Learn more: PMC Disclaimer | PMC Copyright Notice
Homocysteine, hyperhomocysteinemia and vascular contributions to cognitive impairment and dementia (VCID)
Associated Data
Abstract
Homocysteine is produced physiologically in all cells, and is present in plasma of healthy individuals (plasma [HCy]: 3–10µM). While rare genetic mutations (CBS, MTHFR) cause severe hyperhomocysteinemia ([HCy]: 100–200µM), mild-moderate hyperhomocysteinemia ([HCy]: 10–100µM) is common in older people, and is an independent risk factor for stroke and cognitive impairment. As B-vitamin supplementation (B6, B12 and folate) has well-validated homocysteine-lowering efficacy, this may be a readily-modifiable risk factor in vascular contributions to cognitive impairment and dementia (VCID).
Here we review the biochemical and cellular actions of HCy related to VCID. Neuronal actions of HCy were at concentrations above the clinically-relevant range. Effects of HCy <100 µM were primarily vascular, including myocyte proliferation, vessel wall fibrosis, impaired nitric oxide signalling, superoxide generation and pro-coagulant actions. HCy-lowering clinical trials relevant to VCID are discussed. Extensive clinical and preclinical data support Hcy as a mediator for VCID. In our view further trails of combined B-vitamin supplementation are called for, incorporating lessons from previous trails and from recent experimental work. To maximise likelihood of treatment effect, a future trial should: supply a high-dose, combination supplement (B6, B12 and folate); target the at-risk age range; target cohorts with low baseline B-vitamin status.
1. Introduction
1.1 Homocysteine and vascular contributions to cognitive impairment and dementia (VCID)
Brain vascular lesions contribute to dementing illness in the form of vascular dementia, vascular factors exacerbating Alzheimer’s disease (AD), and other cognitive impairment states where diagnostic criteria for dementia are unmet [81,87]. This considerable burden of dementia-related illness is covered by the concept of vascular contributions to cognitive impairment and dementia (VCID) [35,87,107]. The most common cause of VCID is thought to be a common brain vascular pathology called cerebral small vessel disease (SVD) [28,82,87,92].
Homocysteine (HCy) is a thiol-containing non-essential amino acid (Table 1) that is produced in all cells, as a product of normal folate and methionine metabolism. Elevated plasma homocysteine is referred to as hyperhomocysteinemia (HHCy). HHCy is a robust and independent risk factor for stroke [44,48,61,97,98] and for cognitive impairment [101,105] and is associated with pathologically-confirmed AD [16]. HHCy is associated with an increased rate of hippocampal atrophy [16] and with accelerated cognitive decline in AD patients [88] and is now accepted as a risk factor for AD [6]. Plasma HCy concentration is strongly associated with hippocampal atrophy, white matter lesions and lacunar infarcts in cross-sectional studies [32,61,117,123]. The association with hippocampal atrophy appears to be independent of amyloid pathology [14]. White matter lesions reflect vascular damage [92] and it is therefore notable that white matter changes are associated with HHCy [47,49,64,94] and with low vitamin B12 levels [21] and low folate [51,100]. Evidence supporting a causal relationship between HHCy and cerebrovascular injury includes: i) independent, graded association of HHCy with stroke in prospective and retrospective clinical studies [48,61]; ii) genetic association studies of genes regulating HCy metabolism, using Mendelian randomisation [9]; iii) HHCy-induced lesions in experimental animals [4,68,108,111]. Dietary supplementation with the appropriate B-vitamins (B6, B12 and folate) clearly ameliorates HHCy. Thus HHCy may be a modifiable risk factor in VCID.
Table 1
Homocysteine and related molecular species
| Molecule | Structure | Formation |
|---|---|---|
| Homocysteine, HCy |  | Formed in normal cell metabolism of folate and methionine |
| Homocystine |  | Disulphide bonded dimerization of HCy |
| Homocysteine thiolactone (shown as hydrochloride) |  | Condensation of the SH and carboxylic acid groups of HCy |
| Homocysteic acid, HCA (−SO3H dissociates to form homocysteate anion, −SO3−) |  | Uncertain, possibly oxidation of HCy |
Abbreviations. HCy, homocysteine. HCA, homocysteic acid.
1.2 Biochemistry of homocysteine
Rare genetic mutations in folate and methionine metabolizing enzymes (CBS, MTHFR) produce severe HHCy (plasma HCy in excess of 100 µM; up to ∼500 µM). Plasma HCy can also be elevated by smoking, aging, deficiency in folate or vitamin B12, or renal failure. Clinical mild-moderate HHCy is common, especially in older people. Mild-moderate HHCy is various defined, with lower limits ranging from 10–15 µM and upper limits 30–100 µM.
Plasma HCy is derived from cellular export via cysteine and folate transporters [50,56] and circulating HCy levels are maintained at relatively low levels by ongoing enzymatic conversion to either methionine or cysteine (see Figure 1). Plasma total HCy concentration in healthy individuals is typically in the range 5–10 µmoles/litre (µM). Previous detailed reviews are available on HCy biochemistry [52,115]. For the purposes of this review (including data from biochemical assays, cell cultures, experimental animals and clinical populations) we have defined mild-moderate HHCy by homocysteine concentrations in the range 10–100 µM, and severe HHCy as >100 µM.
