AGEs/RAGE in CKD: irreversible metabolic memory road toward CVD?


Hidenori Koyama, MD, PhD, Department of Metabolism, Endocrinology and Molecular Medicine, Osaka City University Graduate School of Medicine, 1-4-3 Asahi-machi, Abeno-ku, Osaka 545-8585, Japan. Tel.: +81-6-6645-3806; fax: +81-6-6645-3808; e-mail:


Eur J Clin Invest 2010; 40 (7): 623–635


Background  Cardiovascular disease is the major cause of death in patients with renal insufficiency, accounting for 50% of all deaths in renal replacement therapy patients. Mortality from cardiovascular diseases in these patients is approximately 9% per year, which is about 30 times the risk in the general population. So far, intensive interventions to the general risk factors, such as high LDL-cholesterol or C-reactive protein, have not been successful in improving their cardiovascular outcomes, suggesting that the beneficial effect of risk reduction may be overwhelmed by accumulated risk memorized by long-term exposure to oxidative stress during the progression of renal failure.

Design  In this review, we propose that this irreversible memory effect in renal failure may be mediated by advanced glycation end-products (AGEs).

Results  The generation of AGEs has been implicated to be deeply associated with increased oxidative stress. Moreover, interaction of the receptor for AGEs (RAGE) with AGEs leads to crucial biomedical pathway generating intracellular oxidative stress and inflammatory mediators, which could result in further amplification of the pathway involved in AGE generation. Several lines of evidence suggest that AGEs/RAGE axis can profoundly be involved in cardiovascular diseases. Recent advances in AGEs and RAGE measurements led us to be capable of understanding more about the role of AGEs/RAGE axis as a risk for cardiovascular diseases in patients with renal failure.

Conclusion  AGEs/RAGE axis could be a crucial mediator of oxidative stress in renal failure. RAGE could be not only a useful biomarker, but also a potentially therapeutic target to overcome the accumulated adverse metabolic memory in renal failure.


Lindner et al. first described in 1974 the higher risk for cardiovascular diseases in patients with maintenance haemodialysis [1], showing that as much as 50% of total death was because of cardiovascular complications, among which 80% (40% of total deaths) were from cardiovascular deaths and 75% of cardiovascular deaths (30% of total) was coronary artery disease. This result was supported by several epidemiological findings [2–5]. It appears that despite the absence of cardiac symptoms, patients with chronic renal failure are already in a very high risk group for coronary artery stenosis at the initiation of renal replacement therapy [6–8]; the prevalence of coronary artery disease was 42–63%. In addition, recent evidence also suggests that this process of cardiovascular damage starts very early during progression in well-defined chronic kidney disease (CKD), long before end-stage renal disease (ESRD) is developed (i.e. stages 1–2 according to GFR) [9].

So far, intensive interventions to the classical risk factors, such as high LDL-cholesterol or C-reactive protein, in ESRD patients have been disappointing in improving their cardiovascular outcomes. Die Deutsche Diabetes Dialyse Studie (the 4D study) showed no significant benefit of atorvastatin therapy with regard to a composite cardiovascular end point in patients with type 2 diabetes who were undergoing haemodialysis [10]. Subsequently in AURORA study for patients undergoing haemodialysis, the initiation of treatment with rosuvastatin lowered the LDL cholesterol and C-reactive protein level, but had no significant effect on the composite primary end point of death from cardiovascular causes, nonfatal myocardial infarction or nonfatal stroke [11]. Similar situation is also observed in type 2 diabetic patients; United Kingdom Prospective Diabetes Study (UKPDS) failed to show the effect of intensive glycaemic control for preventing acute myocardial infarction [12]. Of particular importance, however, risk reductions for myocardial infarction and death from any cause were emerged for intense glycaemic control group during 10 years of post-trial follow-up, despite an early loss of glycaemic differences [13]. Epidemiology of Diabetes Interventions and Complications-Diabetes Control and Complications Trial (EDIC-DCCT) also showed that initial 6·5-year intensive insulin therapy resulted in long-term post-trial reduction in cardiovascular events in type 1 diabetic patients [14]. Considering these trials in diabetes into account, in patients with renal failure, beneficial effect of risk reduction may be overwhelmed by accumulated risk memorized by long-term exposure to oxidative stress during the progression of CKD. This phenomenon could be conceptualized as metabolic memory for oxidative stress.

