) level is associated with an increased risk of cardiovascular disease.
C. Negative studies
Not all studies have shown that H(e) is an independent risk factor
for coronary artery disease. A prospective study in Finland of 7,424
healthy subjects at baseline showed development of stroke in 265
subjects over a 9-yr period (144 ). The fact that the affected subjects
did not have an elevated serum H(e) level has been attributed to the
exceptionally low gene frequency predisposing to hyperhomocysteinemia
in Finland. However, the study itself did not assess the frequency of
mutations in the enzymes concerned, and therefore this explanation must
be regarded as speculative.
In 1997 Verhoef et al. (145 ) reported a prospective, nested case-control study that used baseline samples from the Physicians Health Study. After 9 yr of follow-up, subjects with newly diagnosed angina pectoris, without MI, and with subsequent coronary artery bypass graft surgery were studied. Controls had no clinical diagnosis of coronary artery disease. In this study, total H(e) levels and risk of angina pectoris did not correlate (145 ). These data contrast with other reports from this group, which show that H(e) levels correlate