Methionine is converted to homocysteine by methylation and subsequent hydrolysis. Homocysteine is then either cycled back to methionine via the folate cycle, catalysed by methylenetetrahydrofolate reductase (MTHFR) and the essential cofactor vitamin B12. Alternatively homocysteine can be further metabolized to cysteine via cystathionine beta synthase (CBS) and the essential cofactor vitamin B6. Homocysteine conversion to cysteine occurs primarily in the liver. All other reactions are ubiquitous. For further details on HCy biochemistry, see [52,115]. SAM, S-adenosyl methionine; SAH, S-adenosyl homocysteine.
1.3 Homocysteine and cerebral small vessel disease (SVD)
SVD is a vessel pathology that affects small penetrating arteries (up to 400 µm outer diameter) in deep subcortical brain regions, primarily the basal ganglia and deep white matter [31,77]. It is characterised neuropathologically by fibrous, hyaline thickening of vessel wall, loss of myocytes, and relative preservation of endothelium [40,90]. SVD is common in brains of older people, and is a prevalent cause of microinfarcts and lacunes, lacunar stroke, and diffuse white matter lesions (“leukoaraiosis”) [87,90,92]. Further, neuroimaging [28,114,116] and neuropathology [30,31,104] data suggest that SVD is also the most common cause of VCID[87,90,92]. While increasing age and hypertension are risk factors for SVD, the pathogenesis of SVD is not well understood [34,90].
Multiple studies have shown that mild-moderate HHCy is a risk factor for SVD [32,44,45,61,61,62,117,123]. In case-control studies HHCy was positively-associated with SVD severity, with highest plasma total HCy concentration (plasma [HCy]; average 20 µM) in patients with confluent leukoaraiosis on MRI scans, indicative of more severe SVD [44,61]. The association between SVD and HCy was markedly attenuated after controlling for circulating markers of endothelial dysfunction, suggesting a possible endothelial mechanism for the HHCy-SVD association [44].
2. Cellular actions of HCy
Here we review the actions of HCy at the molecular and cellular level with particular reference to brain cells, specifically neurones and vascular cells. Several reported actions occur at HCy concentrations below 100 µM, within the range of mild-moderate clinical HHCy, thus direct actions of HCy in vivo on neurones, vascular endothelial cells and VSMC in vivo appear likely.
2.1 Effects of HCy on brain neurones
We carried out a systematic review of concentration dependence of the actions of HCy in brain cells (results shown in Table 2). We searched PubMed (on 12th June 2015) for articles in English containing the terms: (brain OR cerebr*) AND (homocyst*) AND (potency OR half-maximal OR IC50 OR EC50 OR ED50 OR affinity). This yielded 109 results. Abstracts were viewed by two independent reviewers (NY, AHH) and 41 were discarded by consensus (not relevant, no quantitative data, or not primary data source). Of the 68 retained, on viewing the full text data were extracted from all where quantitative data were given on the potency of homocysteine actions (Table 2) [2,3,25,27,37,38,46,54,75]. Quantitative data were also extracted on the actions of HCA (see Supplementary table S1).
Table 2
Effects of HCy on brain neurones and glial cells
| Effects observed | Effective concentration | Cell or tissue | Species | Reference |
|---|---|---|---|---|
| Neurite loss and size reduction | >100 µM | Mesencephalic tegmental neurons | Rat | Heider et al. (2004) [46] |
| Inhibition of neural network activity | 400 µM | Embryonic neocortical neurons | Rat | Gortz et al. (2004) [37] |
| Lipid peroxidation | 100 – 1000 µM | Synaptosomes | Rat | Jara-Parado et al. (2003)[54] |
| Inhibition of glutamine uptake | 2 mM | Cerebral neuronal non-synaptic mitochondria | Rat | Albrecht et al. (2000)[2] |
| PAG inhibition | 2 mM | Cerebral neuronal non-synaptic mitochondria | Rat | Albrecht et al. (2000)[2] |
| L-glutamate transporter inhibition | 1000 µM | Synaptosomes, from pre-frontal cortex | Mouse | Griffiths et al. (1989)[38] |
| GABAA receptor binding | Ki(apparent) : >10 mM | Synaptic membranes | Cow | Egubta & Griffiths (1987) [25] |
| Inhibition of high affinity GABA uptake | IC50 >5 mM | Synaptosomes [whole brain] | Rat | Allen et al. (1986) [3] |
| Inhibition of high affinity taurine uptake | IC50: 2.8 mM | Synaptosomes [whole brain] | Rat | Allen et al. (1986)[3] |
| Inhibition of high affinity taurine uptake | IC50: 4.8 mM | Astrocytes from cerebral cortex | Rat | Allen et al. (1986)[3] |
| Inhibition of MMP- 2 activity | IC50: 727 µM | Brain tissue homogenate | Pig | Emara & Cheung (2006)[27] |
| Inhibition of MMP- 9 activity | IC50: 2.35 mM | Brain tissue homogenate | Pig | Emara & Cheung (2006)[27] |
| Inhibition of kynurenic acid production | IC50: 6.4 mM | Cortical neurons | Rat | Luchowska et al. (2005)[75] |
Abbreviations. MMP, matrix metalloproteinase; PAG, Phosphate-activated glutaminase.
Neurotoxicity and morphological changes
Incubation of mesencephalic tegmental neurons from rats with HCy (100µM) for 48 hours produced morphological changes in the neurites of the tyrosine hydroxylase positive neurons (assumed to be dopaminergic)[46]. Fewer and shorter neurites were observed, although there were no significant effects on cell numbers [46]. This suggests that HCy exhibits a mild toxicity towards these dopaminergic neurons at a concentration of 100µM. In vivo injection of HCy into the left ventricle produced dose-dependent neuronal loss [124], with marked neuronal loss. These effects may be mediated by the NMDA and metabotropic glutamate receptors, as co-administration with antagonists of these receptors attenuated the neurotoxic effects of HCy [124].