The links between CKD and cardiovascular disease could be numerous, as both share a number of common aetiological factors [15]. However, even after adjusting for classical risk factors (i.e. age, gender, hypertension, diabetes and dyslipidaemia), the CKD is still shown to be significantly associated with cardiovascular events. Therefore, it is likely that non-traditional risk factors, which are accumulated during progression of CKD, identified and as yet unidentified, could be involved as well. Advanced glycation end products (AGEs) are a potential candidate for accumulated non-traditional risk factor associated with the development of long-term complications of ESRD. Evidence has emerged pointing to a potential role of oxidative stress in generation of AGEs in patients with ESRD. Moreover, interaction of the receptor for AGEs (RAGE) with AGEs leads to crucial biomedical pathway generating intracellular oxidative stress and inflammatory mediators, which could result in further amplification of the pathway involved in AGE generation.

Accumulation of AGEs as a cardiovascular risk in patients with ESRD

AGEs are proteins generated by a series of reactions termed the Maillard Reaction. Classically, AGE formation has been described by a nonenzymatic reaction between proteins and glucose [16]. AGEs derive from the spontaneous reaction of carbohydrates with amino group of proteins, which undergo from the formation of reversible products (Schiff base adducts) to the generation of more stable products (Amadori products). Subsequently, complex reactions occur including intermolecular crosslink formation, and cleavage through oxidation, dehydration, condensation, cyclization and other reactions follows, with generation of AGEs through a late reaction characterized by fluorescent and brown coloration and molecular crosslinkage. Importantly, the formation of glycoxidation products, such as CML and pentosidine, is considered to be the result of a chemical reaction dependent on the concentration of both carbohydrate precursors and reactive oxygen species (oxidative stress). In contrast to the diabetic patients, in whom the accumulation of AGEs in protein may be attributable to both glycative and oxidative stress, increased AGEs in ESRD could be primarily regulated by the second mechanism, oxidative stress [17]. ESRD is a condition of increased (intracellular) oxidative stress, indicated by the increased lipid peroxidation and a decrease in the ratio of oxidized glutathione to reduced glutathione [18–20]. The concentration of lipid peroxidation marker, malondialdehyde-lysine, is also increased in uraemic patients [21]. Oxidative stress in these patients has been attributed to the processes of loss of renal function and/or the renal replacement therapy [20,22]. Suboptimal biocompatibilities in dialysis membranes may acutely further aggravate oxidative stress and related endothelial dysfunction [23]. However, even before the start of renal replacement therapy, renal impairment is associated with a state of increased oxidative stress [24]. Moreover, accelerated oxidative stress, together with the decrease in superoxide excavenging capacity, is already present in early stages of CKD as well [25–27], and alterations in antioxidant system components (such as superoxide dismutase and glutathione peroxidase/reductase) gradually increase with the degree of renal failure [28].

Development for measurements of circulating major AGEs, including pentosidine and Nε-carboxymethyl-lysine (CML), led us to further understand their pathophysiological role in various diseases. Other than diabetes mellitus patients, high plasma and tissue levels of AGEs are observed in patients with ESRD. We have measured both plasma pentosidine and CML in 307 type 2 diabetic patients and examined the effect of renal function on their plasma levels. Serum pentosidine levels, but not CML, are markedly increased in diabetic patients when they suffer from advanced CKD (Fig. 1a) [29]. In ESRD patients (= 383), no difference was noted in serum pentosidine levels between those with and without diabetes mellitus (Fig. 1b), which also let us to believe enhance production and accumulation of AGEs in conditions other than hyperglycaemia. Alternatively, as shown in Fig. 1a, effect of diabetes on serum pentosidine levels could be overwhelmed by an impairment of clearance of the molecule through kidney. To further determine whether the effect of diabetes and ESRD on AGE accumulation could be distinct at tissue level, we measured skin autofluorescence in 389 ESRD subjects including 90 diabetics by using the autofluorescence reader (AFR), which recently developed non-invasive device to estimate accumulation of AGEs in humans [30]. This measure was shown to be useful to estimate AGE accumulation in ESRD patients, as skin autofluorescence was correlated with collagen-linked fluorescence (r = 0·71, < 0·001), pentosidine (r = 0·75, < 0·001) and CML (both r = 0·45, < 0·01) in skin biopsies (= 29) [31]. Interestingly, skin AGEs accumulation was significantly higher in diabetic than non-diabetic ESRD patients (Fig. 1b), suggesting that tissue accumulated, but not circulating, AGE level might reflect longer exposure to hyperglycaemic and oxidative stress in the course of diabetes and CKD progression. Local accumulation of AGEs is observed in patients with Alzheimer disease, rheumatoid arthritis, arteriosclerosis, cancer and other diseases, also suggesting the involvement of inflammation and oxidative stress in the formation of AGEs.

