Sie sind auf Seite 1von 20

Uremic toxins

INTRODUCTION The uremic syndrome can be defined as the deterioration of multiple biochemical and physiological functions in parallel with progressive renal failure, thereby resulting in complex but variable symptomatology [1-7]. Normally, healthy kidneys excrete a myriad of compounds. Uremic retention solutes accumulate in the patient with chronic kidney disease (CKD), including the patient with Kidney Disease Outcomes Quality Initiative (K/DOQI) stage 5 disease or end-stage renal disease (ESRD) [8]. The retention of these solutes is directly or indirectly attributable to deficient renal clearance.

These retained solutes are called uremic toxins when they contribute to the uremic syndrome. Only a few solutes have an established role. Apart from inorganic compounds, only a few compounds conform to the strictest definition of uremic toxins. However, this does not preclude a potential role for various other retention solutes. (See "Overview of the management of chronic kidney disease in adults".)

Uremic toxins can be subdivided into three major groups based upon their chemical and physical characteristics:

●Small, water-soluble, non-protein-bound compounds, such as urea

●Small, lipid-soluble and/or protein-bound compounds, such as the phenols

●Larger, so-called middle molecules, such as beta2-microglobulin (beta2-m)

Views on the uremic syndrome and several uremic solutes have changed substantially during the last two decades [5,9]. The biochemical, physiological, and/or clinical impact of those compounds is presented here.

SMALL WATER-SOLUBLE COMPOUNDS

Urea Despite the extensive number of studies addressing the role of urea in the uremic state, only a few have reported well-defined, adverse biochemical or physiologic effects:

●Urea inhibits Na-K-2Cl cotransport in human erythrocytes, as well as a number of cell volume-sensitive cellular transport pathways [10]. The Na-K-2Cl cotransport is a ubiquitous process that serves numerous vital functions, including cell volume and extrarenal potassium regulation.

●Urea inhibits macrophage-inducible nitric oxide (NO) synthesis at the posttranscriptional level

[11].

●As the most important osmotically active uremic solute, urea may provoke dialysis disequilibrium if the decrease in plasma concentration of urea during dialysis occurs too rapidly. (See "Dialysis disequilibrium syndrome".)

●Urea is a precursor of some of the guanidines, especially guanidino succinic acid, which induce biochemical alterations by themselves. (See 'Guanidines' below.)

●With increasing renal dysfunction, increasing amounts of cyanate are spontaneously transformed from urea. The active form of cyanate, isocyanic acid, carbamylates proteins, thereby affecting their structure and function [12].

●Treatment of adipocytes with disease-relevant urea concentrations induced radical oxygen species production, caused insulin resistance, and increased modified insulin signaling molecules in one study [13].

●Urea induced in vitro intestinal barrier dysfunction in one study, which could play a role in the increased leakiness of the gut for endotoxin, which has a proinflammatory potential [14].

It has been hypothesized, however, that the possible toxic effects of urea are counterbalanced by the simultaneous retention of methylamines [15].

The apparent low degree of toxicity of urea is corroborated by:

●The lack of correlation with symptoms of the uremic syndrome after addition of urea to the dialysate [16]

●The missing impact in the Hemodialysis (HEMO) and Adequacy of Peritoneal Dialysis in Mexico (ADEMEX) studies of an increase in urea removal on outcome [17,18]

Marker of dialysis adequacy Urea is unequivocally recognized as a marker of solute retention and removal in dialyzed patients. The degree of urea clearance also clearly correlates with clinical outcome of patients undergoing maintenance hemodialysis [19]. Thus, urea kinetic modeling is one of the principal tools to estimate and (if necessary) correct the dialysis dose. (See "Prescribing and assessing adequate hemodialysis" and "Patient survival and maintenance dialysis".)

However, high blood concentrations of urea may not necessarily correlate with poor outcome:

●High serum concentrations of urea due to adequate protein intake that are compensated by adequate removal may be relatively harmless, as compared with high urea levels due to inadequate dialysis [20].

●Low urea levels related to low protein intake may negatively correlate with prognosis [21].

It is also unclear whether the kinetic behavior of urea is representative of the behavior of other uremic retention solutes [22]. Data suggest that the dialytic removal of lipophilic protein-bound compounds, as well as that of several other water-soluble compounds, is different from that of urea

The validity of urea kinetic modeling has principally been tested with the small pores found in older dialysis membranes. Since the removal of urea is adequate with such membranes, but the removal of larger molecules is absent or low, the clearance of middle molecules may not be clearly represented by urea kinetic modeling. Therefore, some investigators have asked whether there should be a search for marker molecules that are representative of large and/or lipid-soluble compounds. Findings consistent with this view include overall survival advantages for those enrolled after greater than 3.7 years of hemodialysis treatment, cardiovascular survival advantages for the entire group of hemodialysis patients treated with large-pore dialyzers, and a survival advantage for patients with a lower beta2-microglobulin (beta2-m) concentration, irrespective of the membrane type [17,27-29]. However, these results were from secondary analysis of the HEMO study and need to be confirmed in prospective, sufficiently powered studies.

In the randomized, controlled Membrane Permeability Outcome (MPO) study, the group on high-flux membranes with larger pores had a superior removal of larger molecules as illustrated by lower predialysis beta2-m concentrations over time, compared with the group on low-flux dialyzers with smaller pores [30]. Although removal of small, water-soluble molecules as estimated

by Kt/Vurea was the same for the two membranes, survival was significantly better in patients with serum albumin <4 g/dL, which is the patient group for which the study was originally designed, and, at secondary analysis, in diabetic patients [30]. In a smaller randomized, controlled trial that compared on-line hemofiltration with low-flux hemodialysis, hemofiltr ation resulted in a decrease in beta2-m concentrations and a survival advantage, in spite of a lowering of Kt/Vurea [31].

Nevertheless, as urea removal has been related to survival, urea kinetic modeling should continue to be used as a baseline parameter for the study of dialysis adequacy. One should take into account thatKt/Vurea may not reflect adequacy of dialysis strategies introduced after the proposal of Kt/Vurea as a parameter of adequacy [32,33].

Guanidines The guanidines are a large group of structural metabolites of arginine. Several of the guanidino compounds modify key biological functions:

●Guanidino succinic acid and guanidinopropionic acid inhibit neutrophil superoxide production

[34].

●Guanidino compounds exert both pro- and anti-inflammatory effects upon leukocytes [35,36].

●Guanidino succinic acid, gamma-guanidinobutyric acid, methylguanidine, homoarginine, and creatine induce seizures after systemic and/or cerebroventricular administration in animals

[37].

●A mixture of guanidino compounds suppresses the natural killer cell response [38].

●Guanidines cause structural damage to proteins by deamidation/isomerization, which in turn reduces the binding of homocysteine [39].

Nitric oxide and guanidines Although not observed in all studies [40], most investigations have found that NO synthesis is inhibited in patients with end-stage renal disease (ESRD). This may result in vasoconstriction, hypertension, immune dysfunction, and neurologic abnormalities.

Although arginine (which is also a guanidino compound) markedly enhances the production of NO, some of the other guanidines, as arginine analogs, are strong, competitive inhibitors of NO synthase. Asymmetric dimethylarginine (ADMA), which is significantly increased in ESRD [41], is the most specific endogenous compound with inhibitory effects on NO synthesis. In the brain, ADMA causes vasoconstriction and inhibition of acetylcholine-induced vasorelaxation [42]. It has also been implicated in the development of hypertension, adverse cardiovascular events [43-46], and mortality (see "Risk factors and epidemiology of coronary heart disease in end-stage renal disease (dialysis)"). Administration of ADMA to healthy volunteers resulted in an increase of vascular resistance and blood pressure in a dose-related manner [46]. It was also demonstrated that ADMA increased vascular stiffness and decreased cerebral perfusion in healthy subjects [47]. It has been suggested, based on solute dialysance data, that ADMA was protein bound [48].