Neural network activity suppression
HCY reduced neural network activity in spontaneously active embryonic rat cortical neurons (as measured by basal spontaneous spike rate). The IC50 for HCy (401µM) was above the range for severe HHCy, so the authors concluded that these effects were not clinically relevant. Homocysteate, by contrast, had an IC50 of 1.3µM. Homocysteate is often found to be elevated in patients with HHCy [37] and HCA may be a clinically relevant species.
Lipid peroxidation
HCy (100–1000µM) increased lipid peroxidation in rat brain synaptosomes in a concentration-dependent manner [54]. This action was prevented by an NMDA receptor antagonist [54] supporting a glutamate receptor dependent pathway for HCy-mediated neuronal damage.
Several other cellular actions of HCy are reported, with effective concentrations in the millimolar range, (Table 2). These are unlikely to be relevant to clinical HHCy.
Homocysteate and Homocysteic acid
The physiological metabolic pathways of HCA (see Table 1 for structure) are relatively little known. There is evidence that HCy may be oxidised to form HCA within brain and other tissues [37]. Homocysteate has neurotoxic effects, independent of HCy [37,46]. These appear to be due to NMDA and metabotropic glutamate receptor-dependent excitotoxicity (see Supplementary table S1)[26,36,93,96]. HCA is clearly more potent than HCy in some cellular actions [37,38,46]. While HCA/homocysteate may participate in some of the cytotoxic processes observed in HHCy, native HCA levels appear to be low.
2.2 Effects of HCy on vascular cells
Cell damaging effects of HCy on blood vessels have been suggested for many years. Mild-moderate HHCy could directly cause cell damage in vivo, or could represent a secondary event or biomarker of a pathological process. The salient cellular actions of HCy at clinically relevant concentrations are listed in Table 3 and are discussed below.
Table 3
HCy effects in vascular cells
| Effective concentration | Cell type | Species | References | |
|---|---|---|---|---|
| In vascular smooth muscle cells | ||||
| Increased cell proliferation | 50 −100 µM | Thoracic aortic VSMC | Chick | Dalton et al. 1997 [17] |
| Increased protein kinase C activity | 3 −100 µM | Thoracic and abdominal aorta VSMC | Chick | Dalton et al. 1997 [17] |
| Increased ERK2 activity | 0.1–100 µM, EC50 0.5 µM | Thoracic and abdominal aorta VSMC | Chick | Brown et al. 1998[7] |
| Increased CTGF release, Collagen I and fibronectin synthesis | 50–500 µM | Umbilical vein VSMC | Human | Liu et al. 2008[73] Majors et al. 1997[76] |
| Lysyl oxidase inhibition (by HCy lactone) | 5 – 50 µM, Ki 20 µM | Whole aorta | Calf | Liu et al. 1997 [72] |
| In endothelial cells | ||||
| Reduced cell proliferation | 10 – 50 µM, EC50 10–20 µM | HUVEC | Human | Wang et al. 1997[119] |
| Downregulation of cyclin A | 50 µM | HUVEC | Human | Jamaluddin et al. 2007 [53] |
| Inhibition of endothelial NOS | 50 – 200 µM | Thoracic aorta endothelial cells | Human & Mouse | Jiang et al. 2005[57] |
| Inhibition of endothelial- dependent vasodilatation | 10 – 100 µM, EC50 10 µM | Coronary artery endothelial cells | Pig | Chen et al. 2002 [11] |
| Increased superoxide formation | 30 −1000 µM | Aortic endothelial cells | Pig | Lang et al. 2000 [67] |
| Increased release of MMP-9, loss of Integrin-1 | 12 – 40 µM, EC50 20 µM | Cerebral microvessel endothelial cells | Mouse | Shastry & Tyagi 2004[103] |
| Reduced binding of anti-thrombin III | 10 µM | Aortic endothelial cells | Pig | Nishinaga et al. 1992 [86] |
| Other actions | ||||
| Increased MCP-1, IL-6 release | 10 −1000 µM, EC50 30 µM | Aortic adventitial fibroblasts | Rat | Liu et al. 2012[74] |
| Cell transformation of fibroblasts to myo-fibroblasts | 10–100 µM | Aortic adventitial fibroblasts | Rat | Liu et al. 2012[74] |
| Reduced platelet- fibrinogen binding | 1 – 10 µM | Whole blood | Human | Riba et al. 2004[95] |
| Reduced ApoA- fibrin binding | 8 – 2000 µM, EC50 ∼ 100 µM | Fresh plasma | Human | Harpel et al. 1992 [43] |
| Increased MCP-1, IL-8 release from blood monocytes | 10 −1000 µM, EC50 10–30 µM | Whole blood | Human | Zeng et al. 2003 [125] |
Abbreviations. ApoA, apolipoprotein-A. CTGF, connective tissue-derived growth factor. ERK2, Extracellular signal-regulated kinase. HUVEC, human umbilical vein endothelial cells. IL-6, −8, interleukin-6, −8. MCP-1, monocyte chemotactic protein-1. MMP-9, matrix metalloproteinase-9. NOS, nitric oxide synthase. VSMC, vascular smooth muscle cells.
