Abstract
It has been known for a long time that chronic kidney disease (CKD) is associated with dyslipidemia, but the full extent of abnormalities has been appreciated only recently, because routine laboratory tests fail to disclose the entire spectrum of lipid abnormalities. Lipids, particularly HDL cholesterol, are predictive of cardiovascular events, but a paradoxic inverse relation between cholesterol concentration and cardiovascular death has been noted in uremic patients. This currently is thought to be explained by the confounding effect of microinflammation and possibly calcification, but this is not definitely proved. Several retrospective analyses that included patients with mild or moderate CKD documented benefit from lowering of cholesterol by statins. In contrast, the Die Deutsche Diabetes Dialyse (4D) study and a small Scandinavian study failed to show a benefit from lowering of cholesterol by statins in ESRD. Pathomechanistically, it is possible that nonclassical pathomechanisms override statin-sensitive mechanisms as also suggested by the observation that statins fail to reduce carotid intima-media thickening. Although, experimentally, exposure to lipids (particularly oxidized lipids) aggravates progression, data on the effect of statins on progression in patients with CKD are not definite. The most likely explanation is that the impact of numerous confounders obscures their effect on progression. The increase in urinary protein excretion of patients who are treated with statins had been a cause of concern, but the underlying mechanism (i.e. interference with proximal tubular reabsorption of protein) meanwhile has been well documented.
Since the days of Richard Bright, it has been known that in patients with advanced renal disease, the serum is “milky,” pointing to the presence of hyperlipidemia. More recently, interest in hyperlipidemia of renal disease has been raised by the consideration that the known abnormalities in the lipid spectrum of renal patients might predispose to accelerated atherosclerosis (1). Indeed, the Seattle Group found excess death from coronary heart disease in a large proportion of the first cohort of patients who had been started on hemodialysis in Seattle (2).
Lipid Abnormalities in Renal Disease
The magnitude of lipid changes is not disclosed fully by routine laboratory chemistry, which usually shows normal total and LDL cholesterol, low HDL cholesterol, and high triglycerides. As shown in Table 1, more sophisticated analysis shows striking abnormalities: Increased VLDL; remnants and intermediate-density lipoproteins (3); prolonged persistence of postprandial chylomicron remnants (4); accumulation of small dense LDL; modification of apolipoproteins by glycation, oxidation, and carbamoylation (5); increased lipoprotein(a) [Lp(a)] (6); and accumulation of noncardioprotective acute-phase HDL (7,8). An interesting proposal was made by Shoji et al. (9), who showed that intermediate-density lipoproteins were an independent risk factor for aortic atherosclerosis. They also proposed calculation of non-HDL cholesterol (the sum of LDL and VLDL cholesterol). The latter is particularly elevated in renal failure (10).
Lipid spectrum in renal diseasea
Prognostic Value of Dyslipidemia in Chronic Kidney Disease
In a prospective study, Degoulet et al. (11) made the unexpected—and at first counterintuitive—observation that in dialysis patients, low cholesterol concentrations were associated with high mortality. This key observation meanwhile was confirmed in numerous studies (12,13). In a cohort of 823 patients who were admitted for dialysis, Liu et al. (13) found in the overall cohort that an increased serum cholesterol was associated with a decreased risk for all-cause mortality, whereas in the absence of inflammation/malnutrition (22% of the cohort with serum albumin >3.6 mg/dl, C-reactive protein <10 mg/L, and IL-6 <3.09 pg/ml), an increment in baseline serum cholesterol was associated with increased mortality, as it is in the general population. This study demonstrates that the predictive value of cholesterol is confounded by concomitant inflammation.
Another controversial issue in the past had been the clinical implications of the changes in Lp(a) concentration. Lp(a) is correlated positively with higher cardiovascular risk in the general population (14) but strongly dependent of the genotype and the concomitant isoproteins of apolipoprotein(a) [apo(a)]. Individuals with low molecular weight isoforms have the highest serum Lp(a) concentrations. In renal patients, the average Lp(a) concentration is particularly increased, but the absolute serum concentration of Lp(a) is less predictive than the isoform itself. The prognosis is particularly adverse in patients with the low molecular weight apo(a) isoform compared with the high molecular weight isoform (15).
Evidence for Accelerated Atherosclerosis in Chronic Kidney Disease
Cross-sectional studies documented the high prevalence of carotid artery atherosclerosis and of atherosclerotic peripheral artery disease (16). Autopsy studies showed an excessive prevalence of advanced coronary atherosclerosis (17), and this has been confirmed by coronary angiography (18) and electron beam computed tomography (19). Coronary calcification is regarded as an index of coronary atherosclerosis. Although in renal patients calcification may not be a specific expression of coronary atherosclerosis (and partially reflect media sclerosis as well), longitudinal studies showed accelerated calcification of the coronaries compared with that in nonrenal control subjects (19,20), and this was true even in early renal disease (21).
In the first description of coronary artery disease of dialyzed patients (2), the authors postulated that in uremic patients, atherogenesis was accelerated. Indeed, several groups showed accelerated aortic atherosclerosis in the experimental model of the apoE knockout mouse with reduced renal function (22–24). This is seen even with minor reduction of renal function (e.g., uninephrectomy [22]) and responds to administration of angiotensin receptor blockers (25).