The increase in symmetric dimethylarginine (SDMA), the steric variant of ADMA, is more pronounced than that of ADMA, but, until recently, this compound has been considered biologically inactive. However, SDMA was linked to an increase in reactive oxygen production and an inhibition of NO synthesis [49]. Subsequently, SDMA was associated with an increase in free radical production by leukocytes; this effect was attributed to an increased calcium influx via store-operated Ca 2+ channels (SOCs) [50]. In an extensive study that evaluated a panel of guanidino compounds on several cell systems involved in vascular damage, SDMA most consistently induced proinflammatory effects in various types of leukocytes [36]. In addition, SDMA also induced

monocytic cytokine production, in contrast to ADMA [51]. In the clinical arm of this study, SDMA in chronic kidney disease (CKD) patients was more significantly correlated to interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF-alpha) levels, compared with ADMA [51]. SDMA also modified high-density lipoprotein (HDL) into an abnormal molecule, inducing endothelial damage [52].

Removal of guanidines The generation of the guanidines synthesized from arginine in the proximal convoluted tubule, such as guanidinoacetic acid and creatine, is depressed in ESRD [53]. The reported concentration of guanidinoacetic acid is not elevated among uremic individuals [7]. On the other hand, the synthesis of guanidino succinic acid, guanidine, and methylguanidine is markedly increased, which appears to be due to urea recycling. The reported concentrations of guanidino succinic acid and methylguanidine are increased in uremia, compared with normal concentrations (table 1) [7].

Dialytic removal of the guanidines, despite possessing a low molecular weight, is not consistently comparable with that of urea. Many of these compounds display different intradialytic behavior, suggesting a pluricompartmental distribution [26,54]. In the absence of protein binding, this behavior points to a retardation in the transfer of these molecules from inside to outside of cells. Based on kinetic modeling calculations, these findings have been confirmed from direct estimations [25]. Further mathematical analysis revealed that increasing dialysis time was the preferred strategy to improve removal of guanidinecompounds that had larger distribution volumes (such as methylguanidine, for example), whereas more frequent dialysis preferably impacted upon compounds with a smaller distribution volume (such as guanidino succinic acid) [55]. Optimal results were obtained with a combination of daily and long, slow dialysis [55].

Removal of ADMA is less dependent on renal excretion than on metabolic transformation, which is inhibited in renal failure. Enhancing metabolism might decrease ADMA concentration. Proof of principle is suggested by studies that showed that genetic overexpression of the ADMA-degrading enzyme, dimethylarginine dimethylaminohydrolase, was cardioprotective in mice following heart transplantation [56]. Furthermore, since the transmethylation reaction catalyzed by this enzyme is stimulated by lowering homocysteine levels, ADMA concentration was decreased by coadministration of intravenous methylcobalamin and oral folate [57]. These data illustrate how uremic solutes may influence each others’ concentration and how removal depends not only on renal function and dialysis, but also on metabolism, which can be influenced externally by administration of drugs, vitamins, or nutritional additives.

Oxalate Massive oxalate retention is uncommon in dialyzed patients, except in primary hyperoxaluria [58]. In this disorder, production is increased due to genetically mediated alterations of oxalate metabolism. (See "Primary hyperoxaluria".)

Among ESRD patients without primary hyperoxaluria, serum oxalate concentrations are increased approximately 40-fold, compared with healthy controls [59]. Secondary oxalosis in such patients is characterized by the deposition of calcium oxalate in multiple tissues, which was mostly observed in early renal replacement therapy with inefficient dialysis strategies [60].

Such findings may be seen among those with excessive intake of oxalate precursors, including ascorbic acid, green leafy vegetables, rhubarb, tea, chocolate, or beets. In addition, those with inflammatory bowel disease are at risk for this complication.

The role of pyridoxine (vitamin B6) in uremic oxalate accumulation remains a matter of debate. In rats with renal failure, pyridoxine depletion results in increased urine oxalate excretion and depressed renal function [61]. In hemodialysis patients, pyridoxine at 800 mg/day causes a decrease in oxalate concentration [62]. However, such high doses of pyridoxine may induce gastrointestinal intolerance.

Dialytic removal of oxalate is similar to that of urea and, therefore, is relatively easy with any of the classic dialysis strategies. However, since removal is lower than in healthy controls, serum concentrations remain higher than normal [63].

Uric acid A number of findings point to the involvement of uric acid in the development of uremic complications, especially cardiovascular morbidity and mortality [64-68]. However, a major confounder in epidemiological studies is that factors associated with a high uric acid concentration, such as obesity, diabetes, and hypertension, are cardiovascular risk factors, whereas therapeutic options to decrease uric acid, such as allopurinol, may be the cause of morbid complications.

Uric acid has been linked to mortality in a number of association studies. In the National Health and Nutrition Survey (NHANES) study, high uric acid was associated with overall and cardiovascular mortality, but the association disappeared after correction for kidney function [64]. In a Mendelian randomization study, both uric acid itself as well as a score composed of a number of genetic polymorphisms associated to high uric acid were associated with multiple cardiovascular complications [65].

In observational trials, treatment with allopurinol was associated with arterial stiffness, even after correction for confounders [66]. In a Japanese section of the Dialysis Outcomes Patterns Study (DOPPS), allopurinol was associated with overall mortality, albeit only in the subpopulation without previous cardiovascular events [67].

In a long-term, randomized, controlled trial, the group treated with allopurinol had a lower number of cardiovascular events and a slower progression of kidney failure, but the small sample size of the study was a limitation factor, so this study needs confirmation [68].

Of note, in older studies, uric acid and the group of purines to which it belongs have been associated with calcitriol metabolism and with inhibition of expression of CD14, which acts as a lipopolysaccharide receptor, on monocytes [69-71].

Phosphorus A high serum concentration of phosphates is clearly related to pruritus and hyperparathyroidism, both manifestations of the uremic syndrome [72]. Phosphorus excess also inhibits 1 alpha-hydroxylase and hence the production of calcitriol, the active vitamin D metabolite [73]. Phosphorus retention also alters polyamine metabolism by causing a decrease in intestinal dysfunction and proliferation of intestinal villi [74].

More recently, the phosphorus-vitamin D-parathyroid axis has been linked to fibroblast growth factor-23 (FGF-23), a large molecule that increases in concentration early in the course of CKD [75] and has a strong predictive value for negative outcomes [76]. (See "Overview of chronic kidney disease-mineral bone disease (CKD-MBD)", section on 'Fibroblast growth factor 23'.)

At least in animals, phosphate restriction has an attenuating effect on the progression of renal failure. The results are less compelling in humans [77]. However, dietary phosphate restriction reduces parathyroid hormone (PTH) levels over a wide range of serum calcium concentrations [78].

In observational studies, high serum phosphorus is linked to mortality [79]. The same applies also for low serum phosphorus, very likely because low phosphorus is an indicator of inadequate nutritional status.

Blood phosphorus concentration is the result of protein catabolism and protein intake as well as of the ingestion of other dietary sources. Restriction of oral protein intake increases the risk of malnutrition [72], which can be avoided by the administration of oral phosphate binders [80]. By providing a balanced dietary advice, it is, however, possible to substantially reduce serum phosphorus [81]. This includes the avoidance of processed foods, which many contain excessive quantities of phosphorus [82].

