The Microinflammatory State in Uremia: Causes and Potential Consequences

Causes of Inflammation Related to Renal Failure

Postsynthetic Modification of Proteins. Renal failure itself may possibly contribute to inflammation as a result of the accumulation of proinflammatory compounds or products of metabolism. AGE such as pentosidine are normally associated with diabetes mellitus (22), and levels of advanced lipoxidation end products are also increased in the plasma of patients with renal failure, irrespective of their blood glucose levels. Levels of these substances are increased in diabetes mellitus, generally as a consequence of prolonged hyperglycemia. They result from the reaction of glucose and other carbohydrates with protein, specifically with lysine, via nonenzymatic glycation and oxidation (glycoxidation) of proteins. In diabetes mellitus, these substances are thought to play a role in the progression of renal disease (23).

The kidney normally plays an important role in the metabolism of AGE, which are processed by glomerular filtration followed by tubular uptake and metabolism (24). Because most of the pentosidine in blood is bound to high-molecular weight proteins such as albumin, dialysis has little effect on AGE concentrations (25). AGE may play a role in activating mononuclear cells, thus directly inciting an inflammatory response (24). Inflammation, however, can also play a role in the production of AGE (26). Although AGE and related compounds clearly can modify proteins, including the vascular endothelium (27) and lipoproteins (28), and thus could play a significant role in the development of vascular disease in patients with renal failure, it is not clear why the increased concentrations of these compounds that are observed for all dialysis patients would lead to activation of the acute-phase response or to hypoalbuminemia in only a fraction of dialysis patients.

Although oxidative stress and other postsynthetic modifications of proteins that occur in parallel with decreasing GFR may well explain a component of the inflammation observed for dialysis patients, they do not explain the large skews in the frequencies of increased levels of interleukin-6 (IL-6) (10) or acute-phase proteins (17,19) that were observed among the dialysis patient population (Figure 1A) or the wide temporal variability that we observed (Figure 1B) (29). These observations are more consistent with an episodic process either driving inflammation for this population as a whole or superimposed on a low level of inflammation.

Oxidative injury and protein glycation have also been suggested to play a role, although neither of these processes has been demonstrated to directly increase either acute-phase protein levels or cytokine levels in dialysis patients. However, levels of advanced oxidation products increase as GFR decreases (30) and are closely related to levels of the AGE pentosidine (30). There is a correlation between these products and cytokines, specifically IL-1 receptor antagonist, tumor necrosis factor α (TNF-α), TNF soluble receptor 55, and TNF soluble receptor 75, and these products are associated with evidence of monocyte activation (30), providing at least a theoretical link between advancing loss of renal function and the onset of an inflammatory response.

Oxidative Stress. Renal failure reduces plasma antioxidant activity (31). The increase in oxidative stress is identified by an increase in malondialdehyde levels in red blood cell membranes and a reduction in levels of the reduced forms of glutathione. The loss of antioxidants (32), zinc, selenium, and vitamins C and E (33) occurs during renal failure or as a consequence of dialysis and may contribute to susceptibility to oxidative injury. Plasma glutathione peroxidase activity is also reduced in renal failure (34). Superoxide dismutase and glutathione peroxidase activities are decreased in plasma and in erythrocytes. Increased oxidative stress occurs even before dialysis is instituted. After initiation of dialysis, the opportunity for oxidative injury of circulating blood elements is also present (35). Dialysis does not reduce oxidative injury and may actually exacerbate oxidative stress (36). Increased oxidative stress leads to lipid peroxidation (37) and oxidative alteration of lipoproteins. Cell membranes are also subject to oxidative injury. Stenvinkel et al. (38) established that the decrease in the levels of reduced erythrocyte phospholipids in patients with pre-ESRD was greatest for patients judged to be malnourished, suggesting that there is a relationship between inflammation and oxidative injury.

Levels of IL-6, IL-1, and TNF-α have been found to be significantly increased in patients with renal failure, with no difference being observed between patients undergoing longterm dialysis and patients who have not yet undergone dialysis (39), although others reported an increase primarily among patients undergoing dialysis (40). The process of dialysis provides multiple opportunities for exposure to stimuli that might promote inflammation.

Dialysis Membranes or Techniques. Exposure of blood to bioincompatible dialysis membranes causes activation of circulating mononuclear cells (41) and has been implicated as a potential cause of inflammation induction within the population of patients with ESRD (42). Bioincompatible membranes, such as cellulosic membranes, activate white blood cells (43) and complement (44) and can even exert effects on residual renal function (45). Kaizu et al. (46) observed that dialysis membrane properties represented one factor that predicted serum IL-6 levels in their patients. Activation of cytokines occurred after dialysis using cellulose membranes (47), in contrast to biocompatible membranes (48).

