Discussion
As expected, low plasma HDL cholesterol concentrations were associated with increased CV and overall mortality and increasing HDL cholesterol levels up to 50 mg/dL were associated with improved survival in patients with ESRD receiving MHD. However, increments in HDL cholesterol concentrations >50 mg/dL were associated with a significant increase in CV and overall mortality. These findings are in contrast to the associations observed in the general population in whom increasing HDL cholesterol concentrations were associated with decreased CV mortality and improved survival. Furthermore, in MHD patients increments in total cholesterol to HDL cholesterol concentration ratios were associated with improved survival contrasting the findings in the general population. This observation further supports the finding that very high plasma HDL cholesterol concentrations are not associated with improved survival and instead may be associated with increased adverse outcomes in patients with ESRD. It should be noted that a low total cholesterol/HDL cholesterol concentration ratio may be due to a low plasma cholesterol concentration as opposed to a high HDL concentration. In such cases increased risk of CV and overall mortality may be due to the high burden of systemic inflammation which results in a significant fall in plasma cholesterol concentration. In the subgroup analyses, these paradoxical associations persisted, although they tended to be stronger in diabetics, men, patients <65 years of age and those with higher ferritin and albumin concentrations.
In a series of earlier studies, we found that compared with the HDL from the healthy controls, the HDL from MHD patients exhibits markedly reduced antioxidant activity. This was associated with and, in part, due to the significant reduction of paraoxonase and glutathione peroxidase activity, the key HDL-associated antioxidant enzymes, which are crucial for its antioxidant–anti-inflammatory functions. The reduction of the antioxidant and anti-inflammatory properties of HDL can contribute to the atherogenic diathesis in the ESRD population. However, the association between high HDL cholesterol levels and increased CV and overall mortality in the subpopulation of ESRD patients shown here cannot be solely attributed to the impaired antioxidant and anti-inflammatory activities of HDL. This is because HDL cholesterol levels up to 50 mg/dL were associated with reduced CV and overall mortality in this population. Thus, despite its diminished antioxidant and anti-inflammatory functions, HDL appears to confer some protective action in the majority of ESRD patients. However, in the current study, we also found that the highest HDL cholesterol concentrations were associated with increased overall and CV mortality. This can be explained by a growing body of evidence which indicate that in the presence of oxidative stress and inflammation, HDL is transformed from an antioxidant and anti-inflammatory to a pro-oxidant, pro-inflammatory particle known as acute-phase HDL. In fact, Honda et al. have recently shown that in a cohort of Japanese MHD patients, higher HDL cholesterol concentrations were associated with higher levels of oxidized HDL. Furthermore, they found that elevated oxidized-HDL concentrations were associated with increased CV mortality. They concluded that excess oxidative stress may have yielded dysfunctional HDL in patients on MHD, and patients with high HDL cholesterol under these conditions may have enriched oxidized HDL which can then result in increased CV disease and CV mortality. Yamamoto et al. also recently noted that HDL of patients with ESRD on MHD had reduced antichemotactic activity and increased macrophage inflammatory cytokine response (tumor necrosis factor alpha, interleukin 6 and interleukin 1 beta), when compared with HDL from healthy controls. This is in agreement with the findings of Weichhart et al. who observed that while HDL isolated from healthy individuals inhibited the production of inflammatory cytokines by peripheral monocytes, HDL isolated from the majority of patients with ESRD did not show an anti-inflammatory property and many HDL samples even promoted production of inflammatory cytokines. Therefore, in the setting of inflammation (i.e. patients with ESRD, diabetes and elevated ferritin levels), HDL may not only be dysfunctional but may also have a deleterious effect by promoting inflammation and adding to the CV disease burden.
Another possible mechanism responsible for increased mortality observed with elevated HDL cholesterol levels in our patient population was recently described by Huang et al. in an in vitro study using early endothelial progenitor cells (EPCs), which served as a prognostic indicator of clinical atherosclerosis. In this study, it was demonstrated that while protective at low concentrations, moderate-to-high concentrations of HDL from healthy subjects paradoxically impaired EPCs and related angiogenesis in the absence of oxidized LDL hence providing in vitro mechanistic evidence of potential toxic effects of elevated HDL concentrations under some specific circumstances. In addition, defective unloading of the HDL's cholesterol cargo in the liver which is the final step in RCT may be another possible mechanism which has not been studied to date. Such a defect can lead to a highly atherogenic state despite marked elevation of plasma HDL cholesterol. Another possible mechanism that can explain the association between an elevated serum HDL cholesterol concentration and increased CV mortality is a deficiency of the enzyme cholesteryl ester transfer protein (CETP). CETP deficiency may possibly lead to the development of atherosclerosis despite high HDL cholesterol levels as reported by Nagano et al.. Also, there are studies which show that serum HDL cholesterol elevation caused by CETP-deficient mutations can be associated with a high prevalence of CV disease. Furthermore, Kimura et al. demonstrated that CETP may serve as a protective factor against vascular disease in MHD patients with elevated serum HDL cholesterol levels. Therefore, deficiency or inhibition of the CETP pathway can explain the association of an elevated serum HDL cholesterol concentration with an increase in CV mortality. Overall, the reason for the adverse outcomes seen with elevated HDL cholesterol concentrations in a subgroup of ESRD patients is presently unknown and requires further investigation. However, it is important to note that similar paradoxical associations between elevated HDL cholesterol level and mortality have also been reported in other patient populations such as those with non-ST-elevation myocardial infarction and coronary artery disease. Therefore, it is imperative that future studies focus on determining the mechanisms responsible for these observations.
Several limitations of our study should be mentioned. First, the current findings should be qualified given the observational nature of our study design. In addition, a mixed incident/prevalent MHD population was examined in this study. However, we did adjust for dialysis vintage in all case–mix models and also performed separate analysis for incident (vintage ≤6 months) dialysis patients as shown in Figures 3 and 4. Another limitation is the potential confounding of therapy with lipid-altering agents which was not examined because home medication data were not available systematically in this national cohort. However, given the results of the 4D, AURORA and SHARP studies, it is unlikely that the inclusion of such data would have resulted in significantly different associations. It should be mentioned that our studied cohort represented only 20% of the entire national DaVita database and even smaller fractions for HDL subgroup analyses. However, we do not believe confounding by indication was present given that in 2006, the decision to measure serum HDL levels was made uniformly at the clinic level and was not individualized. In addition, we believe the risk of selection bias was not high given that all DaVita dialysis facilities are under uniform administrative care, and all laboratory tests are performed in one single laboratory with optimal quality-assurance monitoring. Furthermore, we performed a sensitivity analysis to compare DaVita patients included and excluded in this study and did not find any major differences (data not shown).