The effect of oral phosphate binders is often insufficient, especially in subjects with high intake. One major effect related to hyperphosphatemia is the increase in serum Ca x P product resulting in Ca deposition in the tissues, especially the vessel wall [79], which is a negative prognostic factor in patients with CKD [83]. Of note, however, Ca x P product was abandoned by the Kidney Disease:

Improving Global Outcomes (KDIGO) guidelines on chronic kidney disease/mineral bone disease (CKD/MBD) as it was considered a mathematic artifact [84]. This does not exclude the individual roles of both Ca and P in vascular disease of CKD [84]. Application of calcium-containing intestinal phosphate binders may decrease P, but this effect may be counterbalanced by a rise in Ca [85].

The newer generation of phosphate binders, such as sevelamer and lanthanum, may provide a solution to this problem [86]. (See "Treatment of hyperphosphatemia in chronic kidney disease".)

Dialytic removal of phosphate is unpredictable [24] and often followed by impressive rebounds, which is due to the release of intracellular phosphate stores [87]. Slower dialysis techniques may allow a better control of phosphate levels (see "Technical aspects of nocturnal hemodialysis"). In a study strictly controlling all other dialysis parameters, for example, an increase of high-flux dialysis duration from four to eight hours was accompanied by a time-dependent rise in phosphate removal

[88].

Metabolic acidosis Metabolic acidosis can contribute to some of the abnormalities in ESRD. Metabolic acidosis can cause muscle wasting and bone mineral loss and, in children, impair growth [89,90]. (See"Pathogenesis, consequences, and treatment of metabolic acidosis in chronic kidney disease".)

Creatinine Although serum creatinine is a marker for renal function, only limited data suggest that this substance is associated with adverse effects [4]. These effects, such as chloride channel interference, are only reported at creatinine levels not observed with CKD.

Polyamines Polyamines, which are measured at increased levels in uremia, may contribute to anorexia, vomiting, and adverse central nervous system effects [4].

Trimethylamine-N-oxide (TMAO) TMAO, a gut-derived cardiovascular toxin and metabolite of choline, lecithin, and L-carnitine, is markedly increased in kidney failure and has been associated to mortality and progression of kidney disease [91]. The association to mortality in this study was independent of glomerular filtration rate (GFR) and C-reactive protein (CRP).

Carbamylation products Carbamylation products have been associated to increased mortality in ESRD patients in a large, observational series [92]. In spite of the acknowledged link between

urea retention and carbamylation, the correlation between blood urea nitrogen and carbamylation products was relatively weak, pointing to a possible role of other mechanisms in their generation.

PROTEIN-BOUND COMPOUNDS

P-cresol and p-cresyl sulfate P-cresol, a phenol with a molecular weight of only 108 D, has been considered the prototype of lipophilic, uremic, protein-bound toxins. Because of their strong protein binding, their removal by classical dialysis is hampered, with removal of small, water-soluble molecules being completely different [23,93]. It is therefore conceivable that alternative removal methods (eg, adsorption or convective transport) should be developed before adequate elimination of these toxins can be obtained [94,95].

However, evidence underscores that conjugates of p-cresol, p-cresyl sulfate and p-cresyl glucuronate, and not p-cresol are present in the blood of patients with chronic kidney disease (CKD) [93,96]. Previous repeated measurements of p-cresol were related to the fact that most determination methods applied strong acidification for deproteinization, a measure resulting in deconjugation of p-cresyl sulfate and p-cresyl glucuronate by hydrolysis [97,98]. Since this hydrolysis is conceivably proportional to the quantity of conjugates available, former p-cresol measurements as well as estimations of protein binding are likely proportional to what would have been found by direct measurement of the conjugates.

High-flux dialysis, compared with low-flux dialysis, has no beneficial effect on the removal of these toxins [23,94]. However, compared with peritoneal dialysis, p-cresol is cleared better with high-flux hemodialysis [99]. Nevertheless, the plasma concentrations of protein-bound toxins are lower in patients on peritoneal dialysis, compared with those on hemodialysis [99,100]; these findings cannot be explained by differences in dietary protein intake or residual renal function, suggesting a role for metabolism and/or intestinal handling [101].

P-cresol is an end-product of protein catabolism, produced by intestinal bacteria that metabolize tyrosine and phenylalanine [102]. Environmental sources of p-cresol are toluene, menthofuran, and cigarette smoke. Specific concern is warranted regarding menthofuran, which is present in several herbal medicines, flavoring agents, and psychedelic drugs [103]. In view of the known role of the intestine in the generation of p-cresol and its conjugates, several approaches that interfere with intestinal generation of these toxins have been studied and shown to decrease their concentration [104]. (See 'General removal strategies' below.)

In several studies, p-cresol concentration was related to parameters of clinical outcome, including hospitalization rate (particularly due to infection) [98], clinical symptoms of the uremic syndrome [97], mortality [105], and cardiovascular disease [106,107]. The same associations were also demonstrated with p-cresyl sulfate [108,109] and existed among patients with earlier stages of CKD (CKD 2 to 4) [110]. The latter observation suggests that protein-bound solutes may exert a toxic effect on the general population, or at least those not affected by marked kidney disease. In a prospective, observational analysis, p-cresyl sulfate and indoxyl sulfate were independently associated with progression of CKD [111].

The number of data underscoring the biochemical (toxic) effect of the conjugates is steadily growing. Of note, in most of the more recent studies, correct concentrations were applied while also taking into account the effect of protein binding. In one study, it was demonstrated that p-cresyl sulfate stimulated baseline leukocyte activity, thereby pointing to a proinflammatory effect, whereas

the parent compound p-cresol essentially inhibited activated leukocyte function [112]. Another study found that retained p-cresol and p-cresyl sulfate altered endothelial function [113]. Several studies showed effects related to tubular damage and progression of renal failure, such as an enhanced expression of cytokine and inflammatory genes in proximal tubular cells [114], epithelial mesothelial transition, fibrosis, and glomerulosclerosis through the activation of the renin-angiotensin- aldosterone system [115] and suppression of Klotho gene expression leading to fibrosis [116]. Oxidative stress damage has been demonstrated in rat tubular cells [117], and the induction of insulin resistance and reallocation of adipose cells has been demonstrated in different tissues [118]. The leukocyte-activating impact of p-cresyl sulfate has been demonstrated in whole human blood [119]. In one study, the second conjugate, p-cresyl glucuronide, combined with p-cresyl sulfate, had an additive effect on leucocyte oxidative burst activity [119]. P-cresyl sulfate, administered peritoneally or intravenously, increased rolling of leukocytes along the endothelium of peritoneal vessels, as directly visualized by in vivo microscopy, and p-cresyl sulfate and p-cresyl glucuronide induced vascular leakage for albumin [120]. These data suggest a direct toxic effect of these molecules on vascular endothelium and an induction of the cross-talk between vascular endothelium and leukocytes, which is a primary step of a chain of events leading to vascular damage.

The concern has been expressed that, even if correct uremic concentrations were applied in studies on protein-bound toxins, the (near) absence of albumin may have resulted in an increased percentage of non-protein-bound fraction, which is the fraction that exerts biological activity [121]. However, in a systematic review that was limited to experimental studies that used correct free concentrations of p-cresyl sulfate, several studies confirmed the biological effects of p-cresyl sulfate at uremia-relevant concentrations [122]. The analysis, which also included studies on indoxyl sulfate, in total included 27 publications, of which several complied with high-quality standards.