Other investigators have suggested that even dialysis with biocompatible membranes may pose risks for activation of the acute-phase response. Honkanen et al. (41) observed very rapid and similar increases in the levels of IL-1β and the acute-phase protein SAA during 240 min of dialysis with cellulose, cellulose acetate, or polymethylmethacrylate membranes; the latter two types are biocompatible membranes. Cytokine production during in vitro dialysis of whole blood has been demonstrated (49), suggesting that interactions of circulating nuclear cells directly stimulate cytokine production. Dialysis also alters mononuclear cells so that they respond more vigorously to subsequent exposure to endotoxin (50). Therefore, dialysis with a variety of membranes is a potential source of inflammation, although most other investigators, including our group, failed to find convincing evidence of activation of the acute-phase response during dialysis with membranes other than cellulose membranes.

Reuse techniques and the number of times of reuse also may contribute to the interaction of blood with the dialyzer (51), leading to changes in protein loss and possibly changes in the acute-phase response. Pyrogenic reactions in the absence of septicemia are closely associated with reuse (52).

Dialysate Quality. The quality of water used to prepare the dialysate may also contribute to inflammation (53). Mounting evidence suggests that the use of less-than-sterile dialysate or back-leakage of lipopolysaccharide through the dialysis membranes can cause dialysis-related inflammation (54). Several groups recently prepared ultrapure, endotoxin-free water by membrane filtration of the dialysate and observed reduced levels of cytokines (55), which suggests either that monocytes may be activated by endotoxin that remains on the dialysate side of the membrane or that endotoxin can directly cross the dialysis membrane. This latter route may be of increased importance with the use of highly permeable membranes (56).

Neither the exposure of blood to dialysis membranes nor the use of less-than-ultrapure water in the dialysate can explain the inflammation variability observed within the dialysis population (Figure 1A). Furthermore, the acute-phase response also plays an important role in establishing albumin levels in patients undergoing continuous ambulatory peritoneal dialysis (57), a population exposed to neither lipopolysaccharide-containing dialysates nor dialysis membranes. Indeed, in our own patient population, CRP and SAA levels were significantly greater among patients undergoing peritoneal dialysis than among patients undergoing hemodialysis (57). Peritoneal dialysis may, however, carry its own special risk of exposure to mediators of inflammation. Exposure of the peritoneal membrane to plasticizers found in dialysates may be one source of inflammation among this population. Another potential source is transperitoneal access.

The majority of dialysis patients exhibit no evidence of activation of the inflammatory response, despite the fact that all patients are exposed to dialysis membranes and dialysates. The highly skewed distributions of CRP and SAA (Figure 1A) (17) and IL-6 (19) values within the hemodialysis patient population suggest that patient-specific processes, such as the type of vascular access, unrecognized infections, or variable responses to the dialyzer used, may play a role. Qureshi et al. (58) observed that patients with CRP levels of >20 μg/ml exhibited clinical evidence of infection, suggesting that patients with levels of acute-phase proteins or cytokines skewed to high values have clinical infections. However, Bologa et al. (19) specifically excluded patients with evidence of infection and observed that the distribution of IL-6 levels was similar to that we described for acute-phase proteins in hemodialysis patients (17).

Another significant issue is the variability of both cytokine and acute-phase protein levels with time. Cross-sectional studies demonstrated a strong correlation between CRP concentrations and albumin concentrations (17,19) and between CRP and IL-6 concentrations and mortality rates. Those measurements were made at one time point. In longitudinal studies of 37 hemodialysis patients, we observed that levels of two acute-phase proteins, i.e., CRP and α1-acid glycoprotein, varied significantly with time, with no change in dialyzer type or treatment, suggesting that non-dialysis-related processes might be responsible for altered expression of the acute-phase response in these patients (29). The variance in CRP levels was 2 orders of magnitude greater than that for albumin or transferrin (29). Therefore, factors other than dialysis, which are specific to individual patients, might also be responsible for initiating the inflammatory response. It is unclear what specific process is responsible for activation of the acute-phase response in dialysis patients; however, the process does not occur at all times in all dialysis patients.

Infectious Causes of Inflammation

Infections occur commonly among hemodialysis patients. This population is at increased risk of infection as a consequence of impaired humoral and cellular immunity (59) and vascular access, as discussed below.