The effects reported at extracellular HCy concentrations below 100 µM suggest that direct HCy cytotoxicity may be relevant in vivo. We note with caution that there is little information on the clinically relevant range of intracellular HCy concentrations (in physiological conditions, approximately 100 µM). Amino acid transporters are present in vascular smooth muscle cells and endothelial cells, with high affinity for HCy (KM 40–100 µM)[56].
2.2.1 Vascular smooth muscle cells
Cell proliferation
Moderate concentrations of HCy (10–100 µM) enhanced proliferation of vascular smooth muscle cells (VSMC) in primary cultures [17,76]. In VSMC cultured from embryonic thoracic aorta, acute HCy-induced cell proliferation was associated with induction of c-fos and c-myb, and translocation of protein kinase C (PKC) to the plasma membrane [17]. In agreement with a PKC-dependent mechanism, HCy concentrations as low as 3 µM (range 3–100 µM) stimulated diacylglycerol synthesis in these cells [17]. The growth-related MAP kinase ERK2 was also rapidly activated by very low concentrations of HCy (EC50 0.5 µM) in aortic VSMC cultured from chick embryos [7]. How this action relates to normal adult physiology (with ambient plasma HCy well above this concentration) is unclear.
Mitogenic actions of acute HCy treatment were seen in myocyte cultures derived from aorta of adult animals, but only as higher concentrations (500–1000 µM)[112,113]. These agree with earlier in vivo studies in nonhuman primates undergoing chronic intravascular infusion with HCy [42]. In these animals steady state plasma HCy of 160 µM produced intimal lesions with endothelial cell loss and myointimal cell proliferation, resembling arteriosclerosis [42]. The cell proliferative effect of HCy is clearly not a generalized response, as HCy suppresses cell proliferation in endothelial cells, (see 2.2.2 below). HCy may also influence cell fate. HCy (10–100 µM) treatment of adventitial fibroblasts cultured from rat aorta stimulated transformation to myofibroblast-like phenotype (expression of the myocyte-associated proteins smooth muscle actin and SM22α) [74].
Fibrosis and collagenosis
HCy treatment potently increased collagen synthesis in smooth muscle cells in vitro (over a concentration range 50 – 500 µM) [76]. In accord with a pro-fibrotic action, treatment of human VSMC cultures with a moderate concentration of HCy (50 µM) increased expression of the pro-fibrotic growth factor CTGF, collagen type I, and fibronectin [73]. In accord with effects of HCy on vessel fibrosis, HCy lactone (see Table 1) irreversibly inhibited lysyl oxidase (Ki 21 µM) [72]. This enzyme catalyses cross-linking between elastin and fibrillar collagen, and is essential for growth and repair of connective tissue matrices.
2.2.2 Endothelial effects of homocysteine
Mild-moderate HHCy is associated with endothelial dysfunction in humans [10,44,109], experimental primates[71], and rodents [18,24,59,68,108,122].
Inhibition of cell proliferation
HCy suppresses cell proliferation in endothelial cells, contrary to its proliferative action in vascular myocytes. In preparations of human aortic endothelial cells (HAEC), pig aortic endothelial cells and HUVEC, tritiated thymidine incorporation was inhibited dose-dependently by modest HCy concentrations (10 to 50 µM; 24hr exposure) [119]. The half-maximal concentration for L-HCy was approximately 10 µM in these various endothelial preparations, with D,L-homocysteine showing roughly 2-fold lower potency, and L-cysteine without effect, indicating a specific action of L-HCy [119]. HCy-treated HAEC had greatly increased levels of s adenosyl homocysteine (SAH; which is a potent inhibitor of methyl transferase enzyme activity) and reduced carboxyl methylation of p21-ras. Inhibition of endothelial proliferation has also been reported at higher HCy concentrations (100 to 1000 µM) [112].
At these high concentrations, caspase-dependent cell death is seen (500 – 1000 µM) [69]. This pro-apoptotic action of high HCy concentration was abrogated by the nitric oxide donor SNAP, the antioxidant alpha-tocopherol, or by an antioxidant mixture of superoxide dismutase and catalase [69]. High concentrations of HCy (>200 µM) produced endoplasmic reticulum (ER) stress in endothelial cell cultures [89].
Inhibition of endothelial nitric oxide synthase (eNOS)
In cultures of aortic endothelial cells from adult mice, moderate concentrations of HCy (range 5 – 200 µM) inhibited eNOS activity, evidenced by depressed L-citrulline and nitrite formation [57]. This accords with in vivo data showing lower eNOS activity in Cbs null mice relative to wildtype mice, and intermediate eNOS activity in Cbs+/− heterozygotes [57]. Similarly, in coronary artery rings from adult pigs application of HCy (10 – 100 µM) reduced endothelium-dependent vasodilation [11]. Endothelium-independent vasodilation (in response to the NO donor nitroprusside) was unaffected, supporting an endothelial action for HCy. Downregulation of eNOS expression was also seen in human aortic endothelial cells but only at higher HCy concentration (200 µM)[57]. By way of caution, the endothelial preparations in most of these studies came from large arteries, and may not entirely reflect microvascular signalling.
Gene induction effects
Screening for gene induction in endothelial cells following HCy exposure revealed several candidates, [65,102]. In microvascular endothelial cells derived from mouse brain, treatment with moderate concentrations of HCy (up to 40 µM, 24 hours) increased activity of the matrix metalloproteinase MMP-9 concentration-dependently, while growth factor synthesis decreased (VEGF, FGFα, leptin) [103]. Over the same HCy concentration range, shedding of β1 integrin from the endothelial cell surface was also seen[103]. Such effects in intact blood vessels could undermine the extracellular matrix integrity in the intimal layers around endothelial cells, and contribute to the anti-proliferative effects of HCy in endothelial cells.
