Atherosclerosis in Chronic Kidney Disease.

Patients with chronic kidney disease (CKD) are at an increased risk of premature mortality, mainly from cardiovascular causes. The association between CKD on hemodialysis and accelerated atherosclerosis was described >40 years ago. However, more recently, it has been suggested that the increase in atherosclerosis risk is actually observed in early CKD stages, remaining stable thereafter. In this regard, interventions targeting the pathogenesis of atherosclerosis, such as statins, successful in the general population, have failed to benefit patients with very advanced CKD. This raises the issue of the relative contribution of atherosclerosis versus other forms of cardiovascular injury such as arteriosclerosis or myocardial injury to the increased cardiovascular risk in CKD. In this review, the pathophysiogical contributors to atherosclerosis in CKD that are shared with the general population, or specific to CKD, are discussed. The NEFRONA study prospectively assessed the prevalence and progression of subclinical atherosclerosis (plaque in vascular ultrasound), confirming an increased prevalence of atherosclerosis in patients with moderate CKD. However, the adjusted odds ratio for subclinical atherosclerosis increased with CKD stage, suggesting a contribution of CKD itself to subclinical atherosclerosis. Progression of atherosclerosis was closely related to CKD progression as well as to the baseline presence of atheroma plaque, and to higher phosphate, uric acid, and ferritin and lower 25(OH) vitamin D levels. These insights may help design future clinical trials of stratified personalized medicine targeting atherosclerosis in patients with CKD. Future primary prevention trials should enroll patients with evidence of subclinical atherosclerosis and should provide a comprehensive control of all known risk factors in addition to testing any additional intervention or placebo.

C hronic kidney disease (CKD) is currently defined as abnormalities of kidney structure or function, present for >3 months. The most widely used criteria are a decreased glomerular filtration rate (GFR <60 mL/min per 1.73 m 2 ) or urinary albumin creatinine ratio ≥30 mg/g. Patients with CKD are at an increased risk of premature mortality, independently of the criterion (GFR or albuminuria) used for diagnosis [1][2][3][4][5] (Figure 1). In fact, the thresholds of GFR and urinary albumin creatinine ratio that define CKD are also associated with an increased risk of all-cause and cardiovascular death at any age and the risk associated with albuminuria is independent from GFR. 6,7 As recently emphasized, individuals may have CKD even when estimated GFR (eGFR) is normal (>90 mL/min per 1.73 m 2 ) if albuminuria is pathological, and individuals diagnosed of CKD just based on the albuminuria criterion are also at increased risk of cardiovascular death despite normal eGFR. 4 The addition of eGFR and urinary albumin creatinine ratio significantly improved the discrimination of cardiovascular outcomes (cardiovascular mortality, heart failure, coronary disease, and stroke) beyond traditional risk factors in general populations. 8 However, there is still controversy as to the relative contribution to this increased cardiovascular disease (CVD) mortality of atherosclerosis versus other forms of vascular injury, most notably arteriosclerosis, atherosclerosis-independent vascular calcification, myocardial fibrosis, capillary rarefaction, arrhythmia, and disordered hemostasis. 1,9,10 Beyond any academic discussions, unraveling the drivers and relative contributions of these processes may impact on the therapeutic approach to vascular risk in CKD patients. Recent reviews have emphasized the failure of most clinical trials designed to reduce cardiovascular mortality in patients with CKD, with the notable exception of statin-ezetimibe for nondialysis patients with CKD. 1,2,10 Still, the benefits of statins were limited: a 17% decrease in the risk of severe atherosclerosisrelated events, no impact on mortality, and no impact on dialysis patient outcomes. 11 While an association between CKD on hemodialysis and accelerated atherosclerosis was described in the seventies of last century, 12 more recently it has been suggested that the atherosclerosis risk actually increases in CKD stages G1 and G2, remaining high thereafter, and as CKD progresses, that atheroma plaque becomes a smaller component of the overall cardiovascular risk. 1 This very early occurrence of the increased CKD-associated atherosclerosis risk is an interesting concept, since in the absence of pathological albuminuria, a GFR >60 mL/min per 1.73 m 2 is not currently considered CKD in most large-scale epidemiological studies. Thus, most large observational studies lack albuminuria data and define CKD just based on GFR <60 mL/min per 1.73 m 2 , which is a major shortcoming. The association of increased atherosclerosis risk with very early CKD places the focus on pathological albuminuria as a potential driver of atherosclerosis. However, these recent reviews did not discuss atheroma plaque assessment by direct imaging methods. We now review the current state-of-the-art of the role of atherosclerosis in the increased cardiovascular morbidity and mortality of patients with CKD, focusing on reports from the past 15 years, since major changes in the care of patients with CKD may have yielded older assessments of atherosclerosis in CKD outdated. We searched MEDLINE for articles published during the past 10 years (January 2008-December 2018). Search terms were "Atherosclerosis", "Cardiovascular event", "mortality", "Chronic kidney disease", "Renal failure. " Additionally, we searched the reference list of articles identified by the search strategy and selected those from the 21st century that were considered relevant. Reports of direct assessment of plaque, either by vascular ultrasound, coronary imaging, or indirect methods like ankle-brachial index (ABI) were selected because they are more sensitive to detect real atheroma plaque formation, and to differentiate it from medial vascular calcification often seen in these patients, and regarded as a different process.

DEFINING ATHEROSCLEROSIS IN CKD
A key issue is how atherosclerosis is defined (Figure 2). In the earliest report of accelerated atherosclerosis in hemodialysis CKD, it was defined as a presumably atherosclerotic event, such as a myocardial infarction. 12 This has obvious shortcomings, since arrhythmia-related sudden cardiac death, occurring with higher frequency in the perihemodialysis session period, is considered a major contributor to cardiovascular mortality. 2 However, nonwitnessed, out-of-hospital sudden cardiac death due to primary arrhythmia may not be easily distinguished from

Highlights
• Chronic kidney disease is associated with an increase in atherosclerotic burden from early stages. • Progression of chronic kidney disease is associated with progression of atherosclerosis. • Specific risk factors related with chronic kidney disease are involved in the accelerated atherosclerosis observed in these patients. • Subclinical atherosclerotic detection by arterial ultrasound could improve cardiovascular risk prediction in chronic kidney disease patients.
death due to myocardial infarction. A different, functional approach may define atherosclerosis as events that can be prevented by statin therapy. The advantage of this definition is the focus on clinically relevant episodes, providing evidence of causality since the response to targeted interventions is explored. Defining the impact of atherosclerosis in CKD using this definition would require randomized clinical trials (RCTs) targeting known drivers of atherosclerosis, such as high LDL (low-density lipoprotein) cholesterol levels. Indeed, this approach has been tested, and there are lingering doubts on whether the failure of such trials points to a different pathogenesis of CVD in CKD or to the existence of a point-of-noreturn. Finally, a morphological definition of subclinical atherosclerosis may also be used and be based on an increased intima-media thickness (IMT), on the presence of plaques in carotid and femoral ultrasonography, or computed tomogaphy scan evidence of coronary calcification. Issues with this definition include nonatherosclerosis related increases in IMT, as have been reported, for example, for Fabry disease, 13 nonatherosclerosis-related coronary calcification, and a lower clinical relevance than definitions based on events. Most of the imaging information is derived from the assessment of coronary calcification by computed tomography scan (reviewed in Mathew et al 10 ). While these studies suggest a higher atherosclerosis burden in CKD patients within general population cohorts, autopsy studies suggest that in advanced CKD calcification may be located in the medial layer and this may not be as representative of atherosclerosis as in the general population. However, these autopsy reports were published before 2003 and coincided with the widespread use of calcium-based phosphate binders and massive doses of calcitriol. Assessing the relative prevalence of atherosclerosis in CKD using the morphological definition would require large population-based studies comparing CKD with non-CKD patients. An advantage of this definition is the possibility to select patients who have already developed atherosclerosis for clinical trials of atherosclerosis-targeting interventions, instead of the current approach of prescribing atherosclerosistargeting interventions based on population-based risk scores. The current approach may result in prescribing atherosclerosis-targeting interventions to individuals who   do not have atherosclerosis and, thus, may derive little benefit from the intervention. At the clinical trial level, current approaches may limit our ability to quantify the real benefit of drugs that do decrease atherosclerosis events in individuals who do have atherosclerosis. Since a key issue in CKD-associated atherosclerosis is the paucity of RCT with positive results, we chose to focus on atherosclerosis defined as the presence of plaque as a potential approach to address current limitations.