In a longitudinal cohort study of the incidence of and risk factors for hospitalized cases of septicemia among diabetic and nondiabetic hemodialysis patients, using baseline data from the United States Renal Data System, it was reported that, during a 7-yr period, 11.1% of nondiabetic patients and 12.5% of diabetic patients experienced at least one episode of septicemia (60). Risk factors that predicted septicemia were low serum albumin levels and advancing age.

Clinical infections from actively infected grafts, as well as old clotted grafts that do not exhibit signs of active inflammation (61), are also infectious sources of inflammation and may play a very significant role in individual cases. The detection of graft infections may require ¹¹¹In white blood cell scans (61) or other procedures.

The prevalence of tuberculosis is increased among dialysis patients (62). Dialysis patients are also subject to other sources of infection, especially because of the prevalence of diabetes mellitus among this population. Indeed, diabetes mellitus was found to be a risk factor for inflammation in a cross-sectional study of 128 hemodialysis patients (58). Diabetes mellitus poses the risk of peripheral vascular disease and neuropathy, which are both risk factors for foot infections. Peritonitis presents an additional risk of infection for peritoneal dialysis patients (63). It was recently hypothesized that vascular disease itself may be infectious in origin (64). This is reviewed below.

Inflammation as a Cause of Vascular Injury

Possibly the most significant process that is correlated with inflammation is vascular disease. The relationship between these two processes is complex. Endothelial injury involving attachment to and subsequent invasion of the endothelium by macrophages plays an important role in the development of atherosclerosis. Atherosclerotic burden is therefore one potential cause of inflammation. It has been proposed that infections with Chlamydia pneumoniae or Helicobacter pylori may cause both inflammation and atherosclerosis (64). This connection has not yet been verified and, even if it later is, there is no clear connection between this infectious process and the accelerated atherosclerosis in renal failure. It is more likely that inflammation causes both the anthropometric and serologic changes that are attributed to malnutrition in dialysis patients and cause the vascular disease as well.

It was recently established, in large cross-sectional studies, that even small increases in the levels of either cytokines or acute-phase proteins predicted vascular disease among populations of otherwise healthy-appearing adults (4). This observation was based on the assumption, although it was not explicitly stated, that CRP and IL-6 levels are characteristic for specific patients, as LDL levels are. The finding that even small increases in the levels of these markers of inflammation predicted future vascular events generated the term “microinflammatory state.” Markers of inflammation have the same predictive values for dialysis patients (5,6); however, for such patients we know that inflammation is not characteristic for each patient but instead varies with time (29). Elevations in the levels of acute-phase proteins are not subtle, and the incidence of vascular disease is greatly increased. The relationship between inflammation and vascular disease in this population is strong enough to obscure other well known risk factors. Although total cholesterol (and LDL cholesterol) levels are powerful predictors of vascular disease in the general population, the inverse is true for dialysis patients (3).

The relationship between the variability of the acute-phase response and vascular disease in the dialysis patient population remains unknown. Several possibilities present themselves. One is that even a short period of inflammation may result in future vascular events. Inflammation clearly alters the plasma protein composition and both endothelial and lipoprotein structure and function to favor vascular injury, and the time between inflammation and vascular occlusive disease may not be short. After vascular injury occurs, it may be self-propagating, and the inflammatory cells (monocytes) involved in vascular injury may contribute to a chronic increase in the levels of acute-phase proteins and cytokines. We observed, however, that both cytokine and acute-phase protein levels decreased to baseline values after episodes of acute inflammation (29).

There are several ways in which inflammation can promote vascular injury, i.e., through alterations in lipoprotein structure and function, changes in the composition of plasma proteins, alterations in the vascular endothelium, and changes in the expression of specific ligands on the surfaces of platelets, neutrophils, and mononuclear cells.

Several acute-phase proteins are directly associated with vascular disease. One of these is fibrinogen (71). Fibrinogen levels are increased in the subset of patients with inflammation/malnutrition (12), and the levels of this protein, like those of CRP and other acute-phase proteins, vary with time among dialysis patients. Therefore, a specific episode of inflammation not only promotes alterations in the vascular adhesion of mononuclear cells but also increases the concentration of this coagulation factor in plasma. Levels of lipoprotein(a) are also increased among patients with ESRD and are correlated with both vascular injury (carotid artery intimal thickening) and inflammation/malnutrition (12), suggesting that this protein may also be regulated as part of the acute-phase response. Although the concentrations of lipoprotein(a) are generally genetically controlled, the levels of this lipoprotein increase in patients with renal failure (72), in conjunction with increased levels of acute-phase proteins; this increase is in turn associated with increased mortality rates for the hemodialysis patient population (73). This lipoprotein is highly atherogenic when its levels are genotypically increased. The increase in the dialysis patient population may be a component of the acute-phase response and may also contribute to vascular injury.