2.2.3 Homocysteine actions on blood components
Acute exposure to physiological concentrations of HCy (less than 10 µM) increased the affinity and degree of specific binding for fibrin to Lp(a), a complex that includes Apolipoprotein-A and shares structural homology with plasminogen [43]. HCy increased binding by plasmin-modified fibrin (20 fold) and by native fibrin (4 fold). This may be a thiol-dependent action as other small thiol compounds (cysteine, GSH) had a similar effect, whereas sulphur containing compounds that lack the free -SH group including methionine and SAH had no effect. Low HCy (10 µM) also greatly reduced binding of anti-thrombin III to endothelial cells derived from porcine aorta [86].
The pro-coagulant effect of modest levels of HHCy was evident in patients with peripheral occlusive arterial disease (diagnosed as intermittent claudication). In isolated platelets taken from patients with higher plasma HCy (average 19 µM) the inhibitory effect of a nitric oxide donor (GSNO, 1 µM) on platelet-fibrinogen binding was 5-fold lower than that of subjects with normal HCy (average 11 µM)[95]. In HUVEC cultures, binding of tissue plasminogen activator (tPA) to the cell surface was inhibited by high concentrations of HCy (up to 1.5 mM) [41].
3. Experimental animal models relevant to elevated homocysteine
Induction of HHCy in an animal model can be achieved via genetic manipulation or diet. Here we outline animal models that have shown HHCy, sufficient to cause cognitive deficits and vascular adverse events in the brain. Here we discuss transgenic mouse strains with mutations in the genes that underlie severe HHCy in humans (CBS, MTHFR), and animals with diet-induced HHCy. Further information on animal models is given in previous reviews [19,58,110].
3.1 Cystathionine beta synthase
(CBS) converts HCy to cystathionine, which is then converted to cysteine, during normal biochemical processing. In humans, deficiencies in CBS result in severe HHCy and increased risk of thrombosis, and are the most common cause of hereditary HHCy. While homozygous Cbs knockout mice [121] have a neonatal lethal phenotype, inducible Tg-hCBS mice avoid this problem [39]. In Tg-hCBS mice, human CBS is under the control of a zinc-inducible promoter. During pregnancy and lactation drinking water is supplemented with Zn2+ to rescue the neonatal lethal phenotype[120]. At weaning, the zinc supplementation is withdrawn and the progeny develop HHCy (plasma HCy 170 µM, relative to 5 µM in control animals). Another CBS transgenic model reproduces the I278T mutation, which is the second most common allele found in CBS-deficient patients[66]. Tg-I278T Cbs−/− mice have less than 3% wildtype CBS activity and develop HHCy (plasma HCy 290 µM). Tg-I278T Cbs−/− mice also exhibit facial alopecia, osteoporosis, ER stress in the liver and kidney, and reduced mean survival, Tg-hCBS mice do not show any of these phenotypic changes. Although thrombosis is the most common symptom of CBS deficiencies in humans, neither mouse model exhibits any thrombotic or vascular defects.
Cbs+/− heterozygous mice have a 50% lower CBS activity compared to wildtype mice, and develop mild HHCy on normal diet (plasma HCy 6–8 µM) [4,59,121], augmented by high methionine diet for 8–20 weeks (plasma HCy ∼20 µM)[4]. Cbs+/− heterozygote mice exhibit substantial vascular dysfunction. They have thickened arteries, including cerebral arteries, with evidence of endothelial damage [4,59]. Cerebral arteriolar walls are 25% thicker in Cbs+/− heterozygote mice, relative to WT controls [4]. This was accompanied by mild hypertension and blood-brain barrier dysfunction [4]. This vasomotor dysfunction is primarily based on endothelial dysfunction, and is largely attributed to redox imbalance and decreased bioavailability of NO. Compared to Cbs null mice, Cbs+/− mice display a milder HHCy and may be a more useful model to study hyperhomocysteinemic effects on the vasculature.
3.2 Methylenetetrahydrofolate reductase
(MTHFR) is the rate-limiting enzyme in conversion of HCy to methionine. In humans, several MTHFR polymorphisms produce HHCy and neurological conditions, including progressive demyelinating neuropathy and cognitive impairment. MTHFR knockout mice [13] exhibit elevated plasma HCy . Similar to the Cbs mouse model, MTHFR+/− heterozygotes exhibit mild HHCy (plasma HCy ∼20 µM, relative to 3–5 µM in wildtype littermates)[80] and appear outwardly healthy, while MTHFR−/− homozygotes (plasma HCy 31–33 µM, relative to 3 µM in WT animals) survive poorly and present with motor and gait abnormalities within 5 weeks after birth. MTHFR+/− mice exhibit some loss of function in cerebral vessels [85] and high intramural collagen in peripheral arteries [85]. MTHFR−/− homozygotes also present with abnormal lipid deposition in the aorta and disruption of the laminar structure of the cerebellum (though with no obvious changes in the cerebral cortex or cerebrum).
