Atherosclerosis determinants in renal disease: how much is homocysteine involved?

Chronic kidney disease (CKD) is characterized by the progressive loss of renal function, leading to the gradual retention of compounds normally excreted or actively metabolized by the body, termed uremic toxins as a whole [1], among which the protein-bound moieties, such as indoxyl sulfate, asymmetric dimethylarginine, p-cresol and homocysteine, are much investigated at present.

Patients with CKD show a markedly increased cardiovascular risk. For example, according to the recent meta-analysis by Matsushita et al. [2], the adjusted cardiovascular risk steadily increases when the estimated glomerular filtration rate (eGFR) goes below 75 mL/min. Half of these patients die of cardiovascular disease before reaching the final stages and therefore are the most interesting, considering their relative number [3, 4]. Therefore, CKD is considered a coronary heart disease equivalent. On the other hand, in the general population, cardiovascular risk can be explained, at least partially, by an impairment in renal function. In other words, reduced renal function accounts for some of the cardiovascular risk present in the general population. The cardiovascular risk associated with CKD cannot be accounted for in its entirety by the traditional risk factors [5].

Homocysteine is an amino acid, which does not enter protein composition, and is largely elevated in homocysteinuria (>100 µmol/L with respect to 10–12 µmol/L present in healthy individuals), accounting for the high cardiovascular risk present in this inherited disease. In the general population, the ability of hyperhomocysteinemia to cause a cardiovascular risk increase is still debated. This concept is sustained by observational studies analysing the effects of mild hyperhomocysteinemia, as well as by experimental data obtained in various models. However, the Mendelian randomization studies, based on a common methylenetetrahydrofolate reductase polymorphism, are negative. That is, plasma homocysteine levels within the normal range do not seem to influence cardiovascular risk [6]. The last fundamental piece in this mosaic consists of the intervention studies, which utilize folates in order to lower homocysteine levels. Folate is a B vitamin that forces the metabolism of homocysteine through the remethylation pathway, aside from other numerous functions (Figure 1). Also relevant to one-carbon metabolism, we include participation of folate derivatives in the biosynthesis of the purine ring and deoxythymidilate (as DNA building blocks), and, through S-adenosylmethionine (SAM), methylation of DNA (one of the main epigenetic mechanisms for the control of gene expression) as well as RNA, proteins and various small molecules (Figure 1). The folate studies have also been negative with respect to cardiovascular risk [9], but it can be stated that a positive effect on stroke and cognitive dysfunction (Huo et al. [10], and the introduction of the paper by Spence et al. published in this issue), and even cardiovascular risk, is possible, especially in subgroups such as Asian populations and unfortified populations [11–13].

FIGURE 1:

The interlinks among folate metabolism, the methionine–homocysteine cycle and methyl transfer reactions. In the methionine–homocysteine cycle (light grey background upper right), SAM is synthesized by SAM synthetase from methionine and ATP. SAM is the common methyl donor for ∼50 different methyltransferases that recognize various methyl acceptors, including macromolecules (nucleic acids and proteins), as well as a number of small molecules (involved, e.g., in the formation of creatine from guanidinoacetate, phosphatidyl choline from phosphatidyl ethanolamine in the liver, catecholamine biosynthesis). The SAM demethylated product SAH is a powerful competitive methyltransferase inhibitor. Its accumulation is prevented by its rapid enzymatic (SAH hydrolase) in vivo hydrolysis to adenosine and homocysteine. The extent of methyltransferase inhibition exerted by SAH depends on the intracellular [SAM]/[SAH] ratio and on the different Km and Ki values for the various methyltransferases involved, possibly resulting in unbalanced methylation in the presence of hyperhomocysteinemia [7, 8]. In the folate cycle and its role in one-carbon (C1) metabolism (dark background; upper left), homocysteine can be transsulfurated to cysteine or remethylated to methionine; 5-methyltetrahyfolate (5-CH₃THF) is, uniquely, the methyl donor for the latter reaction. THF plays a crucial role in the biosynthesis of DNA building blocks, in that it is strictly related to the biosynthesis of both the purine ring, through 5-formyltetrahydrofolate (5-FTHF) and deoxythimidylate biosynthesis from deoxyuridilate, where 5,10-methylenetetrahydrofolate (5,10-CH₂THF) is used (lighter background; centre left).