Figure 1.

 (a) Serum pentosidine and carboxymethyl-lysine (CML) in diabetic patients with different degrees of renal dysfunction. Chronic kidney disease (CKD) stage was based on glomerular filtration rate (GFR) estimated by MDRD equation: I; GFR ≥ 90, II; GFR 60–89, III; GFR 30–59, IV; 15–29, V; GFR < 15 mL min−1) cited from reference [29]. (b) Skin autofluorescence, but not serum pentosidine, is significantly higher in diabetic than non-diabetic ESRD patients.

AGEs are involved in arterial stiffness resulting from nonenzymatic protein glycation to form irreversible cross-links between long-lived proteins such as collagen and elastin. AGE-linked extracellular matrix is stiffer and less susceptible to hydrolytic turnover, resulting in the accumulation of structurally inadequate matrix molecules. We have measured skin accumulation of AGEs and examined its association with arterial stiffness as measured by pulse wave velocity (PWV) in 120 non-diabetic ESRD patients and 110 age- and gender-matched control subjects with neither renal disease nor diabetes [32]. As shown in Fig. 2a, skin autofluorescence was significantly associated with age in the group of control subjects (Rs = 0·493, Spearman’s rank correlation test), but not in the group of ESRD subjects (Rs = 0·046), suggesting that the effect of age on AGE accumulation is overwhelmed by the effect of uraemia in ESRD subjects. PWV was significantly and positively associated with skin autofluorescence both in the group of control (Rs = 0·246) and that of ESRD subjects (Rs = 0·205) (Fig. 2b). In ESRD subjects, a significant association between skin autofluorescence and PWV was found, independent of age, suggesting that AGE accumulation could be an important predictor for arterial stiffening in ESRD subjects. In contrast to skin autofluorescence, serum pentosidine was not significantly associated with PWV in ESRD and control subjects (H. Koyama, unpublished observations), suggesting that serum and tissue accumulated AGEs may represent distinct pathophysiological conditions. Importantly, experimental studies also showed that inhibition or breaking of AGEs prevents cardiac hypertrophy and arterial stiffness and may restore cardiac function [33]. AGE cross-link breaker is also successfully shown to improve arterial compliance in humans as well [34]. As aortic stiffening is commonly observed pathophysiological aspect of large-artery damage, and is a predictor of all-cause and cardiovascular mortality [35–37], attractive scenario is that irreversible accumulation of AGEs in vessels due to exposure to oxidative stress during progression of CKD results in the increase in arterial stiffness and high risks for cardiovascular events in ESRD patients.

Figure 2.

 Ageing, arterial stiffness and AGEs accumulation. Skin accumulation of AGEs was measured by using the autofluorescence reader, recently developed non-invasive device. Arterial stiffness was measured as pulse wave velocity (PWV). (a) Skin autofluorescence was significantly associated with age in the group of control subjects (Rs = 0·493, Spearman’s rank correlation test), but not in the group of ESRD subjects (Rs = 0·046). (b) PWV was significantly and positively associated with skin autofluorescence both in the group of control (Rs = 0·246) and that of ESRD subjects (Rs = 0·205). Open circles, control subjects; closed circles, ESRD subjects. Cited from Figure in reference [32].

Immunohistochemical analyses reveal that AGE accumulation is predominantly observed in atherosclerotic plaque in the artery in patients with ESRD [38], suggesting that AGEs may also be involved in atherosclerotic plaque formation in addition to arterial stiffening. Alternatively, AGE modifications of lipoproteins may increase vascular deposition of low-density lipoprotein as a consequence of impaired low-density lipoprotein receptor-mediated clearance [39–41], which could be attributable to the augmented atherogenesis. At present, very limited report showed significant association between serum AGEs and carotid intima-media thickness (IMT), a surrogate marker for subclinical atherosclerosis, measured by ultrasound in ESRD patients [42]. In 212 ESRD patients undergoing haemodialysis, the relationship of serum pentosidine or skin autofluorescence with carotid IMT was very weak [43]. Increased AGE levels are also shown to be associated with extensive coronary artery calcification in uraemic patients [44]. Thus, in addition to arterial stiffening, AGE accumulation may also be involved in atherogenic plaque formation in ESRD patients.