PATHOPHYSIOLOGY OF VASCULAR INJURY IN CKD: CLASSICAL AND EMERGING FACTORS
Classical risk factors for cardiovascular mortality and morbidity do not have the same predictive value in patients with CKD, particularly in advanced CKD, as in the general population. 14 This suggests that additional pathogenic processes may be at play that may be reflected in emerging CKD-related risk factors (Table 1). However, the specific molecular pathways triggered by CKD-related factors remain poorly characterized. Despite these suggestions, very few genes were differentially expressed in nondiseased arteries from advanced patients with CKD as compared to normal renal function individuals. 15 In fact, among genes consistently modulated in vascular smooth muscle cells (VSMCs), just 23 genes were downregulated and 8 upregulated, and not always gene expression was accompanied by parallel changes in cellular protein content. Thus, both the mRNA expression and the protein content of the alpha subunit of the hypoxia-inducible factor 3 increased as a consequence of uremia, whereas vimentin content increased with a decreased expression of its mRNA. 15 Four large groups of factors and processes are thought to be involved at different CKD categories ( Figure 3).

Changes in Mineral Bone Metabolism
Some excellent reviews have analyzed the contribution of mineral bone metabolism derangements to CKD-related vascular injury and support a general hypothesis of cardiovascular calcium deposition. 10,16,17 However, there is still controversy on whether medial vascular calcification often observed on patients with CKD can be considered as an atherosclerotic process similar to the one observed in general populations or a totally different process specific for CKD. 18 Thus, in large arteries, vascular calcification is associated to the presence of atheroma plaques, whereas in small arteries, which are rich in VSMC, medial calcification without plaque is often observed. 19 CKD-mineral and bone disorder (CKD-MBD), which among other features is characterized by vascular calcification, starts developing very early in the course of CKD. 20 Indeed, loss of kidney production of the anti-aging and phosphaturic factor Klotho is already observed in CKD G1 (see Figure 1 for CKD categories), when GFR is still normal, but albuminuria may be present. 21 Albuminuria itself is a direct suppressor of Klotho expression, as is local or systemic inflammation. 22,23 Klotho is a coreceptor for the phosphaturic hormone FGF23 (fibroblast growth factor 23), and Klotho deficiency favors the increase of FGF23 levels already observed in CKD category G2 and vascular calcification. 21 Vascular calcification is common in CKD. In general terms, the instability of calcium and phosphate ions in the circulation plus the abnormal differentiation of VSMC to osteoblast/condroblast-like cells are the basis of vascular calcification. Ionic changes are because of the loss of the kidney contribution to calcium-phosphate homeostasis and to the inadequate administration of supplements or drugs (eg, calcium supplements and macrodoses of vitamin D) to correct these changes. On top of this, the balance between factors that promote or prevent soft tissue calcification is skewed to promotion of calcification in CKD. The abnormal dedifferentiation of VSMC depends on multiple extracellular stimuli such as uremic toxins, bioactive metabolites, paracrine signals from infiltrating macrophages or even autocrine signals after resident cell stimulation. [24][25][26][27] Interestingly, 2 of the main actors of CKD-MBD in more advanced stages of CKD, phosphate and calcitriol, can directly modify the VSMC phenotype. VSMCs have both Pit-1 and Pit-2 phosphate transporters and develop an osteoblastic phenotype in the presence of high phosphate concentrations. [28][29][30][31] Some miRNAs may be involved in the regulation of phosphate-induced vascular calcification. Aortas exposed to high phosphate concentrations exhibit a time-dependent differential pattern of expression of 10 miRNAs. 32 miRNA 125b, suggested as a marker of vascular calcification, is not one of these phosphate-dependent miRNAs. 33 The impact of calcitriol is likely dose-dependent and modulated by the interaction with calcium and phosphate availability and other environmental factors. Both a direct calcitrioldriven increase in VSMC calcification and the opposite effect have been described. 34,35 In any case, large doses of calcitriol increase the gut absorption of calcium and phosphate in CKD patients that have a limited ability to excrete the excess of both in urine.
Recent reports from the NEFRONA study (Observatorio Nacional de Atherosclerosis en NEFrologia), 36,37 an observational multicenter prospective study designed to evaluate the prevalence and outcomes of subclinical atheromatosis in CKD patients, as well as the contribution of vascular imaging to a more precise and individualized cardiovascular risk stratification, showed that higher serum phosphate levels were independently associated with plaque prevalence in every CKD category studied (G3, G4, and G5D) whereas higher hsCRP (high-sensitivity C-reactive protein) emerged in G4-G5 and lower 25(OH)D levels in G5D. However, the assessment of peripheral artery disease by indirect methods (ABI) yielded similar but not completely overlapping results. Thus, higher serum phosphate in CKD G3 and lower 25(OH)D levels in CKD G4 and G5D were independently associated with an abnormal ABI. 38 In our view, this emphasizes the importance of directly assessing atherosclerosis in the form of plaques, rather than indirect features. Phosphate is especially interesting given its place as a current major therapeutic target, the consistent association with atheromatosis in all CKD categories studied and its clear interaction with sex. 39 Thus, the association of higher serum phosphate with the presence of plaque started at lower serum levels in men than in women (3.5 versus 5.0 mg/dL; Figure 4C), suggesting that safe serum phosphate levels may differ depending on sex. Major observational studies, clinical trials, and meta-analysis addressing the association of serum phosphate with adverse cardiovascular outcomes or death and the impact of therapy for hyperphosphatemia should be re-analyzed from a gender perspective.
An additional mechanism of vascular calcification relates to calcium deposition in the intimal layer of the artery. 19 Endothelial cell biology is also modified by CKD, but these cells do not develop an osteoblast-like  Figure also shows relationship with early CKD categories (defined by pathological albuminuria categories A2, A3 when glomerular filtration rate [GFR] is preserved, ie, GFR categories G1-G2), and with more advanced CKD categories, defined by GFR <60 mL/min per 1.73 m 2 , ie, GFR categories G3-G5, independently from the magnitude of albuminuria. Very frequently, the magnitude of albuminuria increases as GFR decreases. In GFR categories G1-G2, changes are more dramatic when proteinuria is most severe, ie, nephrotic syndrome. For details, see text. FGF indicates fibroblast growth factor; LDL, low-density lipoprotein; MBD, mineral and bone disorder; and PTH, parathyroid hormone.
phenotype. Thus, the instability of calcium salts in the circulation and endothelial structural and functional changes in response to noxious stimuli, related or not to CKD, are thought to be the main drivers of intimal calcification.

Lipids and Atherosclerosis in CKD
Increased lipid concentrations are one of the main drivers of atherosclerosis in the general population, but lipid deposition in vascular walls as a consequence of changes in plasma concentration is only one aspect of the problem. Thus, chemically modified lipoproteins, such as oxidized lipoproteins, interact with specific receptors in vascular or circulating cells and contribute to vascular damage. 40,41 In CKD, the quantitative changes in serum classical lipids are not particularly proatherogenic. 42,43 In this regard, triglyceride levels were associated with subclinical atherosclerosis in CKD G3, whereas in CKD G4-5 only total cholesterol showed a weak association with plaque presence. 38 Finally, LDL-cholesterol levels were less predictive of coronary risk among CKD patients than in general populations. 44 However, qualitative changes in lipid profile observed in CKD could be associated with a higher atherogenic profile. Thus, CKD has been associated with an accumulation of VLDL (very-low-density lipoprotein) particles, a reduction of LDL particle size and changes in the cholesterol and triglyceride content in LDL and HDL (highdensity lipoproteins; which gain triglycerides and lose cholesterol). 45 Additional striking qualitative changes in the composition of lipoproteins, include chemical modifications in LDL and HDL, such as glycation, oxidation, and carbamylation, which are associated to activation of pathogenic pathways and receptors such as the proinflammatory lectin-like oxidized LDL receptor-1 46-50 Indeed, HDLs from CKD patients lose the vascular protective functions and even may acquire deleterious actions on vascular biology (reviewed in Ferro et al 51 ). In CKD, Lp(a) levels are increased in people with large apo(a) isoforms and in patients with large protein losses, as those with nephrotic syndrome or on peritoneal dialysis. 52