Homocysteine is elevated in CKD patients, an elevation that in patients not supplemented with folates is on average three to four times above normal levels. It has to be mentioned that many patients can receive folate supplementation in order to replace folates lost in dialysis (i.e., not with the specific goal of lowering homocysteine), for example, when megaloblastic anaemia is present. Homocysteine levels can be lowered by folates, although not with low doses, and are not normalized. The reduction in homocysteine levels obtained with folates did not reduce cardiovascular risk in these patients [14]. However, it cannot be ruled out that folates display negative metabolic effects that could offset the benefits. It has been shown that folic acid supplementation does not reduce the intracellular concentration of homocysteine, and it may even disturb the physiological regulation of intracellular one-carbon metabolism, by interfering with the SAM inhibitory effect on methylene tetrahydrofolate (THF) reductase activity [15]. In addition, it is possible that other related compounds, such as S-adenosylhomocysteine (SAH), are the real culprits, which can be either unaffected by folate therapy [15, 16], or even increased, as we have shown in dialysis patients [17]. For example, Zawada et al. [18] showed that SAH is associated with traditional risk factors and subclinical atherosclerosis and that renal function is more closely correlated with SAH than with homocysteine. Another issue to consider is the fact that these patients do not normalize homocysteine levels, as mentioned above, perhaps because of a down-regulation in the folate receptor, and this could be another reason why folates are not effective [19]. A Cochrane review in transplant patients, who do normalize levels with folates, was also negative [20]. However, transplant patients (kidney, liver) are subject to various immunosuppressive therapies, which can also influence cardiovascular risk through hypertension, dyslipidemia, diabetes, etc. Therefore they cannot be considered analogous to a renal failure group in which homocysteine can be normalized.

It has been proposed that homocysteine is simply a marker of renal function, that is, the increase in cardiovascular risk seen in renal disease patients can be ascribed to the loss of renal function, which is inevitably associated with homocysteine increase, and with the increase in several other factors, which are the real mediators. Homocysteine would thus be just an ‘expensive creatinine’ [21]. In fact, adjusting for renal function eliminates the relationship between homocysteine and carotid intima-media thickness and flow-mediated dilation [22]. In a large meta-analysis, homocysteine depended on renal function [23]. In reality, homocysteine and creatinine are very different molecules. Creatinine is a small non-protein-bound molecule that is filtered by the glomerulus and partly secreted and reabsorbed by the renal tubule. The shortcomings of creatinine as a measure of glomerular function are well known. Homocysteine instead is a mostly protein-bound molecule in humans, therefore, little is available for filtration. It is taken up by the peritubular basolateral surface and is metabolized intracellularly [24]. In fact, its presence in the urine is negligible [25]. So, if creatinine and homocysteine reflect renal function, they do so for very different reasons.

There is another way to look at this topic, and it is the logic provided by Spence et al. in this issue. In this cross-sectional study, the authors analyse the effects of renal function, measured through eGFR calculated with the Chronic Kidney Disease Epidemiology Collaboration formula, and homocysteine in a substantial group of patients, ∼2000, with creatinine levels between 300 and 15 µmol/L, on total carotid plaque area and stenosis. These two surrogate end points are considered important ones; in particular, total plaque area is a more reliable predictor of events than carotid intima-media thickness. The findings strongly indicate that homocysteine levels account in part for the effects of renal impairment on atherosclerosis, at least in relation to total plaque area. The role of other toxins is also discussed.

The populations studied by Spence et al. are exactly those where cardiovascular risk starts to increase [26]. Most of the patients analysed were those with an eGFR roughly between 100 and 40 mL/min, that includes patients with Stages 1–3. Stage 3 patients have a cardiovascular risk that is double the risk of patients with normal renal function [26]. This population is perhaps the one in which it is most crucial to understand the reasons why there is an increase in cardiovascular risk, and where intervention could be most rewarding.

In the study, homocysteine is more related to total plaque area with respect to stenosis. We would have expected the opposite, since the plaque determinants are elevated low-density lipoprotein cholesterol, oxidative stress and altered reverse cholesterol transport, while stenosis is more related to inflammation, matrix metalloproteinase activity and plaque thrombosis. Homocysteine is classically considered more a prothrombotic than a proatherogenic agent. This finding therefore corroborates the fact that homocysteine also has a proatherogenic effect [27].

The statistics utilized in the article include the so-called mediation analysis (Sobel test). This test relies on the concept that the link between an independent and its dependent variable is explained by a third variable, a mediator. The independent variable influences the mediator, which in turn influences the dependent variable. Three steps are required. First, it must be shown that the independent variable (in our case, renal function) is a significant predictor of the dependent variable (total plaque area). Second, it must be confirmed that the independent variable (renal function) is a significant predictor of the mediator (homocysteine). Third, when the mediator (homocysteine) and the independent variable (renal function) are used simultaneously to predict the dependent variable (total plaque area), i.e. when the mediator is included in the model analysis with the independent variable, the effect of the independent variable is reduced and the effect of the mediator remains significant. The Sobel test applied requires large sample sizes in order to have sufficient power to detect significant effects. In our case, the sample size is amply sufficient. Homocysteine mediated an average 22% (12–31%) of the effect of renal function on total plaque area.

This article, aside from discussing other uremic toxins that are involved in cardiovascular risk, provides us with another fresh perspective, leading hopefully to new fruitful insights in this area.

CONFLICT OF INTEREST STATEMENT

A.F.P. and D.I. received congress travel and research funds from Gnosis.

(See related article by Spence, Urquhart and Bang. Effect of renal impairment on atherosclerosis: only partially mediated by homocysteine. Nephrol Dial Transplant 2016; 31: 937–944)