Although the conjugates are present in the blood stream, it remains unclear whether, under some conditions, they may not be reconverted to p-cresol intracellularly, especially since certain cell types such as the leukocytes contain sulfatases and glucuronidases.

Previously, only a few studies that examined how to improve removal of the phenols by extracorporeal strategies were performed. One study showed superior dialytic clearance of p-cresol by high-flux hemodialysis, compared with peritoneal dialysis [99], although, as noted previously, plasma concentrations were lower in peritoneal dialysis patients [101]. Among dialysis strategies, convective approaches were more efficient, compared with diffusive ones [123-125]. In another study, increasing dialysate flow and dialyzer surface area while decreasing blood flow in an extended dialysis setting increased removal of protein-bound solutes without altering Kt/V [126]. However, as compared with standard, four-hour dialysis three times per week, a longer session (eight-hour dialysis) three times per week had no significant impact on protein-bound solute concentration; this was in contrast to observations with small, water-soluble compounds and middle molecules [127]. However, a nonsignificant trend for increased removal was nevertheless also observed in a large array of protein-bound molecules in this study, and it is not clear whether extended dialysis with maintenance of high blood and dialysate flows would increase this efficacy.

By applying combined fractionated plasma separation and adsorption, removal of p-cresol could be markedly enhanced, a strategy used as an artificial liver (Prometheus) [128]. Unfortunately, the approach applied in this study resulted in major coagulation disturbances [129]. However, the study

offers proof that adsorption may become an additional asset in the removal of this type of compound. In an in vitro setting, embedding adsorptive material on a high-flux membrane markedly enhanced the removal capacity for protein-bound toxins [95].

As far as convective strategies are concerned, it was shown that postdilution and predilution hemodiafiltration were not different regarding removal of p-cresyl sulfate. Both hemodiafiltration strategies were superior to predilution hemodiafiltration for p-cresyl sulfate and other protein-bound compounds, whereas postdilution hemodiafiltration was superior to predilution hemodiafiltration for all other types of uremic retention solutes (ie, the small, water-soluble compounds and the larger "middle molecules") [130].

The cresols, as several other protein-bound toxins, are the end-product of metabolic transformation of amino acids by intestinal microbiota and by conjugation processes in the intestinal wall or liver [104]. In uremia, the microbiota are modified in favor of species especially generating those uremic toxins [131], which is confirmed by the finding of an increased generation of p-cresol as kidney dysfunction progresses [132]. There are studies suggesting that intestinal bacterial production and intestinal uptake of p-cresol can be altered. As an example, prevention of the intestinal absorption

of p-cresol by administration of oral sorbents (AST-120) decreased its serum concentration [133]. In

a metabolomic study in rats, the same sorbent decreased the concentration of a host of mainly

protein-bound uremic solutes, including p-cresyl sulfate [134]. In addition, the prebiotic arabino-xylo- oligosaccharide (AXOS) decreased p-cresyl sulfate in mice, while decreasing its biochemical impact mainly related to insulin resistance [118]. However, in a mouse model of CKD, sevelamer hydrochloride administration had no impact on the concentration of any of the protein-bound compounds [135], and the same was observed in a subsequent study in humans [136]. Dietary intake may alter the generation of p-cresyl sulfate. In one study of 26 individuals with

normal renal function, the average p-cresyl sulfate excretion (reflecting generation) was 62 percent lower among vegetarians than among participants on an unrestricted diet [137]. A metabolomic comparison between hemodialysis patients with and without colon showed striking differences in protein-bound and other uremic toxin concentrations, including p-cresyl sulfate [138].

The quality of studies assessing interventions to modify the intestinal microbiota or their metabolism

in CKD is in general deceiving and their results contradictory [139]. A randomized, controlled trial on

dietary fiber supplementation showed a decrease in indoxyl sulfate concentration, but not in p-

cresyl sulfate [140].

Homocysteine Homocysteine (Hcy) is a sulphur-containing amino acid produced by the demethylation of methionine. Its retention with uremia results in the cellular accumulation of S- adenosyl homocysteine (AdoHcy), an extremely toxic compound that competes with S-adenosyl- methionine (AdoMet) and inhibits methyltransferases [141]. Hyperhomocysteinemia also disturbs epigenetic control of gene expression by inducing macromolecule hypomethylation [142]. Guanidino compounds modify serum albumin in a way that protein binding of homocysteine is decreased [39].

Patients with chronic renal failure have total serum Hcy levels two- to fourfold above those observed in normal individuals. In addition to the degree of renal failure, its serum concentration also depends upon nutritional intake (eg, of methionine), vitamin status (eg, of folate), and genetic factors.

Moderate hyperhomocysteinemia is an independent risk factor for cardiovascular disease in the general population and is a prevalent cardiovascular risk factor in patients with end-stage renal

disease (ESRD) [143,144]. (See "Overview of homocysteine" and "Risk factors and epidemiology of coronary heart disease in end-stage renal disease (dialysis)".) It may also be present at increased concentrations in kidney transplant recipients with cardiovascular disease [145]. (See "Risk factors for cardiovascular disease in the renal transplant recipient".)

Hcy increases the proliferation of vascular smooth muscle cells, one of the most prominent hallmarks of atherosclerosis [146]. The administration of excess quantities of the Hcy precursor, methionine, to rats induces atherosclerosis-like alterations in the aorta [147]. Hcy also disrupts several vessel wall-related anticoagulant functions, resulting in enhanced thrombogenicity [148]. A detailed discussion of the proatherogenic effects of hyperhomocysteinemia is presented elsewhere. (See "Overview of homocysteine".)

Hcy levels can be moderately reduced by the administration of folic acid, vitamin B6, and/or vitamin B12 [149]. To reduce Hcy, patients with ESRD may require higher quantities of vitamins than the nonuremic population.

Direct clinical proof of the benefit of decreasing Hcy concentration in uremia was not previously available. In addition, low rather than high levels of Hcy were associated with poor outcome in two large, observational studies [150,151]. However, one study found that correction for poor nutritional status and inflammation as confounders permitted the demonstration that Hcy was directly related to mortality in the population without chronic inflammation-malnutrition state (CIMS) [152]. This relationship was masked in patients with CIMS. Vascular events were not reduced in two large, randomized, controlled trials that effectively lowered Hcy by folic acid and B vitamins in patients with CKD [153,154].

In the Homocysteinemia in Kidney and End-Stage Renal disease (HOST) trial, a randomized, double-blind evaluation in approximately 2000 patients, active Hcy-lowering therapy also appeared to have no different impact on outcome, compared with placebo [153]. However, the placebo group was allowed to take folic acid supplements at their discretion, which casts some doubt on the purity of the placebo group. In addition, only one-third of patients had their Hcy levels normalized. A German trial that included more than 600 patients demonstrated no benefit with increased intake of folic acid, vitamin B6, and vitamin B12[155]. However, a meta-analysis by the same authors suggested a benefit in reducing cardiovascular disease associated with the vitamin preparations [156]. Many studies were small in this meta-analysis, which detracted slightly from the credibility of the final analysis.

Dialytic removal of Hcy is thought to be hampered in a similar way as the other protein-bound uremic toxins. Dialysis with the extremely open ("super flux") dialysis membranes, however, could decrease Hcy concentrations, possibly due to a modification of metabolism, rather than to direct removal [157].