Inflammation and Vascular Damage
Inflammation is now considered one of the main mechanisms of atherosclerosis, and CKD is characterized by systemic inflammation. [53][54][55][56] Proinflammatory changes in patients with advanced renal disease include increased A, Prevalence of subclinical atherosclerosis, in each chronic kidney disease (CKD) category in the NEFRONA study (Observatorio Nacional de Atherosclerosis en NEFrologia), crude data. *P<0.01 vs control subjects. Adapted from Arroyo et al. 151 Copyright ©2014 (see: http:// creativecommons.org/licenses/by/4.0). B, Adjusted odds ratio (OR) for prevalent subclinical atherosclerosis in each CKD category in the NEFRONA study using individuals with normal renal function as the reference. Prevalence was adjusted for sex, age, and diabetes mellitus. Derived from Betriu et al. 38 C, Adjusted OR for prevalent subclinical atherosclerosis depending on baseline serum phosphate level, stratified by sex. Prevalence was adjusted for age, diabetes mellitus, smoking status, CKD stage, and CRP (C-reactive protein). Derived from Martín et al. 39 concentrations of circulating CRP and cytokines, an activated phenotype of circulating monocytes (eg, CD14+CD16+ monocytes) and resident vascular cells, and increased synthesis of inflammation-triggered reactive oxygen species. 54,[57][58][59][60][61] In fact, CRP levels were associated with the presence of plaque in CKD patients 38 and with coronary heart disease in the general population. 62 Multiple cytokines have been proposed to be involved in the genesis of the uremia-related proinflammatory status, and a detailed analysis of all of them is out of the scope of this article. A recent review summarizes the published findings in the field of uremia and cytokines and, more interestingly, analyzes the pathogenic consequences. 55 Similar changes have been described in non-CKD atherosclerosis patients, and the question is to what extent reports on non-CKD CVD patients are contaminated by CKD patients since frequently albuminuria levels are not reported or even assessed to diagnose or exclude CKD.
The proinflammatory milieu in CKD is thought to depend both on increased synthesis and decreased clearance of mediators of inflammation. 55 Thus, for instance, calcium-phosphate crystals induce a proinflammatory response in macrophages. 63 Uremic toxins are also involved in the activation of inflammatory responses. Protein-bound uremic toxins generated from gut microbiota metabolites, such as p-cresyl-sulfate and indoxyl sulfate, stimulate proinflammatory responses in macrophages and vascular and parenchymal cells, [64][65][66][67] and promote atherogenesis in genetically modified mice. 68 The finding of a significant correlation between p-cresyl sulfate levels and carotid damage in hemodialysis patients supports the contribution of this toxin to the development of vascular disease in CKD patients. 68 Moreover, these toxins stimulate the cross-talk between macrophages and endothelial cells, 69,70 promoting vascular wall infiltration by inflammatory cells. A particular subset of circulating monocytes, CD14+CD16+ cells, has been proposed to play a relevant role in vascular damage. 71 Conversely, loss of anti-inflammatory factors may also contribute. Thus, kidney-generated calcitriol or Klotho has antiinflammatory properties, [72][73][74] suggesting that a deficit of kidney-generated molecules in the course of CKD also contributes to an inflammatory phenotype.
Cell senescence has more recently been involved in the development of uremia-related vascular damage. The phenotype of senescent cells is characterized by a diminished regenerative ability, loss of physiological functions, accumulation of DNA and oxidative damage, and a particular secretory phenotype commonly known as senescence-associated secretory phenotype. 45,75,76 Vascular wall cell senescence was observed in adult 77 and, even more interestingly, in pediatric CKD patients. 78 In pediatric samples, the senescent phenotype was, at least partially, dependent on calcium-phosphate dysregulation, and through SAP, favored vascular calcification. Damaged VSMC generate exosomes that induce calcification of normal cells. 79 Senescence not only depends on deranged calcium-phosphate metabolism. In cultured human aortic VSMCs and in rats, indoxyl sulfate induces a well-defined senescent phenotype, 80 supporting that both mineral disturbances and uremic toxins are involved in vascular senescence. Additionally, other nonvascular cells may also display senescent features. The previously cited CD14+CD16+ human monocytes have an increased ability for infiltrating vascular walls and exhibit telomere shortening and increased β-galactosidase activity, evidence of a senescent phenotype. 61 The Central Role of the Endothelium in the Genesis of CKD-Related Vascular Injury The endothelium has a critical role in maintaining the normal structure and function of the cardiovascular system. This involves complex mechanisms such as the balance between endothelial regeneration and death, the regulation of interactions with intravascular fluids and cells, and the cross-talk with resident vascular cells, including other endothelial cells. 81,82 The endothelial cell phenotype is responsive to physiological and pathophysiological demands and adjusts the release of metabolites that act as extracellular messengers accordingly.
Uremic toxins interfere with normal endothelial function. Increased inorganic phosphate levels cause endothelial dysfunction, characterized by phenotypic changes, decreased viability, and senescence. [83][84][85] Moreover, vitamin D metabolites regulate endothelial function, by modulating the synthesis of endothelial vasoactive factors 86,87 and the interaction with circulating leukocytes. 88 Other treatments used to control secondary hyperparathyroidism, such as calcimimetics, may also modulate endothelin-nitric oxide homeostasis in endothelial cells by decreasing endothelin-1 synthesis. 89 Carbamylated LDL caused oxidative stress, accelerated senescence and death in endothelial and endothelial progenitor cells, 90,91 potentially contributing to the decreased endothelial regenerative ability that characterizes CKD. 92,93 Asymmetrical dimethylarginine decrease nitric oxide synthesis. 94,95 Similarly, highly protein-bound uremic toxins stress endothelial cells, promoting oxidative damage and adhesion molecule expression, and decreasing proliferation and endothelial progenitor cell-dependent neovascularizarion. [96][97][98] Interestingly, CKD-related endothelial damage results in release of microvesicles and specific miRNAs, such as miRNA92a, that may further promote vascular damage. 99,100 Concerning factors that modulate endothelial growth, plasma VEGF (vascular endothelial growth factor) concentration is increased in patients with advanced CKD, both off or on dialysis, a finding that does not support a pathogenic role of VEGF deficiency in the pathogenesis of CKD-associated endothelial damage. 101 However, there is the theoretical possibility that uremic VEGF has been posttraslationally modified, as it is the case for other proteins, and is no longer as active as nonuremic VEGF.

Proteinuria and CVD
Severe proteinuria leading to nephrotic syndrome has long been known to be associated with a proatherogenic lipid profile characterized by extremely high LDL-cholesterol levels and by high Lp(a) levels. 51,102 More recently, lower levels of proteinuria and even the milder degree formerly called microalbuminuria and now termed A2 albuminuria (>30 mg/g of urinary creatinine) have been associated in epidemiological studies with an increased risk of cardiovascular mortality, heart failure, coronary disease, and stroke and this was independent from the impact of GFR. 8,103 In fact, the discrimination improvement of adding albuminuria to traditional risk factors was larger than that of adding eGFR. 8 The precise molecular links of mild albuminuria with CVD remain unclear. However, albumin only spills into urine when the reabsorptive capacity of proximal tubular cells has been exceeded. That is, even mild albuminuria (and we should remember that in healthy individuals albuminuria is <5 mg/g) is associated with proximal tubular cells overload with filtered proteins and this causes tubular cells stress, decreases Klotho production and triggers an inflammatory response. 22,104 The critical role of proximal tubular cell injury in both CVD and CKD progression is supported by research and clinical data: single cell transcriptomics studies disclosed that differential gene expression in proximal tubular cells is most closely related to GFR and recent RCTs of inhibitors of the proximal tubular cell transporter SGLT2 (sodium-glucose cotransporter 2) have shown protection from both CVD and CKD progression in patients with diabetes mellitus (DM). 105 Thus, new links have emerged between pathological albuminuria, a kidney inflammatory response and CVD, including atherosclerotic events. As discussed below, albuminuria has been associated with evidence of atherosclerosis in children, but studies have several limitations. [106][107][108]