3.3 Dietary induction of HHCy
in mice and rats can be achieved through either a reduction in vitamins B6, B9 and B12, or by diets supplemented with HCy or methionine [4,18,68,108,111]. Vascular dysfunction has been reported in primates (Cynomolgus monkeys) following diet-induced mild-moderate HHCy (plasma HCy 11 µM, relative to 4 µM on normal diet) [70,71]. (Earlier studies in baboons employed direct infusion of HCy [42]). Responses of resistance vessels to endothelium-dependent vasodilators were markedly impaired in HHCy monkeys, and anticoagulant thrombomodulin activity in the aorta decreased by 34 % [71].
Vitamins B6, B9 and B12 act as essential cofactors for the conversion of HCy to methionine or cysteine, combined with enrichment in dietary methionine, causes an increase in conversion to HCy. In 6-month-old rats fed a folate-deficient diet for 8 weeks, increased plasma HCy (increased from 6 µM to 19 µM) was accompanied by ultrastructural changes to cerebral capillaries, endothelial damage, swelling of pericytes, basement membrane thickening and fibrosis [63]. Rats fed a diet high in HCy for 5 or 15 months (plasma HCy 26 µM at both time-points, relative to 10 µM in controls) both showed cognitive impairments, decreased acetylcholine in the brain and cortical micro-hemorrhages[91]. In wild-type mice fed a B-vitamin deficient diet, cognitive deficits resulted from relatively-brief (10 week) diet-derived mild HHCy (plasma HCy 35 µM, relative to 5 µM on normal diet) [111]. The authors found a decline in spatial learning (water maze task) accompanied by rarefaction of hippocampal blood vessels and of microglia [111]. Aged Tg2576 mice on HCy-elevating diet (plasma HCy 27 µM, relative to 5 µM in controls) also showed deficits in spatial reference memory, but not working memory tasks (delayed non-match to position)[5].
Plasma HCy levels were also increased in 3-month-old mice that were fed a diet deficient in vitamins B6, B9 and B12 and also enriched in methionine for 3 months [108]. These mice exhibited cognitive impairments on the two-day radial arm water maze, cerebral microhemorrhages and increased MMP2 and MMP9 activity. This dietary model results in plasma HCy in the range 70–90 µM [108]. While these concentrations are higher than most human subjects classified under mild-moderate HHCy, this level is classified as moderate HHCy in mice [29].
4. Overview of clinical interventional trials relevant to stroke and dementia
If the hypothesis that HHCy causes cognitive decline is correct, then lowering plasma HCy concentrations (e.g. by B vitamin supplementation) should slow cognitive decline and possibly prevent dementia. Multiple clinical trials have tested B vitamin administration (folic acid, vitamins B6 and B12) with a cognitive endpoint. There have been three meta-analyses of these intervention trials [15,33,118].
4.1 Meta-analyses of clinical trials
Three meta-analyses [15,33,118] will be considered here, with respect to the following important features related to trial design [78].
- i)The hypothesis being tested. If the hypothesis is that treatment slows cognitive decline, then the placebo group must exhibit cognitive decline. Unless the placebo group declines, the trial can only show whether the treatment enhances or worsens cognition.
- ii)Age range of the test cohort. If cognitive decline is the outcome, the age of the subjects should be in the range where cognitive decline occurs.
- iii)Trial duration. Is this sufficient to observe cognitive decline? In cognitively healthy elderly people, MMSE declines by only about 0.1 points per year on average.
- iv)The instrument used to assess a treatment effect. Is the instrument sensitive enough to detect change over the period of the trial? The instrument may display ceiling or floor effects, which limit the detectable effect size. Points (ii), (iii) and (iv) will together dictate whether the trial has sufficient power to detect a real effect.
- v)Is the intervention appropriate? For B vitamins, the daily doses should be sufficient to lower plasma HCy by at least 20%.
- vi)Is the group likely to respond to intervention? A B-vitamin supplement will produce no effect if the participants already have adequate B vitamin status. Such trials should be limited either to subjects with high plasma HCy or insufficient B vitamin function. Alternatively, subgroup analyses should be performed to stratify for baseline values of B vitamins and plasma HCy.
The majority of the 26 trials analysed in the three meta-analyses [15,33,118] did not fulfil these requirements. Furthermore, by combining data from different trials (e.g. by using standardised mean differences) the meta-analyses have neutralised some of these requirements, even though they were fulfilled in individual trials. These three meta-analyses [15,33,118] have recently been appraised by others [78] and are outlined below.
In one analysis 7 of the 9 trials were of short duration (≤ 12 months; the two shortest lasted 1 month) with an overall median duration of 6 months [118]. In the trial with the largest number of participants (n=910), which accordingly was given a strong weighting, the doses of folic acid (0.2 mg) and vitamin B12 (1 µg) were too low to influence plasma HCy. In a trial where the doses of vitamins were adequate [79], there was no significant cognitive decline in the placebo group (mean MMSE 29.17 at baseline, 29.32 after 2 y). Thus, the conclusion that folic acid has no effect on the prevention of age-related cognitive decline [118] may be false. The authors concluded that longer trials of folate supplementation are needed, with larger cohorts and with cognitive decline and dementia as outcomes [118].