Indoles Indoxyl sulfate is metabolized by the liver from indole, which is produced by the intestinal flora as a metabolite of tryptophan. Indoxyl sulfate, which is secreted in the normal kidney by organic anion transporter 3 [158], enhances drug toxicity by competition with acidic drugs at protein-binding sites [159] and inhibits the active tubular secretion of these compounds as well as the deiodination of thyroxine 4 (T4) by cultured hepatocytes [160,161].

Uremic retention solutes induce glomerular sclerosis [162], with their removal by peritoneal dialysis or by oral sorbent administration retarding the progression of intact nephron loss. Indoxyl sulfate

may be one of these possible uremic toxins. The oral administration of indole or of indoxyl sulfate to uremic rats causes a faster progression of glomerular sclerosis and renal failure [163,164]. In renal tubular cells, indoxyl sulfate induces free radical production, nuclear factor kappa B (NF-kappaB) activation, and upregulation of plasminogen activator inhibitor-1 (PAI-1) expression [165]. Studies cited above demonstrated indoxyl sulfate-associated enhanced expression of cytokine and inflammatory genes [114], epithelial mesothelial transition, fibrosis and glomerulosclerosis through the activation of the renin-angiotensin-aldosterone system [115], and suppression of Klotho gene expression leading to fibrosis [116].

Among patients on peritoneal dialysis, indoxyl sulfate was correlated with interleukin-6 (IL-6) [166]. Risk factors of atherosclerosis are associated with indoxyl sulfate concentration in hemodialysis patients [167]. In a study that included patients with CKD stages 2 to 5, serum indoxyl sulfate was predictive of mortality, even after adjustment for major cardiovascular risk factors [168].

Endothelial cell dysfunction is common in uremia. Indoxyl sulfate may play a role by inhibiting endothelial cell proliferation and repair [169], while it induces a significant production of free radical species in endothelial cells [170,171]. Endothelial dysfunction induced by indoxyl sulfate is also illustrated by an increase in microparticle release by cultured endothelial cells [172]. Endothelial- leukocyte interaction, resulting in leukocyte adhesion to the endothelial wall, has also been demonstrated [173]. In a rat model, indoxyl sulfate was demonstrated to stimulate vascular smooth muscle cell proliferation [174]. Indoxyl sulfate also has direct profibrotic, prohypertrophic, and proinflammatory effects on cardiac fibroblasts and myocytes [175].

It has been suggested that indoxyl sulfate plays a role in aortic calcification [176] and elements of bone dysfunction, such as resistance to parathyroid hormone (PTH), osteoblast dysfunction, and downregulation of pathways of PTH gene expression and low turnover bone disease [177-179].

A thrombogenic effect has also been attributed to indoxyl sulfate, essentially by the demonstration

of interference with the generation of thrombogenesis tissue factor 3 via the aryl hydrocarbon

receptor pathway [180-182].

In one study, indoxyl sulfate was shown to enhance adhesion and extravasation of leukocytes to

the same degree as lipopolysaccharide, as well as an almost total stagnation of blood flow within the studied vessels [120]. In an attempt to unravel the mechanism of these profound alterations, it was shown that, in the presence of indoxyl sulfate, the glycocalyx, a molecular layer protecting the endothelium, was affected. Heparan sulfate, one of the main components of the glycocalyx, was found in increased concentration in the serum of indoxyl sulfate-treated rats. Of note, a study had

shown in a similar way increased free-floating heparan sulfate in the plasma of hemodialysis patients [183].

A systematic review that included multiple studies revealed a negative impact of indoxyl sulfate

[122]. This was also observed with p-cresyl sulfate (see 'P-cresol and p-cresyl sulfate' above). That biological effects of indoles are not restricted to indoxyl sulfate was demonstrated in a study

showing that indole-3-acetic acid activated the inflammatory nongenomic aryl hydrocarbon receptor(AhR)/p38MAPK/NF-kappaB pathway, hence inducing the proinflammatory enzyme cyclooxygenase-2 [184]. A clinical arm of this study also showed a higher mortality among patients with the highest indole-3-acetic acid levels [184].

Indoles are found in various plants and herbs, and some are also produced by the intestinal microbiota. Several metabolites are retained in uremia, and kinetic behavior appears to differ among them. Some of these substances do not even conform to the definition of uremic retention solutes, as their concentration is low in patients with stage 5 (ESRD) (eg, tryptophan, melatonin).

Removal of protein-bound indoles is hampered during dialysis [23]. It does not correlate with that of urea or creatinine [24]. In conventional hemodialysis, super-flux triacetate membrane was superior to low-flux cellulose triacetate regarding removal of indoxyl sulfate [185]. Convective strategies are superior to diffusion with regards to removal [125,130]. Adsorptive removal strategies have the potential to provide added value [95]. In a group of CKD patients not yet on dialysis, the sorbent AST-120 (Kremezin R) actively decreased indoxyl sulfate in a dose-dependent way, as appreciated from a short-term, prospective, dose-finding study. There were no differences in decline of kidney function, however, possibly due to the short observation period [186].

On the other hand, in a prospective, observational analysis, p-cresyl sulfate and indoxyl sulfate were independently associated with progression of CKD [111]. In small, controlled studies with a long follow-up period, the group on AST-120 showed a slower decline of estimated glomerular filtration rate (eGFR), was started later on dialysis, and survived longer once dialysis was started [187-189]. However, a largeAmerican/European randomized, controlled trial could not confirm the beneficial impact of AST-120 on the progression of kidney failure [190]. The study contained no control of whether the group on AST-120 showed a decrease in plasma concentration of protein- bound uremic toxins.

Oral administration of bifidobacteria in gastro-resistant capsules modifying intestinal flora to hemodialysis patients reduces serum levels of indoxyl sulfate by correcting gastrointestinal flora [191,192]. Dietary intake may also alter serum levels of indoxyl sulfate. In the study cited above of 26 individuals with normal renal function, the average indoxyl sulfate excretion (reflecting generation) was 58 percent lower among vegetarians than among participants on an unrestricted diet [137]. In a randomized, controlled trial of hemodialysis patients, dietary fiber supplementation induced a decrease in indoxyl sulfate concentration, but not in p-cresyl sulfate [140].

Furanpropionic acid 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), a urofuranic fatty acid, is a strongly lipophilic uremic solute and a major inhibitor of drug protein binding [160]. This toxin inhibits the renal uptake of para-amino hippuric acid (PAH) in rat kidney cortical slices [193] and causes a decrease in renal excretion of various drugs, metabolites, and endogenously produced organic acids that are removed via the PAH pathway. In vivo renal CMPF clearance in the rat is inhibited by PAH and probenecid [194]. CMPF also inhibits hepatic glutathione-S-transferase [195], deiodination of T4 by cultured hepatocytes [161], and adenosine diphosphate (ADP)- stimulated oxidation of nicotinamide adenine dinucleotide (NADH)-linked substrates in isolated mitochondria [196].

There is a correlation between neurologic abnormalities and the plasma concentration of CMPF [197]. Since CMPF is virtually 100 percent protein bound, its removal by hemodialysis strategies is virtually nonexistent. CMPF levels can be lowered substantially only with peritoneal dialysis [198].

MIDDLE MOLECULES Middle molecules, arbitrarily defined as those of a molecular weight in excess of 500 D, were previously thought responsible for the uremic syndrome. However, at the

time of the formulation of the middle-molecule hypothesis, it was extremely difficult to isolate specific responsible molecules.

Nevertheless, several clinical, metabolic, and/or biochemical disturbances are caused by uremic compounds with molecular weights in this middle-molecule range:

●Chromatographic fractions of uremic ultrafiltrate with a molecular weight between 1 and 5 kD inhibit appetite and suppress food intake in animals [199].