INSIGHTS FROM ANIMAL MODELS FOR CKD-RELATED ATHEROSCLEROSIS
The relationship between CKD and atherosclerosis is supported by results of animal models of atherosclerosis in CKD performed in genetically modified mice (either ApoE or LDL receptor null mice) or in rabbits, with severe (5/6 nephrectomy) or mild (uninephrectomy) renal impairment. All reports so far detected an increase in the severity of atherosclerosis. [109][110][111] Additionally, studies in which arterial segments with atheroma plaque were transplanted into normal or CKD mice demonstrated that CKD inhibits regression of the disease. 112,113 This suggests that CKD-related alterations in the internal milieu promote atherosclerosis. However, the individual molecular contributors are unclear. Small molecule uremic toxins may be implicated since an oral adsorbent that binds uremic toxins decreased atherosclerosis in uremic mice. 114 Potential mechanism of action includes recruitment of the inflammatory response. Thus, the uremic toxin indoxyl sulfate increased atherosclerosis by inducing proinflammatory macrophage activation through triggering Notch signaling. 115 Macrophages obtained from CKD mice show a higher migratory activity in response to chemoattractants 116 and a higher cellular accumulation of cholesterol, independently of extracellular concentrations due to decreased ABCA1 (ATP-binding cassette transporter 1) levels. 117 The low ABCA1 levels may be mediated by NF-κB (nuclear factor kappa-B) activation by the renin-angiotensin system (RAS) or other factors. Indeed, RAS blockade decreased atherosclerosis in experimental CKD. 118 CKD activates other proinflammatory molecules in macrophages, such as MPO and IL (interleukin)-17A, both of which increase plaque burden in experimental animals. 111,119 Beyond low ABCA1 levels, CKD causes both quantitative and qualitative changes in circulating lipid levels that can be atherogenic. 45 Thus, LDL carbamylation favors atherosclerosis in animal models. 92 Furthermore, Lp(a) levels are increased in CKD and Lp(a) overexpression in uremic mice increased the extent of atherosclerosis. 120 Alterations in endothelial, smooth muscle, or even periarterial cells may be linked to CKD-MBD and inflammation and contribute to CKD-induced atherosclerosis. Low active vitamin D levels increased adhesion molecules in endothelial cells 88 and induced a senescent phenotype in smooth muscle cells, 26 a feature found in the neointima of atheroma plaques. 121 High levels of iron, clinically derived from transfusions or intravenous iron, also increased adhesion molecules expression in uninephrectomized mice. 122 Epigenetics may also contribute. Low miR-142-3p levels have been associated with CKD-mediated endothelial dysfunction, as a mimic miRNA restored acetylcholineinduced vascular relaxation of uremic arteries. 123 On the contrary, dual inhibition of endothelial miR-92a-3p and miR-489-3p decreased CKD-induced atherosclerosis. 124 Furthermore, extracellular matrix and elastin degradation by cathepsin S contribute to CKD-induced plaque formation. 125 Finally, RAS activation in periaortic adipose tissue increased atherosclerosis in CKD mice. 126

PREVALENCE OF ATHEROSCLEROSIS IN CKD
The prevalence of atherosclerosis in CKD has been studied using diverse methodological approaches and in adults as well as in children.

Coronary Artery Disease
Postmortem studies have confirmed that CKD patients have extensive CAD with a higher proportion of calcified plaques, greater media thickness and the presence of medial calcification compared with those without CKD. [127][128][129] Several approaches have been used to assess coronary atherosclerosis in living patients with CKD. Angiographic studies revealed that a reduced kidney function is significantly associated with CAD severity, independently of other risk factors. [130][131][132] Similarly, the COURAGE (Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation) cohort study revealed that triple vessel CAD was more prevalent in patients with CKD than in non-CKD. 133 Noninvasive assessment of CAD based on the coronary artery calcium score improves cardiovascular risk prediction in subjects with CKD. 134 Indeed, coronary artery calcium score increases as GFR declines 132,135 and is associated with a worse outcome. 136 Intravascular ultrasound has been extensively used to characterize the impact of renal function on coronary plaques volume and composition. Impaired renal function was related to a higher lipid core and a lower fibrous volume in coronary plaques. 137 Furthermore, intravascular ultrasound showed that the plaque composition of coronary culprit lesions evolved from necrotic core-rich to extensively calcium-rich plaques as renal function declined. 138 Similarly, patients with CKD undergoing percutaneous coronary intervention showed a more extensive and severe atherosclerosis with a greater necrotic core and less fibrous tissue in coronary plaques. 139 An optical coherence tomography study showed that lower GFR and DM were independent risk factors for a larger lipid index. Patients with CKD had a larger lipid index with a higher prevalence of calcium, cholesterol crystals, and plaque disruption compared with non-CKD patients. 140

Ankle-Brachial Index
Clinically, an ABI of ≤0.90 is used to diagnose peripheral artery disease. 141 It is considered as a risk indicator of generalized atherosclerosis 142 and an independent predictor of cardiovascular or all-cause mortality in the general population. [143][144][145] By contrast, an ABI of >1.40 is used as a marker of increased vascular stiffness 146 and arterial calcification. 147 Peripheral artery disease is the most common cause of amputation in patients in endstage renal disease. 148 In addition, ABI <0.90 is significantly correlated with increased risk of vascular access failure in hemodialysis patients. 149 Among patients with CKD, the percentage of ABI >1.40 rises as renal disease progresses. 150,151 A recent meta-analysis of 18 studies enrolling 60 467 participants showed that an abnormally high ABI was associated with an increased cardiovascular and all-cause mortality (4.28-fold and 67% higher in CKD/hemodialysis patients versus 84% and 45% higher in the general population). 152 In addition, the negative impact of an abnormal ABI on all-cause mortality (hazard ratio [HR], 2.26; 95% CI, 1.60-3.18) and cardiovascular mortality (HR, 3.06; 95% CI, 2.30-4.07) was more pronounced among patients with hemodialysis compared with nondialyzed CKD patients. 153

Carotid and Femoral Atherosclerosis
In the general population, several longitudinal studies have shown that an increased carotid IMT (cIMT) is associated with a higher cardiovascular risk, independently of known traditional risk factors as reviewed elsewhere. 154,155 An association between kidney dysfunction and carotid atherosclerosis was observed in several general population studies. Indeed, impaired renal function is strongly associated with a faster change in cIMT. 156 In middle-aged males and postmenopausal women a mild impairment in renal function was independently associated with an increased cIMT. 157 cIMT can predict ischemic events and long-term mortality in predialysis 158 and hemodialysis patients. 159,160 Renal replacement therapy affects cIMT. In this regard, a recent cross-sectional analysis showed that peritoneal dialysis is independently associated with lower IMT values than hemodialysis in. 161 However, although cIMT values were higher in CKD G3 than in non-CKD controls, they did not differ between more advanced CKD G categories and non-CKD controls, indicating that cIMT conveys information that differs from the assessment of plaques. 151 Similar to the general population, plaque presence in carotid arteries is more prevalent among diabetics, smokers, males, and older CKD patients. 38 Moreover, the presence of carotid atherosclerosis is higher in patients on dialysis than in predialysis patients. 162 In this regard, the NEFRONA study showed a clear trend towards higher prevalence of plaques with more advanced CKD categories, confirming that severity of CKD is an independent factor influencing subclinical atheromatosis ( Figure 4A and 4B). 38 In addition, the cause of CKD also influences the prevalence of atheromatosis. Thus, patients with diabetic nephropathy are at higher risk of subclinical atheromatosis than patients with other causes of CKD. 163 The presence of atherosclerosis in femoral arteries has been less explored. Moreover, to the best of our knowledge, no study apart from the NEFRONA study, addressed simultaneously the characterization of subclinical atherosclerosis in carotid and femoral artery territories (6 carotid and 4 femoral territories). Femoral artery plaque was also significantly associated with CKD prevalence, even among those with normal ABI. 164 However, although higher values of both cIMT and femoral IMT are associated with atherosclerosis burden in patients with CKD, cIMT had a higher sensitivity for detecting subclinical atherosclerosis in this population. 165 Arterioscler