A larger meta-analysis combined 19 trials (n= 5,398 participants) [33] of which 8 were in the previous analysis [118]. In this analysis, only 6 of the trials (n=785 participants) were performed on subjects with cognitive impairment and the results were reported separately for this group and for participants without cognitive impairment. This study concluded “It remains to be established whether chronic treatment with B-vitamins is associated with better cognitive outcomes, or if their use could reduce the risk of dementia in later life” [33]. This meta-analysis also suffered from lack of statistical power to detect differences associated with small effect sizes (<0.2) between the supplemented and placebo groups. Another limitation is that the analysis combined data from 62 different cognitive tests, by converting the individual test scores to standard deviation scores, in order to compute standardised mean differences. Thus, there was considerable clinical heterogeneity in the combined data. In order to analyse the data, the authors assumed zero correlation between the assessments at baseline and study endpoint. This may be a serious limitation, as baseline cognitive test performance is a well-known predictor of later test performance. Thus, in any analysis of cognitive decline it is imperative to use a model that controls for the baseline value (which is not possible with the widely-used standardised mean difference method). A further limitation of this method is that it depends upon the t test and so the analysis cannot be adjusted for important co-variates (e.g. gender and education) that are known to influence cognitive change.
A third meta-analysis[15] aimed to test the effect of lowering plasma HCy on ‘cognitive ageing’ (though this was not defined). The analysis considered 11 trials (n=22,000) of which 5 had been included in one or both of the previous two studies, and concluded “B-vitamin supplementation caused HCy-lowering but was without significant effect on cognitive function or cognitive aging”[15]. Trials on subjects with a prior diagnosis of cognitive impairment or dementia were specifically excluded[15]. Furthermore, in most subjects in the included trials, cognitive decline could not be measured since baseline global cognitive measures were not available for 76% of subjects. It is therefore unclear what can be concluded from such trials, apart from a lack of a cognitive enhancing (or a harmful) effect of B vitamins.
4.2 The FACIT and WAFACS trials
Two of the trials (n=2,825) that were included in the above meta-analyses did report cognitive scores at baseline. Both these trials showed significant effects of B vitamins on cognitive change in those subjects with high plasma HCy [23], or with poor B vitamin status [60]. The FACIT trial (n=818; mean age 60 y) recruited people from the Netherlands with plasma HCy was between 13 and 26 µM, but with adequate vitamin B12 status[23]. Folic acid (0.8 mg/day) was administered for 3 y. At the end of the trial plasma HCy concentration was 26% lower than in the placebo group. The trial assessed several cognitive domains and found that performance on sensorimotor speed, information processing speed, and complex speed had all declined in the placebo group after 3 y but that the decline was slowed by folic acid treatment. There was an improvement in memory performance in the placebo group (likely due to learning effect) but a greater degree of improvement occurred in the folic acid group. Global cognitive performance also improved more in the folic acid group. Notably, there was a larger beneficial effect of folic acid on information processing speed in those whose base-line plasma HCy was above the median than in those below the median. The authors estimated that folic acid treatment conferred cognitive performance of someone 4.7 y younger for memory, 1.7 y younger for sensorimotor speed, 2.1 y younger for information processing speed, and 1.5 y younger for global cognitive function. Overall, this trial was consistent with the view that lowering plasma HCy slows decline in those cognitive domains that are sensitive to ageing[23].
In the US-based WAFACS trial (n=5442, all female, age >40y) a daily combination of B vitamins (2.5 mg folate, 50 mg B6, 1 mg B12) was tested for secondary prevention of cardiovascular disease [60]. Cognitive testing was carried out in a substudy (n=2009, age>65y). While cognitive change from baseline did not differ between the vitamin-treated and placebo groups in the cohort as a whole, there was a significant protective effect in those participants with low baseline dietary intake of folate [60].
4.3 The VITACOG trial
VITACOG was designed to test whether lowering plasma HCy would slow the accelerated rate of brain atrophy that occurs in people with MCI [106]. Volunteers (n = 168) with MCI were randomized to placebo or a daily combination of folic acid (0.8 mg), vitamin B12 (0.5 mg) and vitamin B6 (20 mg) for two years. Volumetric MRI scans were performed at the start and end of the trial. B vitamin treatment led to a highly significant slowing in the rate of global brain atrophy (average 30%). The effect of B vitamin treatment depended on baseline plasma HCy, with a 53% slowing of atrophy in those in the top quartile (> 13 µM)[106].
A further analysis of the MRI scans from VITACOG subjects revealed strong effects of B vitamin treatment on regional brain atrophy in MCI [22]. There was 7-fold less regional brain atrophy in the B vitamin group compared with the placebo group, though this effect was significant only in people with plasma HCy above the median. The gray matter regions protected by B vitamin treatment included: structures of the medial temporal lobe, the precuneus, angular gyrus and supramarginal gyrus, all regions known to be vulnerable to the AD disease process. It was concluded that the B vitamin treatment had slowed the atrophy of AD-related regions of the brain, in people with raised plasma HCy [22].
VITACOG was not powered to detect any effect of homocysteine-lowering on cognition. Nevertheless, a pre-specified analysis stratifying for plasma HCy showed that in people with baseline plasma HCy above the median (11.3 µM), cognitive decline was virtually prevented in the following domains: episodic memory, semantic memory and global cognition (MMSE) [20].
4.4 Importance of initial homocysteine concentration
The VITACOG trial implies a threshold effect of plasma HCy on biological outcomes such as brain atrophy and cognition. Slowing of brain atrophy and of cognitive decline occurred only in people with plasma HCy > 11 µM, while improvement in clinical measures was only found with plasma HCy > 13 µM. These findings are consistent with previous observations. For example, there is a concentration-related (> 10 µM) association of plasma HCy with incidence of dementia. In the OPTIMA study, only concentrations above 11 µM were associated with an increased rate of atrophy of the medial temporal lobe[16]. At concentrations > 10 µM there was a concentration-dependent increase in the rate of cognitive decline in AD patients under 75 y [88]. If the threshold for effects of plasma HCy is about 10–11 µM that could explain why some studies, for example in countries that already employ mandatory folic acid fortification, do not find associations of plasma HCy with cognitive or brain outcomes. Thus, an 18-month trial of high-dose B vitamins in patients with AD in the USA (baseline plasma HCy 9.16 µM) did not find an overall effect of treatment on cognitive decline[1]. It should be noted that those patients with mild AD did show a protective effect of B vitamins after 15 months, which may indicate that timing of HCy-lowering intervention is critical.