●A 500 to 2000 D subfraction of uremic serum inhibits apolipoprotein (apo) A-I secretion in a human hepatoma cell line [200].

●A compound of molecular weight between 750 and 900 D inhibits osteoblast mitogenesis

More than 20 compounds have been identified that conform to the strict definition of middle molecules [8,202]. Several of those have biological effects, especially a proinflammatory impact [2].

Dialysis membranes with a capacity to remove middle molecules (high-flux membranes) have been related to lower morbidity and mortality [203-207]. However, all these data have been collected in observational studies.

In the Hemodialysis (HEMO) study, there was no significant improvement in overall survival with high-flux membranes (removing middle molecules), compared with low-flux membranes (with virtually no removal of middle molecules) [17]. Upon secondary analysis, however, a benefit for high-flux membranes was found for cardiovascular risk [17], with the difference especially marked for patients who had been dialyzed for more than 3.7 years prior to enrollment in the study [27]. In addition, based on data collected during the HEMO study, it was shown that, independent of the membrane used, beta2-microglobulin (beta2-m) concentrations were directly related to mortality [28,29]. In another subanalysis of the HEMO study, a decrease in cerebrovascular events was observed for dialysis with high-flux compared with low-flux membranes [208].

Subsequent to the HEMO study, two other studies showed superiority of high flux regarding outcome. Although both analyses were based on prospective studies [209,210], the evaluation of flux was performed by secondary analysis, and results are therefore less robust since the primary endpoints in those studies were not directly related to dialysis. In both studies, the superiority of high flux was demonstrated.

The next study in this context is the Membrane Permeability Outcome (MPO) study [30,211]. This study compares high flux versus low flux in European hemodialysis patients. A superior outcome for the high-flux membrane was shown in patients with a serum albumin below 4 g/dL, the subgroup of dialysis patients for whom the study protocol originally had been developed [30,211].

In an observational analysis of a Dialysis Outcome Practice Patterns Study (DOPPS) cohort, superiority of high-volume hemodiafiltration was suggested [212]. A limited controlled trial showed borderline survival superiority for on-line hemofiltration over low-flux hemodialysis [31]. An Italian multicentric study showed superiority of convective strategies with regards to intradialytic hemodynamic stability [213]. Two randomized, controlled trials showed survival superiority of hemodiafiltration over hemodialysis only in a secondary analysis for the subgroup with the highest ultrafiltration volume [214,215], creating a selection bias in favor of the patients with the better vascular status, in whom it is easier to accomplish high exchange volumes.

A third study, with ultrafiltration and substitution volume systematically kept high in all patients who

were randomized to hemodiafiltration and finished the study, showed a significant survival benefit for hemodiafiltration [216]. Unfortunately, the study did not contain a complete intention-to-treat analysis, since patients who dropped out of the hemodiafiltration arm because high exchange volumes could not be obtained were censored. In addition, substantially more patients were censored from the hemodiafiltration group because of transplantation.

A Cochrane meta-analysis showed no overall survival advantage nor an effect on hospitalizations

for hemodiafiltration and a modest advantage with regards to cardiovascular events and dialytic

hypotension [217]. The study, however, also pointed to an important overall risk of bias and limitations in trial methods.

Beta2-microglobulin Beta2-m (molecular weight of approximately 12,000 D) is a component of the major histocompatibility antigen (see "Major histocompatibility complex (MHC) structure and function"). Dialysis-related amyloid, which can be observed in patients being maintained on long- term dialysis, is to a large extent composed of beta2-m (see "Dialysis-related amyloidosis"). Nevertheless, some patients may suffer from amyloidosis after only one to two years of dialysis [218,219]. Since the last 15 years, a decline in incidence of dialysis-related amyloidosis has been recorded, at least in Europe [220], likely reflecting an improvement of dialysis quality, especially of dialysis water.

In a proteomic analysis searching for biomarkers for peripheral vascular disease in a population

reportedly without major kidney dysfunction, beta-2-m was the only discriminator, suggesting a

potential link to the development of vascular disease [221].

Beta-2-m has also been shown to impair cognitive function and act as a pro-aging factor [222]. However, no activation of circulating leukocytes could be observed in the presence of uremic concentrations of beta-2-m in an vitro study [223]. This does not exclude other biological actions of that molecule.

Advanced glycosylation end products (AGEs) may affect the pathophysiologic impact of beta2-m. AGE-modified beta2-m has been identified in amyloid of hemodialyzed patients [224]; it also enhances monocytic migration and cytokine secretion [225], suggesting that foci containing AGE beta2-m may initiate inflammatory response, leading to bone/joint destruction [224,225].

Serum beta2-m levels may be lower in peritoneal dialysis patients than in hemodialysis patients [226]. This may be due to a better conservation of endogenous residual renal function with peritoneal dialysis since peritoneal dialysis alone poorly clears beta2-m [99]. Although the clinical expression of dialysis-related amyloidosis disappears after kidney transplantation, the underlying pathological processes, such as bone cysts and tissue beta2-m deposits, remain [227].

In prospective studies, a progressive decline of predialysis beta2-m levels has frequently been demonstrated in patients dialyzed with membranes with a larger pore size [228]. The question arises whether this removal is sufficient to prevent the development of beta2-m amyloid. In a retrospective study, AN69 dialyzers with a large pore size were associated with a lower prevalence of amyloid disease than small-pore cuprophane [229]. However, cuprophane also induces a more profound inflammatory reaction, which is known to enhance the development of amyloid disease.

Because beta2-m is only removed by dialyzers with large pore size, it may be representative of other large molecules in its kinetic behavior. Apart from its role in amyloidosis, the biological impact of beta2-m seems minor.

Finally, a post-hoc analysis of the HEMO study found that serum beta2-m levels correlated with mortality, with each 10 mg/L increase in predialysis level being associated with an 11 percent increase for all-cause mortality [28]. Later analysis revealed that this mortality was related to infectious causes [29]. This result may lend support to using beta2-m levels as a marker for middle- molecule clearance.

Theoretically, long, daily hemodiafiltration may provide optimal clearance [230]. Prolonging only high-flux dialysis time, however, already causes a significant increase in removal of beta2-m [88].

Parathyroid hormone Parathyroid hormone (PTH), a middle molecule with a molecular weight

of approximately 9000 D, is generally recognized as a major uremic toxin. However, its increase in

concentration during end-stage renal disease (ESRD) is attributable to enhanced glandular secretion, rather than to decreased removal by the kidneys. Excess PTH gives rise to an increase in intracellular calcium, resulting in disturbances in the function of virtually every organ system [231].

A downregulation of PTH/PTHrP receptor mRNA expression is observed in the liver, kidney, and

heart of rats with advanced chronic renal failure, thereby blunting the cellular response to excess

PTH and creating resistance to PTH [232]. Parathyroidectomy does not entirely prevent PTH/PTHrP receptor downregulation [233], suggesting that this alteration depends on more than elevated PTH alone.

The increased PTH concentration in uremia is the result of a number of compensatory homeostatic mechanisms. Hyperparathyroidism results at least in part from phosphate retention, decreased production of calcitriol (1,25-dihydroxyvitamin D), and hypocalcemia. (See "Overview of chronic kidney disease-mineral bone disease (CKD-MBD)".)

Advanced glycosylation end products Glucose and other reducing sugars react nonenzymatically with free amino groups to be converted after weeks into AGEs through chemical rearrangements and dehydration reactions [234]. Several AGE compounds are peptide-linked degradation products (molecular weight 2000 to 6000 D) [235]. Among the postulated structures for AGE are imidazolone, pyrrole aldehyde, pentosidine, and N epsilon-(carboxymethyl)lysine.