CKD and Atherosclerosis in the Pediatric Population
There is less information on the relationship between CKD and atherosclerosis in children, and the size of the cohorts is usually small, precluding firm conclusions. However, the evidence to date suggests that indeed albuminuria and decreased eGFR are associated with atherosclerosislike changes from childhood. More than 20 years ago, endothelium-dependent dilation of the brachial artery was reported impaired in children (mean age 12 years) with CKD (mean GFR 17.5 mL/min per 1.73 m 2 ) who did not have co-existing risk factors for atherosclerosis. 166 cIMT was higher in children (mean age 9 years) with CKD (mean eGFR 39 mL//min per 1.73 m 2 ) than in healthy controls, and it was also increased in kidney transplanted children (mean age 14 years), but the impact of albuminuria was not assessed. 167,168 In adolescents (median age 15 years), eGFR correlated with the Pathobiological Determinants of Atherosclerosis in Youth, a validated risk score to estimate the probability of advanced coronary atherosclerotic lesions in young adults. 169 An independent association of albuminuria with early features of atherosclerosis is also suggested by studies in children with type 1 DM or nephrotic syndrome. In a large study of 406 adolescents (mean age 14 years) with type 1 DM, aortic IMT but not cIMT was independently associated with albuminuria. 170 However, a cross-sectional association between nephrotic range proteinuria and cIMT has been repeatedly reported in small series of children and adolescents with idiopathic nephrotic syndrome or systemic lupus erythematous. 106,107 Prospective studies to validate these findings are ongoing. 108

PROGRESSION OF ATHEROSCLEROSIS IN CKD
Progression of atherosclerosis in CKD has been evaluated in observational studies, and RCTs have addressed whether targeting known atherosclerosis-related factors in the general population modifies the incidence of cardiovascular events in patients with CKD.

Observational Studies
Two distinct processes can be recognized. A slow asymptomatic buildup of atherosclerotic plaques, and symptomatic occlusive thrombus formation triggered by plaque rupture or endothelial erosion. Theoretically, CKD can influence one or both of these processes. Few observational studies, outside the NEFRONA cohort, 36,37 have addressed the progression and prognostic value of subclinical atherosclerosis in CKD patients. In a relatively contemporary Canadian study, the rate of carotid plaque progression was lower in patients with advanced CKD (eGFR category G4) than in patients with better renal function (eGFR category G2-3a). Median plaque growth rate was 0.4 mm 2 /y for patients with creatinine clearance <23 mL/min per 1.73 m 2 versus 5.0 mm 2 /y for patients with creatinine clearance >43 mL/min per 1.73 m 2 , which is counterintuitive given preclinical studies linking the uremic environment to vascular injury. 171 Thus, despite the overall agreement on the association of CKD with an increased incidence of cardiovascular events, there is less information on the prevalence and progression of atherosclerosis in CKD and no consensus on whether advanced CKD is associated with accelerated or slower progression of atherosclerosis. 1,2,171 In this regard, the HR for risk for CVD events conferred by 3 atherosclerosis-related parameters (coronary artery calcium score, IMT, and ABI) was lower for individuals with CKD than in those without CKD, but the baseline risk was higher for CKD. Coronary artery calcium score was associated with the highest risk for CVD, heart failure, or coronary heart disease events both in CKD and non-CKD populations but was a significant stroke predictor only in non-CKD individuals. 134 Although in end-stage renal disease patients, progression of atheromatosis is better associated to cardiovascular events than the baseline number of plaques, 172 progression of atheromatosis is more likely in patients with CKD who already have atherosclerosis. 173 In the NEFRONA study, atheromatosis progressed within 2 years in 69% (95% CI, 62-76) of CKD patients with baseline presence of plaque and in 40% (95% CI, [35][36][37][38][39][40][41][42][43][44] of patients without baseline plaque. Interestingly, patients with progressive CKD (defined as a doubling of serum creatinine or the start of renal replacement therapy within 2 years) were more likely to show atheromatosis progression ( Figure 5). 173 Therefore, and although the design of the study does not address cause-effect relationships, there was a close link between the progression of both CKD and CVD. This is a key piece of information that should guide future primary prevention RCTs in patients with CKD: one-third of the CKD population does not have baseline subclinical atheromatosis and more than half of these do not develop atheromatosis within 2 years. Inclusion of this population without baseline subclinical atheromatosis in RCTs that use as inclusion criterion the high cardiovascular risk associated to CKD will result in a population in which no response to therapy is expected in the intervention or placebo arms, thus diluting the sample and decreasing the study power.
Similar to the general population, traditional risk factors such as age, sex, smoking habit, and blood pressure have been associated to progression of atheroma burden in CKD. [173][174][175][176] An stratified analysis by CKD category revealed that higher phosphate and lower 25(OH)D levels were associated with plaque progression in CKD category G3 (Table 2). 39,88,173,[177][178][179] This has practical interest, since these patients are not recommended to be referred to a nephrologist by KDIGO guidelines (Kidney Disease: Improving Global Outcomes), but they already display CKD-specific risk factors for atheromatosis progression. 7 Moreover, a diagnosis of dyslipidemia, higher ferritin or IMT values, total cholesterol levels below 180 mg/dL, higher uric acid and lower 25(OH)D levels were also independent predictors of faster atheromatosis progression. 173,[180][181][182] The fact that higher cholesterol levels were only significantly associated with progression in dialysis patients, but counterintuitively, with slower progression, underscores the marginal role of cholesterol in progression of atheromatosis in advanced CKD stages and is in line with the reverse epidemiology phenomenon already described for cholesterol in dialysis patients. 183 The fact that high ferritin, and not transferrin saturation, is associated with progression in dialysis patients points to a role of inflammation, and not iron deficiency, in progression of atherosclerosis. The association with DM also merits discussion since glycosylated hemoglobin targets are higher for CKD patients with advanced microvascular complications than for the non-CKD population. 184 While most patients with DM just have small reductions of GFR and no proteinuria, those CKD with patients advanced microvascular complications are the ones at the highest cardiovascular risk. However, therapeutic nihilism and glycosylated hemoglobin levels above 7.9% are risk factors for mortality in dialysis patients, suggesting that glycosylated hemoglobin targets in CKD patients may be revised once safer glucose-lowering agents are available. 2 An additional analysis confirmed that the modality of dialysis was not associated with atheromatosis progression. 185 As it was the case for plaque progression, CKDrelated factors independently associated to an accelerated progression of IMT in the NEFRONA cohort. These factors included higher serum phosphate in CKD G3 and G5D, and low 25(OH)D levels in G4. Additionally, parathyroid hormone levels over the recommended range in CKD G4 and outside the recommended range (either higher or lower) in CKD G5D were also associated with IMT progression. Parathyroid hormone is of interest because, although a statistically significant association was only observed in G4 and G5 categories, the pattern suggested a U-shaped relationship. Therefore, values under or over the recommended range show a clear trend to be associated to atheromatosis progression and could be explaining part of the association of out of range parathyroid hormone levels with adverse patient outcomes. 186,187 Therefore, the factors associated with progression of IMT and plaque show some agreement. Among Novel factors found in the NEFRONA study (Observatorio Nacional de Atherosclerosis en NEFrologia) being independently associated with progression of subclinical atherosclerosis and cardiovascular events in CKD patients. 173,186,237,243,276,294 CKD indicates chronic kidney disease; FGF, fibroblast growth factor; IMT, intima-media thickness; PTH, parathyroid hormone; and TWEAK, TNF-related weak inducer of apoptosis.
CKD-related factors, higher phosphate levels are associated with progression of both, IMT and plaque burden in stage G3. This is especially important as changes in CKD-MBD related factors (eg, decreased Klotho production, increased FGF23 levels) occur early in CKD patients and are related to phosphate handling. Although FGF23 levels were not assessed in NEFRONA, it is likely that they are higher in patients with higher serum phosphate as higher serum phosphate leads to higher FGF23 levels. Indeed, higher FGF23 levels were a marker of subclinical atherosclerosis in different studies [188][189][190][191][192][193] and could be involved in the early increase of atherosclerosis in CKD.
Prior atherosclerotic events are associated to a higher risk of recurrent major adverse cardiovascular events. [194][195][196] Data from the SMART study (Second Manifestations of Arterial Disease; 7216 patients) and REACH Registry (Reduction of Atherothrombosis for Continued Health; 48322 patients), 2 prospective cohorts of patients with clinical atherosclerotic disease were used to predict the risk of major adverse cardiovascular event according to American Heart Association/American College of Cardiology scores. The highest incidence of recurrent major adverse cardiovascular event (stroke, myocardial infarction, or cardiovascular mortality) was found in patients with an eGFR <45 mL/min per 1.73 m 2 , suggesting that indeed, CKD is a risk factor for reccurrence. 197 Thus, there is evidence for a counterintuitive situation: advanced CKD is associated with slower progression of atherosclerosis and a less clear link between evidence of atherosclerosis and cardiovascular events, yet the risk of a cardiovascular event is increased, likely because of additional changes in the uremic milieu that increase the risk of an event once atherosclerosis is present or because of nonatherosclerotic cardiovascular injury.