Dyslipidemia and Progression of Renal Disease
It had been known for a long time that in diverse models of renal damage, high-lipid diets or other maneuvers to induce hyperlipidemia aggravate glomerulosclerosis and progressive renal dysfunction (26–28). Oxidized lipids are particularly injurious in this context (29,30). Although various parameters of dyslipidemia predict accelerated progression in humans (31), a beneficial effect of statins on progression has not been documented beyond doubt. In several studies and meta-analyses that examined the effect of lipid lowering by statins, a trend for slower progression was seen (32), but this finding has not been consistent and significant (33). The most likely explanation is that in renal patients, the adverse effect of dyslipidemia (and the benefit from its reversal), which has been documented clearly in animal experiments, is confounded and overridden by factors that are difficult to control fully even in prospective studies (e.g., BP, proteinuria). In populations with cardiovascular disease, a recent meta-analysis showed that statin therapy seems to reduce proteinuria modestly and results in a small reduction in the rate of kidney function loss (34).
Effect of Statins on Cardiovascular Disease in Patients with Chronic Kidney Disease
In large, prospective statin trials, post hoc analysis of the proportion of the included patients who had impaired renal function (mainly stages 2 and 3 chronic kidney disease [CKD], e.g., in the Anglo-Scandinavian Cardiac Outcomes Trial [ASCOT] [35] or Cholesterol and Recurrent Events [CARE] study [36]), suggested a benefit. In the CARE study, this depended on baseline GFR and was seen primarily in the small group of patients with a GFR of <40 ml/min (37).
A similar observation had also been made in patients who received a transplant in the Assessment of Lescol in Renal Transplantation (ALERT) study. There was a tendency for fluvastatin to reduce cardiovascular death (38), including patients with reduced renal function. In a long-term open follow-up observation, the reduction of cardiovascular events became statistically significant (39).
These studies left unresolved the issue of whether statins lowered significantly cardiovascular death in patients with stages 4 and 5 CKD. This possibility was not a priori certain. Fahti et al. (40) drastically lowered LDL cholesterol by atorvastatin in patients with advanced CKD as well as in nonrenal patients with coronary artery disease. As shown in Table 2, this maneuver reduced the maximum intima-media thickness in nonrenal patients with coronary artery disease but failed to affect this index in patients with CKD.
Intensive lipid-lowering effect on carotid intima-media thickness: Comparison of renal and nonrenal patients (39)a
To resolve this issue, in the prospective Die Deutsche Diabetes Dialyse (4D) study, 1255 patients who had type 2 diabetes and had been on hemodialysis for <2 yr were randomly assigned to receive placebo or 20 mg/d atorvastatin (41). In the verum group, LDL cholesterol decreased by 42% from 125 to 72 mg/dl compared with essentially no change (−1.2%) from 121 mg/dl at baseline in the placebo group during the 4-yr observation period. These values correspond to what LaRosa et al. (42) had achieved in his widely known Treating to New Targets (TNT) study, which documented impressive reduction of cardiovascular end points. Although LDL cholesterol was lowered equally effectively in the 4D study, there was a minor trend at best for an improved outcome with respect to the primary composite end point (cardiac death, fatal myocardial infarction, stroke), which failed to reach statistical significance, however. This finding is in full agreement with a recent small Scandinavian study that found significant lowering of cardiovascular end points in patients with pre–end-stage CKD but no effect whatsoever with the same dose of atorvastatin in the patients who were on hemodialysis (43).
The reasons for this unexpected failure to reduce the primary end point despite substantial lowering of LDL cholesterol are unresolved. It might be due to the operation of alternative pathomechanisms such as inflammation or calcification (44), which might override the deleterious effect of dyslipidemia. Certainly cardiovascular death in these patients was due primarily to noncoronary (or nonischemic) causes. For this reason, the power of the study in retrospect was inadequate. Another possibility to consider is that patients who would have been particularly susceptible to statin treatment may have died before reaching ESRD, because a major proportion of renal patients die from cardiovascular causes before reaching ESRD (45). In this context, it is of note (Table 3) that the major cause of death was sudden death and congestive heart failure, in good agreement with data of other studies and the US Renal Data System (www.usrds.org). Indeed, death from confirmed coronary artery disease was lowered by 19% per mmol lowering of cholesterol even in the 4D study, but patients who died from coronary heart disease composed only 9% of the patients who died. We shall have to wait for ongoing studies such as the Study of Heart and Renal Protection (SHARP) (46) and Heart and Renal Protection (HARP) Study (47,48) to be able to make evidence-based recommendations for the treatment of dyslipidemic patients who are in terminal renal failure.
Causes of death in the 4D studya
What Practical Conclusions Can We Draw for Patients’ Treatment on the Basis of the Current Incomplete Knowledge?
The cardiovascular risk is excessive in patients with even minor renal dysfunction (49–51). However, at this stage of CKD, classical cardiovascular risk factors have an impact that is similar to what is seen in nonrenal patients (48). Therefore, at that stage of CKD, the same rationale applies as in the similar high-risk population of diabetes (52): in our opinion, the high cardiovascular risk justifies the use of statins as if it were secondary intervention (as in patients with a history of coronary heart disease) and not primary intervention.
The current data do not support the general use of statins in patients who are on dialysis. In the 4D study, at best, a delayed trend of a reduction of the primary cardiovascular end point was observed (albeit much less pronounced compared with patients without renal disease and statistically NS). In retrospect, the study was underpowered to evaluate specifically coronary death. In our opinion, in dialysis patients with documented coronary heart disease, the use of statins can be justified, although the benefit is not yet proved, because the adverse effect profile of the statin was not altered in renal failure; adverse effects indeed were extremely low with no cases of rhabdomyolysis and no excess frequency of creatine kinase elevation.
- © 2006 American Society of Nephrology