4.5 Interactions with other risk factors
Retrospective analysis of the VITACOG data revealed an interaction with omega-3 fatty acids. It emerged that the protective effect of B vitamin treatment on both brain atrophy and cognitive decline in MCI only occurred in those participants with a good status of long-chain omega-3 fatty acids [55]. We also observed that the beneficial effect of B vitamins on brain atrophy was absent in subjects taking aspirin [106]. These results emphasize the importance of considering risk factor interactions. This may be a reason why some trials of B vitamins have failed, if the treated population were taking aspirin regularly, as in many trials included in the meta-analysis by Clarke et al.[15], or had a poor omega-3 fatty acid status. Further trials are needed to test B vitamins in combination with omega-3 fatty acids, in people not taking regular aspirin.
4.6 Future trials of B vitamin supplementation for HHCy lowering
In our view further trails of combined B vitamin supplementation are called for. These should incorporate the lessons from previous trails, and from recent experimental work. In order to maximise likelihood of treatment effect, future trials should aim to target individuals who are in the at-risk age range, but not demented. Cohorts with inadequate baseline B vitamin status should be targeted, and a full combination supplement (B6, B12 and folate) supplied at high dose. With the benefit of extant trial data, accurate power analyses should be possible, to allow conservative design in terms of sufficient cohort size and treatment duration.
A possible scenario could be a government-sponsored scheme within areas of social deprivation, where long-term vitamin inadequacy is prevalent. Participants would be randomised to a highly-flavoured tablet formulation, such as a vitamin C preparation, without B vitamins (placebo) or with a substantial dose of B6/B12/folate (treatment). Recruitment would be offered to all older persons in a public setting. Recruitment, registration and supply would be at a routine public forum, such as weekly pension collection, or GP drop-in clinics for cardiovascular monitoring.
For a future trial in VCID, suitable endpoints may be MRI-derived measures of disease progression [99]. Such surrogate biomarkers may be of utility in trials for VCID for two main reasons. First, a large proportion of individuals who exhibit VCID and VCID biomarkers will not convert to dementia. Second, MRI-based endpoints will permit a test of efficacy in a smaller cohort than would be required to detect a difference in cognitive decline.
4.7 Preclinical studies lowering Hcy levels
While no animal studies have been done to lower Hcy levels and test cognition or neuropathology, some animal studies have been done with the aim of studying cardiovascular effects of Hcy lowering. One study used the CBS−/− genetic model, induced hypercholesterolemia, and then targeted CBS expression using a viral vector to restore low Hcy levels. Using this approach, the investigators found that the restoration of low Hcy levels improved infarct healing and attenuated remodelling after a myocardial infarction [83]. The same group used the same genetic approach in a model of pressure overload-induced cardiomyopathy and found that Hcy lowering reduced mortality and lowered oxidative stress [84]. A few studies have examined the effects of various B-vitamins, the most common being folic acid, on learning and memory, however, these studies did not examine homocysteine levels [8,12].
5. Conclusions
The molecular targets of HCy that are relevant to VCID are not fully defined. While numerous biochemical and cellular actions have been reported, on inspection most resulted from HCy concentrations above the clinically relevant range of 10–100 µM. Reported actions of HCy below 100 µM were primarily vascular. These included: PKC activation and proliferation of vascular myocytes; vessel wall fibrosis and collagenosis; reduced proliferation and suppression of nitric oxide signalling in endothelial cells; heightened superoxide formation, and release of inflammatory mediators (MCP-1, IL-6, IL-8, MMP9); various pro-coagulant actions on different blood components (fibrinogen, ApoA, antithrombin-III). Mouse models of HHcy and the extensive body of epidemiology data indicate a strong relationship between HHcy and cognition. Further, it is clear from the mouse studies that HHcy can influence AD pathology, a further consideration for clinical studies. Future studies are clearly required to fully characterize and identify the key molecular pathways linking Hcy to VCID.
At least 19 clinical trials relevant to VCID have tested HCy-lowering interventions. Many were compromised by the challenges of performing a cognitive clinical trial (trial duration, statistical power, cohort age, B vitamin status). Three trials supported a beneficial effect (FACIT, WAFACS, VITACOG). The question as to whether or not lowering plasma HCy slows cognitive decline requires further well-designed trials. The extensive clinical and preclinical data strongly support Hcy as a key mediator for VCID. The challenge is now to design clinical studies that fully address this link. Preclinical animal model studies will be essential for the identification of the appropriate time-points when Hcy effects can be manipulated and reversed.
Acknowledgments
We are very grateful to AD Smith and H Refsum for their helpful comments on clinical trials. Funding of Dr Hainsworth’s laboratory comes from ADDF (grant no.20140901), Alzheimers Society (PG146/151) and Alzheimers Research UK (PPG2014A-8). Funding for Dr. Wilcock’s work on hyperhomocysteinemia comes from the National Institute of Neurological Disorders and Stroke / National Institutes of Health grant 1R01NS079637.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.