Interventional Studies
RCTs of pharmacological interventions to reduce the risk of cardiovascular events in CKD patients have frequently targeted factors known to promote atherosclerosis in the general population. Regrettably, some interventions to reduce traditional risk factors have failed to decrease cardiovascular events and mortality in patients with advanced CKD. However, the primary outcome definitions were not always limited to atherosclerotic complications.
Lipid-lowering drugs very effectively prevent cardiovascular events in patients with normal renal function or CKD, but they lack efficacy in patients with advanced CKD on dialysis (Table 3). 11,[198][199][200][201]204 Trials comparing different statins with placebo obtained similar results. Moreover, the combination of simvastatin plus ezetimibe in the SHARP (Study of Heart and Renal Protection) RCT also failed to prevent cardiovascular events in dialysis patients, although it was beneficial in nondialysis CKD. 11 A meta-analysis of 51 099 patients concluded that lipidlowering drugs reduce all-cause and cardiovascular mortality and events in nondialysis CKD patients but not in dialysis patients. 203 A post hoc analysis of the 4D RCT in diabetics on dialysis disclosed that atorvastatin reduced the risk of mortality and cardiac events only in patients with low intestinal cholesterol absorption, as reflected by cholestanol-to-cholesterol ratios. 204 Still, suboptimal trial design may have negatively impacted the ability to detect benefit. The 4D study primary end point included cardiovascular events not primarily related to atherosclerosis such as sudden death. 198 A recent critical analysis of current lipid-lowering approaches in CKD emphasized that in SHARP, the subgroups of patients that tended to derive the most benefit from the intervention also had the largest residual risk, raising the question of whether target-oriented approaches to statin prescription (which differs from the current fire-and-forget paradigm enshrined in KDIGO guidelines and based on SHARP results) or use or more potent drugs such as PCSK9 (proprotein convertase subtilisin/kexin type 9) inhibitors may have resulted in further improvement in outcomes. 226,227 These high-risk patients included those with baseline LDLcholesterol ≥100 mg/dL, urinary albumin creatinine ratio >30 mg/g, and body mass index ≥28 kg/m 2 , the interaction being significant for body mass index. 11,227 Antihypertensive drug intervention studies have reported mixed results about atherosclerotic complications outcomes in patients with CKD, some studies finding a beneficial effect on cardiovascular events. [205][206][207][209][210][211] A meta-analysis of 5 RCT in hemodialysis patients, observed a decreased risk of cardiovascular events with the use of antihypertensive drugs. This was likely because of an improved control of blood pressure since it was not significant when the meta-analysis included normotensive patients receiving placebo. 206 However, the optimal blood pressure target in dialysis is still under discussion. 228,229 The use of telmisartan versus standard therapies on top of ACE (angiotensin-converting enzyme) inhibitors reduced both overall and cardiovascular mortality in hemodialysis patients with chronic heart failure. 208 However, heart failure is not a primary atherosclerotic event. Two RCTs that compared atenolol versus lisinopril, highlighted the importance not only of blood pressure control but also of the effect on the sympathetic system in dialysis populations. 210,211 Thus, treatment with atenolol resulted in a greater reduction of aortic pulse wave velocity compared to ACE inhibitors 210 and reduced cardiovascular morbidity and all-cause hospitalizations in hemodialysis patients. 201 Many pharmacological interventions have targeted CKD-MBD. Several meta-analyses support a beneficial effect of noncalcium-phosphate binders and, specifically sevelamer on all-cause mortality and coronary calcification when compared with calcium-based phosphate binders. 215,230,231 The calcium receptor agonist cinacalcet did not significantly reduce the risk of death and cardiovascular morbidity either in hemodialysis or in CKD G3 to G5 patients. 213  patients, which reported that paricalcitol improved endothelium-dependent vasodilatation, as compared with the placebo group. 217 However, this is a surrogate end point and in RCTs in nondialysis patients for indications other than hyperparathyroidism, 2 µg/d paricalcitol was associated with higher risk of a hypothetical composite end point of dialysis, myocardial infarction, stroke, or death. 232 Erythropoiesis-stimulating agents may be harmful in terms of cardiovascular events, mainly stroke, in CKD patients, especially if used at high dose in iron-deficient patients and attempting to normalize hemoglobin levels without actually normalizing them. [218][219][220] Additional interventions that failed to find any benefit include the use of antioxidant or homocysteine-lowering therapies. 2  However, studies so far have been of small size and generally of suboptimal quality. 222 The reasons behind this lack of clinical efficacy as opposed to multiple preclinical studies supporting the role of oxidative stress in tissue injury is unclear. However, the clinical results are not surprising given that chronic antioxidant therapy is not in clinical use for any chronic condition with the potential exception of N-acetylcysteine for lung disease, but this case is confounded by the mucolytic properties of the compound. Potential explanations include an insufficient understanding of redox regulation in humans and sup-optimal doses or compounds explored. Antiplatelet drugs did not protect from cardiovascular mortality but reduced the risk of myocardial infarction in CKD patients. 224 In any case, whether recommendations on antiplatelet drugs developed from studies in the general population can be extrapolated to patients with advanced CKD or dialysis patients is hotly debated, given the increased risk of bleeding and a potentially different risk:benefit ratio. 233,234 Finally, inflammation has also been targeted. The CANTOS (Canakinumab Anti-inflammatory Thrombosis Outcomes Study) secondary prevention trial enrolled inflamed patients and showed that reducing inflammation through IL-1β inhibition significantly reduced vascular risk, beyond that achievable with lipid lowering and the reduction in cardiovascular mortality was especially striking (31%) among patients who achieved the largest reductions in hsCRP. 235 In this regard, in the 1875 CANTOS patients with baseline eGFR <60 mL/min per 1.73 m 2 , canakinumab reduced the risk of major adverse cardiovascular events (HR, 0.82; 95% CI, 0.68-1.00; P0.05) and, as in the full trial population, this was especially noteworthy among those achieving on-treatment hsCRP levels below 2 mg/L (HR, 0.68; 95% CI, 0.53-0.86; P=0.0015). 225 There has been much debate about the reasons for the failure of multiple trials to meet the primary end point, and they are likely multifactorial. Among potential reasons, the intervention may be really not effective. CKD impacts on multiple biochemical and functional parameters and may render ineffective those interventions that are protective in the general population. As previously discussed, not only newer pathogenic factors emerge in the context of CKD, but the relationship between factors contributing to cardiovascular injury may change as a consequence of CKD. In some cases, the outcome was not appropriate for the intervention studied. Thus, cardiovascular mortality may not be an appropriate end point for interventions targeting atherosclerosis, given the high impact of arrhythmia and sudden death in more advanced CKD. A pessimistic view of the issue sustains that in patients with advanced CKD, it may be too late to observe any benefit. An alternative explanation is that the time-tobenefit may be longer. In general, cardiovascular intervention trials in CKD patients have enrolled lower numbers of patients and have been of shorter duration than general population cardiovascular trials. As an example, a trial of sevelamer versus placebo with primary end point mortality randomized 2103 patients with a mean follow-up of 20 months. The trial did not meet the primary end point, but the risk of death was 23% lower in those over the age of 65 years. 212 Since the drug was hypothesized to potentially decrease the risk of death through protection from vascular calcification, a mean follow-up of <2 years may not have been enough for differences in vascular calcification to develop and for these differences to be of sufficient magnitude to impact mortality, except in patients with more severe baseline calcification, that is, those over the age of 65 years. However, one of the key issues that in our opinion has hampered major clinical trials is the lack of a holistic approach to patient care, that provides optimal care for all potential cardiovascular risk factors to both placebo and intervention arms. This may be compounded by the rigidity of some clinical trial protocols. An egregious example of both failures is the TREAT trial (Trial to Reduce Cardiovascular Events With Aranesp Therapy) of darbepoetin in type 2 DM patients with CKD, with a primary end point of the composite outcomes of death or a cardiovascular event (nonfatal myocardial infarction, congestive heart failure, stroke, or hospitalization for myocardial ischemia) and of death or end-stage renal disease. 220 Hemoglobin increased in the placebo arm (instead of dropping from around 10.5-9.0 g/dL as expected) pointing to the enrollment of patients who had iron-deficient anemia rather than uremic anemia. The standard of care would have called for correction of iron deficiency before enrollment. The fact that even this basic trial-related standard of care was not met, suggests that a holistic approach to cardiovascular risk reduction was not provided (eg, <60% of participants were on statins). As a consequence of these failures and of the rigidity of the trial protocol, intervention patients received huge doses of darbepoetin, unheard of in routine clinical practice: the median dose in this population with a median GFR of 33 mL/min per 1.73 m 2 was higher than that needed for a contemporary hemodialysis cohort. 220,236 We hypothesize that the apparent discrepancy between the result of the trial in the 4047 patients enrolled (HR for the primary cardiovascular composite outcome-death or a nonfatal cardiovascular event-1.05; 95% CI, 0.94-1.17; P=0.41) and the much smaller (and underpowered by around 12-fold) Western Europe/Australia group (n=370; HR, 0.66; 95% CI, 0.43-1.01) may represent potential differences in the standard of care in different countries.
In summary, few pharmacological interventions decrease cardiovascular events or mortality in CKD patients and, in any case, the residual risk is still large. A built-in holistic approach to patient care in future RCTs may improve the outcomes. Additionally, an optimized risk stratification strategy may maximize the chances of success, as suggested by the CANTOS trial success. In this regard, a very recent study explored whether 10 potential circulating biomarkers related to CVD predicted atheromatosis progression in the NEFRONA cohort. The analysis identified higher FGF2 levels as an independent predictor of lower odds of progression. Thus, for every 50 pg/mL increase in FGF2 concentrations, the odds of progression decreased by 14%. 237 While the biological basis of the association should be explored in detail, FGF2 increases endothelial cell proliferation, decreases endothelial adhesion molecule expression and protects from endothelial dysfunction and from experimental atherosclerosis. [238][239][240][241] Another new biomarker associated with atheromatosis progression is the TNF (tumor necrosis factor) superfamily cytokine TRAIL (TNF-related apoptosis-inducing ligand). Indeed, lower TRAIL levels were associated with progression of atheromatosis after adjustment for other potential confounding factors. 242 A third biomarker found in the NEFRONA cohort is the rs495392 Klotho polymorphism. 243 Thus, patients presenting the TT or TG variants were less likely to progress (adjusted OR: 0.52; 95% CI, 0.33-0.82 and 0.70, 95% CI, 0.53-0.92, respectively) than GG homozygous patients. The impact of rs495392 on Klotho expression or function is currently unknown.