Clinical data assessing AGEs as a cardiovascular risk predictor in patients with ESRD are quite limited. Recent observations show that the tissue accumulation of AGEs as estimated by skin autofluorescence has been implicated as a risk predictor for cardiovascular mortality in patients with ESRD as well as diabetes [31,45], although the cohort sizes are small (ESRD; = 109, diabetes; = 117). As these findings are quite attractive and are important pieces in puzzles to clarify underlying mechanisms for cardiovascular events in ESRD patients, their finding should be confirmed in lager cohort. As opposed to these observations for tissue AGEs accumulation, the relationship between serum AGEs and mortality in haemodialysis patients is rather controversial: some observed a strong relationship between serum AGE levels and survival [46,47], whereas others did not [48,49]. Although it is not clear at present how to explain these discrepant observations, different methods to determine the AGE levels (immunoassay versus serum fluorescence), influence of dialysis modality and timing or effects of food and smoking may partly contribute to these results. It is also possible that serum AGEs levels are strongly influenced by renal function, thus not necessarily simply represent metabolic memory for entire period of chronic kidney disease. In contrast, the tissue levels of AGEs are considered to be hardly removed and could be irreversibly accumulated. Alternate explanation could be that high serum AGE levels may reflect better nutritional support (e.g. high serum albumin) and low inflammation (e.g. high C-reactive protein), both of which are known predictors for mortality in ESRD patients [50,51]. Taken altogether, AGEs levels, particularly in tissues, could represent accumulated oxidative stress during the progression of CKD, and their measurements would be extremely useful for stratification of the cardiovascular risks in patients with ESRD.

Receptor for AGEs (RAGE) and atherosclerosis

Receptor for advanced glycation end-products (receptor for AGEs, RAGE) is a multi-ligand cell-surface protein that was isolated from bovine lung in 1992 by the group of Schmidt and Stern [52,53]. RAGE belongs to the immunoglobulin superfamily of cell surface molecules and has an extracellular region containing one ‘V’-type immunoglobulin domain and two ‘C’-type immunoglobulin domains [52,53] (Fig. 3). The extracellular portion of the receptor is followed by a hydrophobic trans-membrane-spanning and then by a highly charged, short cytoplasmic domain, which is essential for intracellular RAGE signalling. RAGE is initially identified as a receptor for CML modified proteins. Three-dimensional structure of the recombinant AGE-binding domain by using multidimensional heteronuclear NMR spectroscopy revealed that the domain assumes a structure similar to those of other immunoglobulin V-type domains [54,55]. Three distinct surfaces of the V domain were identified to mediate AGE-V domain interactions [54]. The site-directed mutagenesis studies identified the basic amino acids, which play a key role in the AGE-binding activities [55]. RAGE also interacts with other non-glycated peptide ligands including S100/calgranulin [56], amphoterin (also termed as high mobility group box 1 protein, HMGB1) [57,58], amyloid fibrils [59], transthyretin [60] and a leukocyte integrin, Mac-1 [61]. The common characteristics of these ligands are the presence of multiple β-sheets [61–63]. RAGE is thought to interact with these ligands through their shared three-dimensional structure.

Figure 3.

 Numerous truncated forms of RAGE. There are three major spliced variants of RAGE: full-length, N-terminally truncated and C-terminally truncated. The C-terminally truncated form of RAGE is secreted from the cell and is named endogenously secreted RAGE (esRAGE). esRAGE has a V-domain, which is essential for binding with ligands and is capable of competing with RAGE signalling as a decoy receptor. There are other forms of soluble RAGE (sRAGE) that are cleaved from cell-surface RAGE by a shedase ADAM10 or MMP9. The ELISA assay for sRAGE measures all soluble forms including esRAGE in human plasma, whereas the ELISA for esRAGE measures only esRAGE, using polyclonal antibody raised against the unique C-terminus of the esRAGE sequence. Cited from reference [29].

Activation of RAGE is associated with diabetic microvascular complications including nephropathy, retinopathy and neuropathy. By use of RAGE-overexpressing and deficient mice, RAGE is shown to have pathological role in both early- and advanced-phase diabetic nephropathy [64,65]. In humans, RAGE gene polymorphisms are also shown to be related to the occurrence of diabetic nephropathy [66,67]. RAGE expression in the peripheral nervous system rises cumulatively and relates to progressive pathological changes, and mice lacking RAGE have attenuated features of neuropathy and limited activation of potentially detrimental signalling pathways [68,69]. In mesangial [70] and endothelial [71] cells, RAGE activation results in a burst of reactive oxygen species (ROS). The exact mechanism for this is unknown but is thought to involve nicotinamide adenine dinucleotide phosphate (reduced) [NAD(P)H] oxidase [72], which could alone contribute to cellular oxidative stress and dysfunction. In addition, RAGE signals via phosphatidylinositol-3 kinase (PI-3K), Ki-Ras and the MAPKs, Erk1 and Erk2 [70,73]. These signalling pathways initiate and sustain the translocation of nuclear factor-κB (NF-κB) from the cytoplasm to the nucleus in a number of cell types including circulating monocytes and endothelial cells [74–76], leading to prolonged inflammation, resulting in a RAGE-dependent expression of proinflammatory mediators such as monocyte chemoattractant protein-1 (MCP-1) and vascular cell adhesion molecule-1 (VCAM-1) and organ damage [56,77–79].