The increased accumulation of AGE is not the result of enhanced glucose levels or reduced removal of modified proteins by glomerular filtration. With uremia, it is more likely due to increased concentrations of small carbonyl precursors. Thus, uremia can be described as a status of increased carbonyl stress, resulting from increased oxidation or decreased detoxification of these carbonyl compounds [236]. Together, these processes help result in increased levels of AGEs

AGEs affect multiple processes. These compounds:

●Cause an inflammatory reaction in monocytes by the induction of interleukin-6 (IL-6), tumor necrosis factor-alpha (TNF-alpha), and interferon-gamma [237]. Proinflammatory effects had previously been demonstrated with artificially prepared AGEs. A similar proinflammatory impact was present with genuine AGEs as they were accumulated in uremia [238].

●Modify beta2-m, which (as previously mentioned) may play an important role in the formation of dialysis-related amyloid.

●React with and chemically inactivate nitric oxide (NO), a potent endothelium-derived vasodilator, antiaggregant, and antiproliferative factor [239].

●Induce oxidative protein modification [240].

In addition to renal failure, AGEs are also retained in diabetes mellitus and aging, settings in which they have been implicated in tissue damage and functional disturbances. Specific receptors for AGEs have also been identified (RAGE), with their expression being enhanced during moderate uremia [241].

However, it is unclear whether all AGEs retained in uremia have a biological impact. It is also unknown whether one AGE is kinetically representative for the molecules of the group. Pentosidine has been advanced as a marker for these compounds; however, with respect to dialysis, inhomogeneous behavior of pentosidine has been demonstrated, a finding that parallels what is known for uremic toxins in general [242]. In an observational study, high, rather than low, carboxymethyllysine levels were linked to better outcomes [243].

This group of molecules highlights growing interest to dialytically remove more of the larger molecules that are retained in uremia, perhaps including more specific removal via adsorption columns.

Removal is also more efficient with membranes with a larger pore size [244]. This includes protein- leaking dialyzers [245]. This was shown in one study in which two groups of 13 hemodialysis patients were treated for six months with protein-leaking and non-protein-leaking dialyzers [245]. A significant decrease in total pentosidine values was observed with protein-leaking dialyzers (-43 percent), a decrease not observed with the other dialyzer.

This removal could counterbalance some of the deleterious effects of AGE accumulation (eg, apolipoprotein B production) [246]. Whether this removal will be sufficient to neutralize the clinical complications potentially attributed to the AGE is unclear.

Other middle molecules The kinetic behavior with dialysis of molecules with a molecular weight above 12 kD should be comparable with that of the somewhat smaller middle molecules. Several peptides of molecular weight above 12 kD, which can be recovered from uremic sera, ultrafiltrate, or peritoneal dialysis dwell fluid, dampen various polymorphonuclear cells functions involved in the killing of invading bacteria [247,248]. The exact concentrations in uremic sera or biological fluids were not reported in these studies.

Leptin, a 16 kD plasma protein that suppresses appetite [249] and induces weight reduction in mice [250], is retained in renal failure [251]. The increase in serum leptin levels is almost entirely due to a rise in free (non-protein-bound) concentration [251]; it has been suggested to play a role in the decreased appetite of uremic patients (see "Pathogenesis and treatment of malnutrition in maintenance dialysis"). However, there are conflicting findings concerning the possible biochemical role of leptin in renal failure (see "Leptin and end-stage renal disease"):

●Increased leptin is associated with low protein intake and loss of lean tissue [252]. Data suggest an inverse correlation in uremia between leptin and indices of nutritional status such

as serum albumin or lean body mass [253] and a direct correlation with C-reactive protein (CRP) [254].

●However, leptin levels are also elevated in obese people and are not necessarily related to reduced appetite. Body fat and serum leptin correlate positively in uremia [254].

The dinucleoside polyphosphates are molecules characterized by the presence of two nucleotides at the extremes, linked by a variable number (mostly two to six) of phosphates. They have a molecular weight of approximately 1000 Dalton and are protein bound. The diadenosine polyphosphates induce proliferation of smooth muscle cells [255,256] and enhance free radical production by leukocytes [257]. Uridine adenosine tetraphosphate is a strong vasoconstrictor, which is released by endothelium [258].

Fibroblast growth factor-23 (FGF-23), a middle molecule enhancing renal phosphate excretion and associated with disturbances of bone metabolism, is increased from the early stages of CKD and has been linked with increased mortality. Although generally considered a marker molecule rather than a toxin, increasing FGF-23 up to uremic levels induced cardiac hypertrophy in mice [259]. The role of FGF-23 in mortality and other outcomes is discussed elsewhere. (See "Patient survival and maintenance dialysis", section on 'Disorders of mineral metabolism'.)

Further middle molecules of potential importance are complement factor D, adrenomedullin, atrial natriuretic peptide (ANP), ghrelin, resistin, immunoglobulin light chains, neuropeptide Y, and various cytokines. Plasma IL-6 is independently associated with mortality in patients with CKD [260]. In vitro, however, neither IL-6 nor any of the other major cytokines induced leukocyte activation at concentrations observed in dialysis patients [261]. The only cytokine that produced an effect was TNF-alpha, however, only at high concentrations, which are no longer observed with dialysis strategies employed today.

GENERAL REMOVAL STRATEGIES The main strategies that have been used to decrease uremic solute concentration are conventional hemodialysis and peritoneal dialysis. However, dialysis is nonspecific and also removes essential compounds. In addition, lipophilic compounds, which may be responsible at least in part for functional alterations in uremia, are inadequately removed by current dialysis strategies.

With maintenance hemodialysis, treatment with high-flux membranes was suggested to provide superior removal of middle molecules, possibly resulting in improved survival. However, the Hemodialysis (HEMO) study found no overall survival difference with high- versus low-flux membranes at primary analysis [17]. However, differences were found for the entire cohort in cardiovascular mortality and in the subgroup treated for >3.7 years upon enrollment for overall mortality with secondary analysis [27]. In the same database, a direct relation between beta2- microglobulin (beta2-m) concentration and mortality was also demonstrated [28]. In addition, the Membrane Permeability Outcome (MPO) trial described above subsequently demonstrated better outcomes associated with high-flux compared with low-flux membranes in the population with serum albumin <4 g/dL [30]. Of note, this low serum albumin group composes a large section of the current population on dialysis.

Adding convection by increasing ultrafiltration and equivoluminous substitution with sterile saline or ultrapure dialysate (hemodiafiltration) further adds to middle-molecule removal [262-264]. In an observational analysis of the Dialysis Outcomes and Practice Patterns Study (DOPPS) database, a

survival advantage for convection was found, at least if sufficient exchange volumes were pursued

[212]. Large exchange volumes can be obtained if ultrapure dialysate is used for substitution (on- line hemodiafiltration). A study comparing convective strategies at strictly identical conditions showed superiority of postdilution hemodiafiltration and predilution hemofiltration as far as beta2-m removal was concerned. Postdilution hemodiafiltration was superior when considering other aspects

of solute removal (small, water-soluble and protein-bound compounds) [130]. With regard to

survival outcome studies, a limited Italian study showed better outcomes associated with on-line hemofiltration, compared with low-flux dialysis [31]. A larger, Italian, multicenter study showed superiority of convective strategies with regards to hemodynamic stability [213].