CARDIOVASCULAR EVENTS AND MORTALITY IN CKD
In CKD, the risk of CVD and of death is increased for multiple forms of CVD, including causes with divergent pathogenic mechanism from atherosclerosis to heart failure and sudden death 1,2 (Table 4). Sudden cardiac death accounts for up to 25% of hemodialysis patient deaths. 262 Standardized mortality rates from pulmonary embolism were 12-fold higher in dialysis patients than in the general population, as compared with 11-and 8-fold higher for atherosclerosis-related forms of death, such as myocardial infarction and stroke. 263 These data emphasize the need for a precise and optimized definition of cardiovascular end points in future RCTs that are in accordance with the putative mechanism of action of the intervention.
Both eGFR and proteinuria are independent predictors for negative cardiovascular outcomes, providing an even higher risk than other traditional cardiovascular risk factors, such as DM, hypertension, dyslipidemia, or smoking. 244,245,[247][248][249][250]264,265 This negative association has been described in all ethnic and age groups. An exception may be patients over the age of 55 years, in whom a U curve has been observed in large studies (over 2 million participants) in the association between eGFR and mortality. 6 While these studies confirmed the association of eGFR <60 mL/min per 1.73 m 2 with higher mortality, they also observed an association between high eGFR and mortality. This is thought to represent the shortcomings of current estimation of GFR equations, since for these large populations, they rely on serum creatinine, and malnourished or sarcopenic individuals may have low serum creatinine values leading to overestimation of eGFR. However, there is a marked interindividual variability in predictive factors for the development of cardiovascular events in patients with CKD. 250,257 Diverse community-based studies showed a robust association between lower renal function and atherosclerosis-related complications, including myocardial infarction, stroke, [252][253][254][255] or peripheral artery disease. 256,257,266 These negative outcomes have been observed in any category of CKD and interestingly also in patients with mild declines of renal function, in which CKD is diagnosed based on albuminuria rather than on eGFR. In this regard, the use of noninvasive techniques for the early diagnosis of silent atherosclerosis has been proposed. Differences in the cIMT and in prevalence of carotid plaques were already present in subjects with normal renal function when separated by eGFR tertiles using 85 and 99 mL/min per 1.73 m 2 as cutoff points. 260 In Spain, 24% of those aged ≥65 years in the general population had an eGFR <90 mL/min per 1.73 m 2 or overt proteinuria. 267 By contrast, a majority of patients with a prior atherosclerotic event had low eGFR. 268 Thus, in a cohort of 704 ischemic heart disease patients hospitalized with acute coronary syndrome with a mean age of 61 years, 76 % had an eGFR below 90 mL/min per 1.73 m 2 when evaluated 7 months after the acute event. 268 Of those with decreased eGFR, 38% had above normal levels of FGF23, suggesting a negative impact of decreased eGFR over CKD-related parameters.