Inappropriate chronic inflammation associated with progressive CKD reflects sustained activation of inflammatory cells, like monocytes/macrophages, where accumulation of AGEs may play important role through binding with the RAGE. It has been shown that in peripheral monocytes from subjects with varying severity of CKD, RAGE expression is closely associated with worsening CKD and is strongly correlated with plasma levels of pentosidine, a marker for AGEs [80]. In ESRD subjects with high-grade inflammation, stimulation of mononuclear cells with AGE-modified human serum albumin causes a rapid dose-dependent rise in NF-κB activity that could be completely blocked by an anti-RAGE antibody [81]. Linden et al. recently demonstrates an association of excess AGE burden with increased peripheral blood mononuclear cell RAGE mRNA and in vivo endothelial dysfunction in patients with CKD [82]. They hypothesized that endothelial dysfunction in CKD may be partly mediated by AGE-induced inhibition of endothelial nitric oxide synthase through RAGE activation. At the site of endothelium, in contrast to normal endothelial cells which do not constitutively express RAGE, arterial and capillary endothelial cells of uraemic patients do express RAGE [83]. Indeed in endothelial cells in vitro, AGEs positively regulate human RAGE gene transcription through NF-κB-mediated mechanisms [84]. Thus, enhanced RAGE expression in ESRD may amplify AGEs-induced perturbation and contribute to systemic and local vascular inflammatory disease causing atherosclerotic and non-atherosclerotic vascular lesions in CKD patients.

Possible involvement of RAGE in atherosclerosis has been extensively examined in diabetic human and animals. It has been demonstrated that RAGE is upregulated in human diabetic atherosclerotic plaques, the expression of which colocalized with inflammatory markers such as cyclooxygenase-2 (cox-2) and matrix metalloproteinases (MMPs), particularly in macrophages at the vulnerable regions of the atherosclerotic plaques [85,86]. In a murine model of accelerated atherosclerosis, apolipoprotein (apo) E null mice, induction of diabetes was found to be associated with a significant increase in atherosclerotic lesion area at the aortic sinus compared with non-diabetic apo E null mice [87]. This diabetes-associated atherosclerotic lesion has been found to exhibit increased accumulation of AGEs and S100/calgranulins and enhanced expression of RAGE [88]. Of importance, daily treatment of the mice with genetically engineered murine soluble RAGE (sRAGE) suppressed diabetes-associated acceleration of lesion area and complexity with the effect independent of glycaemic and lipid profiles [87]. Similarly, sRAGE has been found to prevent progression of atherosclerosis in apo E null mice with insulin-resistant type 2 diabetes (db/db background). Of note, vascular inflammatory phenotype such as overexpression of VCAM-1, tissue factor and matrix metalloproteinase in the mice was also prevented by administration of sRAGE [89].

Involvement of RAGE in accelerated atherosclerosis in diabetes has recently been confirmed by using RAGE-deficient mice [90]. It is becoming apparent that RAGE is also involved in atherogenesis even in non-diabetic conditions in a mouse model of atherosclerosis [90–92]. Harja et al. demonstrated critical roles for RAGE and its ligands in vascular inflammation, endothelial dysfunction, and atherosclerotic plaque development, and highlighted that endothelial RAGE and its ligands mediate vascular and inflammatory stresses that culminate in atherosclerosis in the vulnerable vessel wall [91]. Similarly by using apo E-deficient mice, Soro-Paavonen et al. [90] recently showed that in diabetic conditions, RAGE deficiency significantly decreased the atherosclerotic lesion formation which was associated with attenuation of leukocyte recruitment, decreased expression of proinflammatory mediators and reduced oxidative stress in aorta. In their report, RAGE deficiency significantly suppressed the atherosclerotic lesion area even in non-diabetic condition, which was associated with aortic inflammatory gene expression. We have also recently shown that RAGE-deficiency attenuates progression of atherosclerosis in non-diabetic condition through inhibiting adiposity and increasing serum adiponectin [92]. sRAGE administration was also found to significantly stabilize atherosclerotic lesion area and complexity in non-diabetic apo E null mice [93]. Thus, these reports implicated the functional role of RAGE in endothelial cells in atherogenesis, possibly through regulation of inflammatory signalling adiposity.