Two controlled trials showed no superiority of on-line hemodiafiltration over hemodialysis at primary analysis [214,215], but only in the subgroup analyses of the patients with the highest exchange volumes. This is a risk for selection bias since patients achieving the highest exchange volumes are potentially those with the best vascular access, which has itself been associated to better survival. One trial managed to maintain a large group of patients on high-volume exchange hemodiafiltration,

in a comparison to a group on hemodialysis [216]. In this study also, however, patients were

censored if they could not maintain the high-exchange volumes and were, to the best of our

knowledge, not included in an intention-to-treat analysis.

A previously ignored component of adequate dialysis is the treatment time. Lower morbidity and

mortality are observed in patients submitted to long dialysis sessions [265,266]. Compounds may be cleared more efficiently with continuous or long-lasting strategies because removal is more gradual (see "Technical aspects of nocturnal hemodialysis"). In a strictly comparable setting for all dialysis parameters except treatment length, prolonging high-flux hemodialysis from four to eight hours resulted in a time-dependent increase in beta2-m and phosphorus removal [88]. However,

prolonging hemodialysis did not decrease protein-bound solutes in a similar study [127].

Optimal removal for each type of molecule may be obtained with a different type of extracorporeal treatment (eg, by using large-pore membranes and/or dialyzers or devices with a high adsorptive capacity for some or several of the uremic toxins). This includes, for example, protein-leaking membranes, which are designed to allow passage of albumin, other similarly sized proteins, and uremic toxins in the 35 to 60 kD size range [267].

Research centers on alternative measures adding to the removal capacity of classical dialysis. These measures include adsorption [95] or changing the physical conditions within the dialyzer, enhancing the free fraction of protein-bound toxins [268].

Removal is also influenced by intestinal intake (especially for the protein-bound solutes) and preservation of renal function:

●Intestinal uptake can be reduced by influencing dietary uptake or by oral administration of absorbents. Approaches that have been shown to result in a decrease in concentration include a low-protein diet [137,269], administration of prebiotics such as resistant starch [270], or probiotics such as bifidobacterium [104,271]. The active intestinal sorbent AST-120 (Kremezin R) has essentially been associated with absorption of indoxyl sulfate [186] and subsequent preservation of renal function, but this substance also absorbs p-cresol as well as other compounds [133,134].

●Preservation of residual renal function may also be an important manner to pursue additional removal of retention solutes. In a number of studies, a correlation was found between concentration of several uremic toxins and residual renal function [272,273]. This relationship appeared stronger than that to adequacy of dialysis. Residual renal function was better preserved after administering the sorbent AST-120 [111,187,188].

Finally, it should be considered that, in uremia, not only strategies decreasing solute concentration are important, but interventions countering their biological impact also play a role [6]. Typically, this can be obtained with already developed drugs, such as angiotensin-converting enzyme (ACE) inhibitors in neutralizing Ca influx due to symmetric dimethylarginine (SDMA) [274], but also with drugs still to be developed, and reaches a much broader population than with removal strategies (ie, not only the 0.1 to 0.2 percent of the global population with stage 5 chronic kidney disease [CKD], but all those with stage 3 CKD or more [5 to 10 percent]). Similarly, interventions increasing metabolic degradation of toxic solutes may be considered [57]. Some experimental studies suggest that the concentration of indoxyl sulfate can be decreased by interventions increasing the activity of sulfotransferase, the enzyme responsible for the sulfation of the parent compound, indole [275].

SUMMARY The uremic syndrome is a complex mosaic of clinical alterations that may be attributable to one or more of these different solutes. Knowledge about the nature and kinetic behavior of the responsible compounds may help when new therapeutic options are considered in the future.

The following factors, which have frequently been neglected, may interfere with uremic solute concentrations and their impact on biological functions:

●In addition to classical sources of uremic solutes such as dietary protein breakdown, alternative sources such as environmental contact, food additives, natural stimulants (coffee, tea), intake of herbal medicines, or addiction to psychedelic drugs may play a role in uremic toxicity.

●Many solutes with toxic capacity enter the body through the intestine. Changes in the composition of intestinal flora or changes in intestinal production, absorption, transfer, and metabolization may alter serum concentration of these toxins.

●Some uremic solutes interfere with functions that directly affect the biochemical action of other solutes (eg, the expression of parathyroid hormone [PTH] receptors, the response to 1,25(OH)2 vitamin D3, as well as the protein binding and breakdown of several other solutes).

●Most patients with renal failure use multiple medications. Interference of drugs with protein binding and/or tubular secretion of uremic solutes influences their biological effect.

●Lipophilic compounds may be responsible at least in part for functional alterations in uremia; they are inadequately removed by current dialysis strategies.

●The main strategy that has been used to decrease uremic solute concentration is dialysis. However, dialysis is nonspecific and also removes essential compounds.

●Uremic solutes accumulate not only in the plasma, but also in the cells, where most of the biological activity is exerted. Removal of intracellular compounds during dialysis through the cell membrane may be hampered, resulting in multicompartmental kinetics and inadequate detoxification, unless dialysis length and/or frequency are increased.

Lower morbidity and mortality are observed in patients submitted to long dialysis sessions, although the data mentioned have been obtained in an uncontrolled setting [265,266]. A controlled trial demonstrated a positive impact of frequent nocturnal hemodialysis versus conventional dialysis on left ventricular mass and quality of life [276]. A randomized, controlled trial compared six times weekly to thrice-weekly hemodialysis over a 12-month follow-up period [277]. Six times weekly dialysis was associated with a decrease in the primary outcome, which was a composite of death or change in left ventricular mass, and with improved control of hypertension. Compounds may be cleared more efficiently with continuous or long-lasting low-efficiency strategies because removal is more gradual. (See "Technical aspects of nocturnal hemodialysis".)

Biochemical alterations are provoked by a broad spectrum of compounds:

●Some are small and water soluble (eg, urea, the guanidines, phosphate, oxalate).

●Some are lipophilic and/or protein bound (eg, p-cresyl sulfate, 3-carboxy-4-metyl-5-propyl-2- furanpropionic acid [CMPF], homocysteine [Hcy], indoles).

●Some are larger and in the middle-molecule range (eg, beta2-microglobulin [beta2-m], PTH, advanced glycosylation end products [AGEs]).

●Some compounds from one group may impact generation and behavior of compounds from another group (eg, guanidines increasing generation of tumor necrosis factor-alpha [TNF- alpha]) [35].

Optimal removal for each type of molecule may be obtained with a different type of extracorporeal treatment (eg, by using large-pore membranes and/or dialyzers or devices with a high adsorptive capacity for some or several of the uremic toxins). Adsorption with the dialysis devices currently available is of little importance. More specific devices with large surface areas must be developed before adequate adsorption can be obtained.

Removal is also influenced by intestinal intake and preservation of renal function:

●Intestinal uptake can be reduced by influencing dietary uptake, by oral administration of sorbents, or by influencing intestinal flora by administration of prebiotics or probiotics.

●Preservation of residual renal function may also be an important manner to pursue additional removal of retention solutes.

●Therapeutic intervention includes administration of drugs countering biological impact of uremic solutes and reaching a much larger population than with removal strategies.

Finally, the choice of marker molecules for uremic retention and dialytic removal should be reconsidered. It has become increasingly evident that the current markers, which are all small, water-soluble compounds (urea, creatinine), are not always representative in their kinetic behavior for middle molecules, lipophilic/protein-bound compounds, and even some other hydrosoluble compounds. On the other hand, it may be derisory to seek a marker that is representative for all solutes or even groups of solutes responsible for the uremic syndrome.