Subclinical Atherosclerosis and Cardiovascular Events and Mortality
The Framingham cardiovascular risk equations are inaccurate in cardiovascular event prediction in patients with CKD. 264 A recent meta-analysis showed that addition of eGFR and albuminuria to the Framingham equations improved the concordance (C) index for cardiovascular mortality in 1.67% (95% CI, 1.31-2.02), for coronary heart disease in 0.73% (95% CI, 0.50-0.97), for stroke in 1.28% (95% CI, 0.76-1.8), and for heart failure in 2.58% (95% CI, 1.69-3.48). 269 The presence of carotid atheroma plaque was associated to the presence of coronary artery disease. 270 When plaque presence is included, the prediction capability of cardiovascular events increases significantly. [271][272][273] Similarly, plaque extent is a strong predictor. 274,275 In both predialysis (Table 5) and dialysis ( Table 6) patients, addition of plaque burden as evidence of subclinical atheromatosis to the equation, increased the C index even further. The net reclassification index was positive (11.0% for predialysis and 16.9% for dialysis patients), indicating that the risk of 11% of predialysis patients and almost 17% of dialysis patients would be more accurately calculated if plaque information was added to the algorithm. 276 Similarly, plaque presence was independently associated to cardiovascular events in dialysis patients. [277][278][279] On the contrary, cIMT is a weak predictor of cardiovascular events in the general population 280 276 Several specific CKD-related biomarkers have been associated to cardiovascular events. In CKD patients not on dialysis, low vitamin D levels are cardiovascular event predictors. 88,177,276,281 The consistent association of low 25(OH)D levels with cardiovascular outcomes, either atheromatosis progression or events is striking and any potential cause-and-effect relationship debatable. In this regard, a recent general population trial concluded that supplementation with vitamin D did not result in a lower incidence of cardiovascular events. 282 The trial did include patients with CKD, except for those with renal failure or dialysis. However, it was marred by mean baseline 25 (OH)D levels within the normal range. Thus, the results cannot be extrapolated to patients with vitamin D deficiency. Second, high hyperkalemia levels are   285 Finally, results about cholesterol levels, cardiovascular events, and all-cause of mortality are paradoxical in endstage renal disease patients [286][287][288] possibly due to the malnutrition-inflammation syndrome commonly present that usually includes lower serum cholesterol levels. 289 In the recent years, novel CKD-related biomarkers of subclinical atherosclerosis have been described. Lower serum levels of the TNF superfamily cytokine TWEAK (TNF-related weak inducer of apoptosis) and higher serum ACE2 levels were independently associated with the presence of atherosclerosis plaques. [290][291][292][293] Additionally, higher serum TWEAK levels were associated with a slower plaque progression (b=−0.135; P=0.001), independently of renal function. 294 It should be reminded that serum TWEAK levels are lower in CKD than in the general population, 292 but higher levels had previously been associated with mortality in hemodialysis patients. 295 This apparent paradox may result from the increased expression of the TWEAK receptor Fn14 in stressed tissues leading to increased sensitivity to TWEAK. 296 Lower circulating TWEAK levels were also an independent predictor of cardiovascular death or events, increasing the area under the ROC curve built with only traditional risk factors. 297 Furthermore, the combination of TWEAK with 2 biomarkers linked to vascular calcification, osteopontin and osteoprotegerin, significantly increased the accuracy of the model to predict cardiovascular events, including cardiovascular mortality. Thus, patients with osteoprotegerin levels >869 pg/mL, osteopontin levels >35.5 ng/ mL and TWEAK <369 pg/mL which represented 6.4% of the sample, displayed the highest cardiovascular event rate, independently of other variables like age, sex, DM, or CKD category. 298 Finally, soluble ADAM (a disintegrin and metalloproteinases) activity, derived from circulating ADAM17, was associated with an increased risk of cardiovascular events. 299

FUTURE DIRECTIONS
The only way to advance in clinical medicine is to demonstrate the efficacy of the intervention in an RCT. Thus, improving RCT design is a key priority (Table 7). Key aspects to take into account are that length of the study and primary outcomes should be in line with the presumed mode of action of the drug. As an example, sudden cardiac death or pulmonary thromboembolism should not be considered appropriate outcomes for interventions purportedly targeting atherosclerosis and are much more common in CKD patients than in the general population.
Another key aspect is that all patients in the trial should be treated according to current standards of CVD prevention. Thus, either the RCT is performed in countries that do provide all those current standards by default and not subject to the specific insurance status of the participant or the RCT provides coverage for all these standards. Failures like the TREAT trial should not be repeated. These current standards keep evolving. Thus, any trial designed or initiated in 2019 should require type 2 DM CKD patients to be on statins, RAS blockade, and SGLT2 inhibitors and using noncalcium-based phosphate binders (if a phosphate binder is required) unless intolerant or even the intolerant exception may be dropped. In hemodialysis patients, dose and modality should meet state-of-the-art standards before considering enrolling the patient, and iron stores should be replete before starting erythropoiesis-stimulating agents. 300 Otherwise, we risk missing the efficacy of an intervention because patients are suboptimally treated regarding CVD prevention. As a next step, interventions potentially targeting complementary pathways may be tested in the same trial. This may allow pharma companies to share the costs and burden of the RCT, and be more ambitious, especially regarding length of follow-up. Finally, the target population should be narrowed as much as possible to the mechanisms of action of the intervention. An intervention targeting atherosclerosis should only enroll patients having atherosclerosis at baseline, as example, as demonstrated by imaging. Otherwise, the trial may be burdened by a subset of patients that do not have atherosclerosis and may not develop it in natural history within the horizon of the trial. For an intervention targeting vascular calcification, this should be present at baseline for the same reason. The therapeutic strategies that Changes in concordance index when adding eGFR, proteinuria and number of plaques in a competing events regression model to evaluate risk of cardiovascular events in CKD G3 and G4 patients of the NEFRONA study (Observatorio Nacional de Atherosclerosis en NEFrologia). CKD indicates chronic kidney disease. Changes in concordance index when adding dialysis vintage and number of plaques in a competing events regression model to evaluate risk of cardiovascular events in the CKD G5D patients of the NEFRONA study (Observatorio Nacional de Atherosclerosis en NEFrologia). CKD indicates chronic kidney disease. merit specific CKD trials based on the success in general populations and the available results in CKD patients are targeting inflammation with canakinumab and targeting LDL-cholesterol with PCSK9 inhibitors. 225,227 These should be performed within a 5-year horizon. A stratified personalized medicine strategy may be used for patient enrollment that may test both interventions, based on evidence of baseline inflammation of residual LDL-cholesterol and Lp(a) levels while on statins.

CONCLUSIONS
A recent analysis of the Global Burden of Disease 2016 in Spain concluded that at the current rate of increase in the number of deaths attributable to CKD, CKD will become the second most frequent cause of the death, after Alzheimer, in a not so distant future. 301,302 Thus, action is needed to reverse this predicted course of events. CKD patients suffer from a high incidence of cardiovascular events, and patients in secondary prevention are treated as high cardiovascular risk patients. However, despite the recent KDIGO lipid-lowering therapy guidance, 226,227 there is still controversy among practicing physicians on whether all CKD patients should be classified as high atherosclerosis risk patients and treated pharmacologically as such. The NEFRONA study has shed some light on the significance and complications of subclinical atherosclerosis in CKD. The use of a simple, inexpensive, and noninvasive method like vascular ultrasound could help to stratify risk in CKD patients without a previous cardiovascular event, potentially preventing unnecessary pharmacological treatment in some patients, thus decreasing cost and pill burden in a population already highly medicated, while indicating therapy for some age groups without current statin indication for primary prevention (eg, <50-year-old; Figure 6). Multitarget intervention Table 7

. Proposals to Improve the Design or RCTs With Cardiovascular End Points in CKD Patients
Adequate length of follow-up and end point to the expected mechanism of action of the intervention Holistic approach to patient care: perform RCT in countries that offer this by default or include these requirements and medications in the trial design: Optimal anemia control: replete iron stores before prescribing ESAs Optimal acid-base control (avoid acidosis and alkalosis) Optimal CKD-MBD therapy: PTH within recommended range, 25(OH)D, calcium and phosphate levels within normal range, avoid calcium-based phosphate binders Optimal dialysis dosing, frequency (consider 4× a week for patients prone to volume overload), duration, ultrafiltration (adjust with techniques such as lung ultrasound), and modality Optimal treatment of comorbidities (eg, for type 2 diabetes mellitus this entails use of RAS blockade, statins, and SGLT2i for patients not on dialysis) Optimal management of hyperkalemia (avoid hypokalemia-or hyperkalemia-associated deaths) CKD indicates chronic kidney disease; ESAs, erythropoietin stimulating agents; MBD, mineral and bone disorder; PTH, parathyroid hormone; RAS, renin-angiotensin system; and RCT, randomized clinical trial. This atherosclerosis-focused approach should be complemented with measures to prevent nonatherosclerosis cardiovascular disease (CVD). Magnitude of risk in both graphs is not necessarily to scale. The area corresponding to G1-G2 in B is punctate, since this population was not studied in NEFRONA, but there is evidence from other studies supporting an increased risk for atherosclerosis progression in early CKD. 171 addressing all risk factors for which experimental evidence of potential cause-and-effect relationship exists should be considered. For some risk factors (eg, serum phosphate) sex-specific therapeutic targets may be considered. Ideally, this paradigm should be tested in clinical trials enriched in really high-risk patients (ie, those that already have developed atherosclerosis). These interventional studies should address whether the use of vascular ultrasound increases the reclassification of CKD patients, and whether this strategy may be cost-effective in interventional clinical trials, thus reducing the unacceptable cardiovascular morbimortality in this population. An atherosclerosis-focused approach should be complemented with measures to prevent nonatherosclerosis CVD.