It is still not clear whether RAGE accumulation is increased in atherosclerotic lesions in uraemia. However, chronic renal failure markedly accelerates atherogenesis in apo E-deficient mice [94,95], whereas blockade of RAGE reduces the proatherogenic effects of uraemia, possibly through a systemic decrease in oxidative stress [96].

C-terminally truncated form of RAGE (sRAGE) and cardiovascular diseases

Numerous truncated forms of RAGE have been described [97–103] (Fig. 3). Two major spliced variants of RAGE mRNA, N-terminal and C-terminal truncated forms, have been most extensively characterized [98]. The N-truncated isoform of RAGE mRNA codes for a 303-amino-acid protein lacking the N-terminal signal sequence and the first V-like extracellular domain. The N-truncated form is incapable of binding with AGEs, as the V-domain is critical for binding of the ligand [52]. The N-truncated form of RAGE appears to be expressed on the cell surface similar to the full-length RAGE, although its biological roles remain to be elucidated [104]. The C-terminal truncated form of RAGE lacks the exon 10 sequences encoding the transmembrane and intracytoplasmic domains [98]. This spliced variant mRNA of RAGE encodes a protein consisting of 347 amino acids with a 22-amino-acid signal sequence and is released from cells. This C-truncated form is now known to be present in human circulation and is named endogenous secretory RAGE (esRAGE) [98]. esRAGE was found to be capable of neutralizing the effects of AGEs on endothelial cells in culture [98]. Adenoviral overexpression of esRAGE in vivo in mice reverses diabetic impairment of vascular dysfunction [105]. Thus, the decoy function of esRAGE may exhibit a feedback mechanism by which esRAGE prevents the activation of RAGE signalling. It has also been suggested that some sRAGE isoforms that could act as decoy receptors may be cleaved proteolytically from the native RAGE expressed on the cell surface [106], suggesting heterogeneity of the origin and nature of sRAGE. This proteolytic generation of sRAGE was initially described as occurring in mice [107]. Recent findings by screening chemical inhibitors and genetically modified mouse embryonic fibroblasts suggest that a disintegrin and metalloprotease 10 (ADAM10) and MMP 9 as membrane proteases responsible for RAGE cleavage [108,109]. ADAM is known as a shedase to shed several inflammatory receptors and can be involved in regulation of RAGE/sRAGE balance. Thus, the molecular heterogeneity of the diverse types of sRAGE in human plasma could exert significant protective effects against RAGE-mediated toxicity. However, the endogenous action of sRAGE may not be confined to a decoy function against RAGE-signalling. In HMGB1-induced arthritis model, for example, sRAGE is found to interact with Mac-1 and act as an important proinflammatory and chemotactic molecule [110]. Further analyses are warranted to understand more about the endogenous activity of sRAGE.

As sRAGE and esRAGE may be involved in feedback regulation of the toxic effects of RAGE-mediated signalling, recent clinical studies have focused on the potential significance of circulating sRAGE and esRAGE in a variety of pathophysiological conditions. First, total sRAGE levels were shown to be significantly lower in non-diabetic patients with angiographically proven coronary artery disease than in age-matched healthy controls [111]. In contrast to non-diabetic population, evidence so far with regard to the association between sRAGE and vascular disease in diabetes is contradictory. A study shows that serum sRAGE is positively associated with coronary artery disease in type 1 diabetic patients [112]. However, a recent longitudinal study in small numbers of type 1 diabetic subjects (= 47) showed that annual increase in carotid IMT was inversely associated with arithmetic average of plasma sRAGE [113], although this report failed to show significant relation between basal sRAGE and IMT. Similarly, the findings regarding plasma levels of the sRAGE in diabetes are quite confusing; many reports showed increased levels, whereas substantial contradictory findings also exist. These are summarized in our recent reviews [29,114]. These discrepancies might be the results of co-existing renal insufficiency, which markedly influences plasma sRAGE levels as discussed later.