Apolipoprotein A-I is a negative acute-phase protein, and its synthetic rate is decreased during inflammation (74). SAA displaces apolipoprotein A-I from HDL and alters HDL structure and function (Figure 2), causing HDL to become adherent to the vascular endothelial surface (75). Paraoxonase activity is reduced, and HDL is then no longer capable of reducing oxidized LDL. Furthermore, HDL that has been thus altered acts as a chemoattractant to macrophages, serving as a bridge between them and the endothelial surface (Figure 2) (76).

Inflammation also alters the expression of soluble vascular cell adhesion molecule-1 (77) and soluble intercellular adhesion molecule-1, i.e., adhesion molecules that promote monocyte binding to endothelial cells and are associated with vascular disease (78).

P-selectin levels are increased during acute inflammation, increasing the adherence of platelets to the endothelium. P-selectin (platelets) and E-selectin (endothelial cells) are surface glycoproteins that mediate leukocyte rolling on the endothelial surface, a step that immediately precedes adhesion. P-selectin is able to initiate the cascade of events that increases cell adherence and leukocyte infiltration into injured tissues, by first promoting leukocyte rolling along the vascular endothelium (79). P-selectin levels are increased by hemodynamic stress (80) and are regulated by TNF-α during inflammation (81). Therefore, inflammation also alters the expression of endothelial adhesion molecules that favor the attachment of mononuclear cells and neutrophils, which then initiate oxidative bust activity and further the inflammatory response, contributing to increased oxidative modification of lipids and proteins. Patients with renal failure have a decreased antioxidative reserve, as described above, and thus are less likely than others to be able to prevent this process from becoming self-perpetuating.

Effects of Inflammation and Oxidation of Lipoproteins

During inflammation, SAA is incorporated into HDL and displaces apolipoprotein A-I. The triglyceride content of HDL is thus increased (Figure 2). Both paraoxonase and platelet-activating factor-acetyl hydrolase are thought to play a role in preventing the peroxidation of other lipoproteins, including LDL. The activities of both paraoxonase and platelet-activating factor-acetyl hydrolase are reduced. Paraoxonase activity remains suppressed long after levels of the acute-phase proteins have returned to normal after a single episode of minor surgical stress (82). Paraoxonase plays an important role in preventing the oxidation of HDL. Oxidative modification of LDL has been implicated in atherogenesis (83). This interaction between acute inflammation and lipoprotein structure (specifically HDL structure) provides one link between accelerated vascular disease and inflammation among patients with ESRD. Inflammation itself can directly yield lipoprotein oxidation, in addition to removing potential defense mechanisms against lipoprotein oxidation.

The leukocyte enzyme myeloperoxidase oxidatively modifies LDL via the oxidative nitration of tyrosine and other modifications of the tyrosine residue (Figure 2) (84). Both ascorbate and probucol inhibit the myeloperoxidase-catalyzed oxidation of LDL, providing the possibility for therapeutic intervention. Ceruloplasmin carries copper and is an acute-phase protein whose concentrations increase during inflammation. It also acts as an oxidant for lipoproteins during acute inflammation.

In the absence of inflammation, the antioxidant activity of HDL is usually normal in hemodialysis patients (85), although total HDL levels are reduced. Reductions in HDL particle size are linked to insulin resistance and accompany reductions in LDL particle size and hypertriglyceridemia.

The predictors of death resulting from cardiovascular causes are low HDL cholesterol levels and high triglyceride/HDL ratios. Indeed, patients with the lowest total cholesterol levels, not those with the highest LDL levels, succumb to vascular disease (3). The combined effects of inflammation and oxidative injury on lipoprotein structure and function may provide some insight into this apparent contradiction.

Inflammation among patients with renal failure is common and may present with markers of malnutrition, e.g., reduced albumin or prealbumin levels, decreased muscle mass, or high Subjective Global Assessment scores. Inflammation contributes to both anemia and erythropoietin resistance. Inflammation predicts death, including death resulting from cardiovascular causes, and is likely to explain the relationship between malnutrition and death resulting from cardiovascular causes among this population. The cause of inflammation is likely to be multifactorial and, although it may reflect underlying vascular disease, inflammation is also a potent factor causing vascular injury. This effect is most likely mediated by changes in lipoprotein structure and function, alterations in endothelial function, adhesion to mononuclear cells and platelets, and alteration of the composition of plasma proteins to promote hemostasis and improve host defenses, with the unwanted consequences being vascular injury and death.