Following the development of an ELISA system to specifically measure human esRAGE [115], we measured plasma esRAGE level and cross-sectionally examined its association with atherosclerosis in 203 type 2 diabetic and 134 non-diabetic age- and gender-matched subjects, and found that esRAGE levels were inversely correlated with carotid and femoral IMT when analysed in all subjects [116]. Stepwise regression analyses revealed that plasma esRAGE was the third strongest and an independent factor associated with carotid IMT, following age and systolic blood pressure. However, when non-diabetic and diabetic groups were separately analysed, inverse correlation between plasma esRAGE level and IMT was significant in non-diabetic population only, but not observed in type 2 diabetic subjects [114]. No association of plasma esRAGE with IMT in diabetes was also reported in other study with 110 Caucasian type 2 diabetic subjects [117]. In contrast, significant inverse relationship between plasma esRAGE and carotid IMT was reported in type 1 diabetic subjects [118,119]. Recently, the same research group also longitudinally examined the predictive significance of plasma esRAGE and sRAGE on progression of carotid atherosclerosis and found that low circulating esRAGE level as well as sRAGE level was an independent risk factor for the progression of carotid IMT in type 1 diabetic subjects. In Chinese type 2 diabetic patients, plasma esRAGE is recently shown to be decreased in angiographically proved patients with coronary artery disease than those without it [120]. In contrast to plasma sRAGE levels, we and other groups have consistently found that plasma esRAGE level is significantly lower in type 1 and type 2 diabetic patients than in non-diabetic controls [116,118]. Recent study with large numbers of type 2 diabetic and non-diabetic subjects in Chinese population (= 1320) also confirmed the findings [120].

Important component that can affect plasma sRAGE is the presence of CKD, which may explain controversial findings of plasma sRAGE in diabetes (Table 1). Circulating sRAGE levels have been shown to be increased in patients with decreased renal function, particularly those with ESRD [117,121,122]. We have also carefully examined the effect of renal function on plasma esRAGE in diabetic subjects [114,123]. Although plasma esRAGE levels in type 2 diabetic subjects without CKD are lower than non-diabetic controls, it gradually elevated in accordance with the progression of CKD [114] or diabetic nephropathy [123]. Thus, plasma sRAGE and esRAGE are markedly affected by the presence of CKD, which might make the previous findings regarding comparison between non-diabetic and diabetic subjects quite controversial. It remains to be determined whether the increase in plasma esRAGE in CKD is caused by decreased renal function alone or whether esRAGE levels are upregulated to protect against toxic effects of the RAGE ligands. Successful kidney transplantation resulted in significant decrease in plasma sRAGE [124], implying that the kidneys play a role in sRAGE removal.

Table 1.    Levels of circulating soluble RAGE in patients with CKD
 Carotid atherosclerosisInverse association[125]
 Chronic kidney diseaseIncreased[117,121,122,136,137]
 Chronic kidney diseaseIncreased[114,123,126,137]
 Cardiovascular outcomeInverse association[126,129]

Association of circulating sRAGE or esRAGE with vascular diseases in CKD or ESRD subjects is an important topic to be elucidated. So far, only limited reports are available (Table 1). To examine clinical significance of circulating esRAGE on cardiovascular outcomes in ESRD subjects, we performed a cohort study consisting 206 ESRD patients including 35 diabetics. Of note, even in patients with ESRD, plasma esRAGE levels at base line were still inversely associated with body mass index (Fig. 4a), as was shown for healthy and diabetic population [116]. Plasma esRAGE was also significantly associated with plasma adiponectin in these populations (Fig. 4a). Moreover, 98 ESRD patients were also examined for subclinical atherosclerosis by using carotid artery ultrasound and found that plasma esRAGE was significantly and inversely associated with carotid IMT (Fig. 4b). This observation was in good agreement with Basta’s recent finding that plasma sRAGE was inversely associated with IMT or plaque numbers in carotid artery in 142 CKD patients [125]. This cohort was followed up for the median of 111 months, and 132 patients were confirmed to be alive on haemodialysis and 74 to have died, 34 of which was due to fatal cardiovascular events at the end of the follow-up. Importantly, the subjects in the lowest tertile of plasma esRAGE levels exhibited significantly higher cardiovascular mortality, but not non-cardiovascular mortality (Fig. 4c), even though the plasma esRAGE levels at baseline were higher in ESRD subjects than in those without kidney disease [126]. In the subpopulation of non-diabetic subjects alone, low circulating esRAGE level was also a predictor of cardiovascular mortality, independent of the other classical risk factors. It is not known at present how esRAGE is involved in cardiovascular mortality. In our ESRD cohort, neither plasma pentosidine nor carboxymethyl-lysine level predicted cardiovascular mortality. Moreover, the inverse correlation between low circulating esRAGE level and cardiovascular mortality was not dependent of plasma AGEs levels. Thus, the protective effect of esRAGE against cardiovascular mortality may not be entirely dependent on neutralization of toxic AGEs. Other endogenous ligands for RAGE, such as S100A12, may also be involved in the function of esRAGE. The plasma level of S100A12 has been shown to be increased in diabetes and inversely correlated with serum sRAGE level [127,128]. Similar to our findings, low sRAGE levels are also shown to be associated with a 2–3 times higher risk for mortality especially after correction for creatinine clearance in a cohort of 591 patients after renal transplantation [129]. Thus, low circulating esRAGE level appears to be a predictor for atherosclerosis and cardiovascular events in patients with ESRD.

Figure 4.

 (a) Plasma esRAGE level is inversely associated with body mass index (BMI), and positively with serum adiponectin in ESRD subjects. (b) Inverse association of plasma esRAGE with carotid intima-media thickness (IMT) in ESRD patients. (c) Low plasma esRAGE level is associated with cardiovascular, but not with non-cardiovascular mortality (cited from reference [126]).

It is important to determine whether currently available pharmacological agents can regulate plasma sRAGE or esRAGE. Forbes et al. [130] showed that inhibition of angiotensin-converting enzyme (ACE) in rats increased renal expression of sRAGE, and that this was associated with the decreases in expression of renal full-length RAGE protein. They also showed that plasma sRAGE levels were significantly increased by inhibition of ACE in both diabetic rats and in human subjects with type 1 diabetes. Thus, one attractive scenario is that the protective effect of ACE inhibition against progression of renal dysfunction is mediated through regulation of RAGE versus soluble RAGE production. Other potential agents that may affect circulating soluble RAGE include the thiazolidinediones [131] and statins [132,133], both of which are known to modulate the AGEs-RAGE system in culture. A randomized, open-label, parallel group study was performed with 64 participants randomized to receive add-on therapy with either rosiglitazone or sulfonylurea to examine the effect on plasma soluble RAGE [134]. At 6 months, both rosiglitazone and sulfonylurea resulted in a significant reduction in HbA1c, fasting glucose and AGE. However, significant increases in total sRAGE and esRAGE were only seen in the rosiglitazone group. Thus, thiazolidinedione could be one promising candidate which increases circulating levels of esRAGE and sRAGE. Tam et al. recently reported changes in serum levels of sRAGE and esRAGE in archived serum samples from a previously randomized double-blind placebo-controlled clinical trial that explored the cardiovascular effects of atorvastatin in hypercholesterolemic Chinese type 2 diabetic patients, and found that atorvastatin can increase circulating esRAGE levels [135]. Finally, we have started the randomized clinical trial comparing the effect of pioglitazone with glimepiride on plasma sRAGE and esRAGE, expression of RAGE on peripheral mononuclear cells and RAGE shedase gene expression in type 2 diabetic patients (UMIN000002055). This study will be of particular importance to understand the regulatory mechanisms of sRAGE and esRAGE in clinical setting.

The findings discussed here implicated the pivotal role of AGEs-RAGE system in initiation and progression of cardiovascular disease in patients with CKD (Fig. 5). There appears to be vicious positive feedback mechanisms in AGE accumulation through AGE-RAGE induced oxidative stress in the course of CKD progression. sRAGE or esRAGE could be potentially important in cutting this vicious cycle. Measuring tissue AGE accumulation could be useful to estimate how long and what extent the patient have been exposed to oxidative stress. Moreover, plasma sRAGE or esRAGE could serve as a novel biomarker for estimation of the risk stratification of atherosclerotic disorders. Further examination of the molecular mechanisms underlying RAGE and esRAGE regulation will provide important insights into potential targets for the prevention and treatment of cardiovascular diseases.

Figure 5.

 Vicious positive feedback mechanisms for AGE accumulation in CKD. AGE/RAGE induced oxidative stress could further amplify the synthesis of AGEs. sRAGE or esRAGE, which could be a natural decoy inhibitor for RAGE, may be protective for this vicious cycle, thus inhibit atherogenesis and cardiovascular diseases.


The authors thank all colleagues in the Osaka City University Graduate School of Medicine and Kanazawa University Graduate School of Medical Science for their unflagging support of our projects. We apologize to all colleagues whose work we could not cite other than indirectly through other publications, due to limitation of space. This study was supported in part by a Grant-in-Aid for Scientific Research (20591067 to H.K.) from Ministry of Education, Culture, Sports, Science and Technology, Japan.


Department of Metabolism, Endocrinology and Molecular Medicine, Osaka City University Graduate School of Medicine, 1-4-3 Asahi-machi, Abeno-ku, Osaka 545-8585, Japan (H. Koyama, Y. Nishizawa).