Could Phosphate Provide a Second Chance for Statin Therapy in Kidney Failure?

Individuals with CKD are at markedly higher risk of cardiovascular disease and mortality than those without CKD (¹). The excess risk of these outcomes is particularly evident for those who transition to kidney failure (²). This has led to an enormous amount of attention on targeting classic cardiovascular risk factors in patients with CKD in the hopes that more aggressive risk factor modification could reduce their excess risk of cardiovascular mortality, the leading cause of death in individuals with CKD. Hyperlipidemia is an established risk factor for cardiovascular disease and death, and multiple large, randomized controlled trials enrolling individuals at high risk for cardiovascular disease have proven that reducing elevated cholesterol concentrations via β-hydroxy-β-methylglutaryl-CoA reductase inhibitors (hereafter referred to as statins) is safe and efficacious for the primary or secondary prevention of cardiovascular disease events (³,⁴). Importantly, however, these landmark studies routinely excluded individuals with severe kidney disease, making the efficacy of statin therapy for reducing cardiovascular disease risk unclear in those with kidney failure. Given that individuals requiring maintenance dialysis are at exceptionally high risk of cardiovascular risk, it was only natural to hypothesize that aggressive lipid-lowering therapy using statins would lower rates of cardiovascular disease events and mortality in this patient population.

Surprisingly, however, three seminal studies enrolling thousands of patients with kidney failure showed that despite demonstrating substantial efficacy in lowering LDL concentrations, statin therapy (atorvastatin, rosuvastatin, or simvastatin plus ezetimibe) did not lower rates of cardiovascular disease events or mortality as compared with placebo (^56–7). Moreover, secondary or post hoc analyses attempting to find a subset of participants of these studies for whom statin therapy may be efficacious proved to be largely null. Much has been written about the negative nature of these findings, with potential explanations ranging from lack of sufficient power to detect relatively modest effects to the possibility that dyslipidemia does not represent the major etiologic pathway leading to cardiovascular disease in patients with kidney failure. In support of this latter possibility, sudden cardiac death represents a disproportionately common cause of cardiovascular-related mortality in patients with kidney failure (⁸), suggesting that classic atherogenic pathways may not be the predominate cause of cardiovascular disease for those on dialysis. Whatever the correct interpretation of the data, the findings from large clinical trials have been stubbornly clear to date: there appears to be no good justification for prescribing statins to individuals on maintenance hemodialysis (or peritoneal dialysis per the subgroup findings of the Study of Heart and Renal Protection trial [⁵]) to reduce cardiovascular disease events or mortality. However, this does not definitively rule out the possibility that there may still be a subset of individuals who could benefit—such as individuals requiring secondary prevention after an acute event (myocardial infarction or stroke) or individuals within the first 3 months of initiating dialysis when cardiovascular event rates are the highest—supporting continued efforts to identify individuals who might still benefit from therapy.

Along these lines, the study by Massy and colleagues in this issue of CJASN (⁹) may provide new hope for identifying individuals with kidney failure who might benefit from statin therapy—in essence by using their serum phosphate concentrations as a potential barometer of statin efficacy. The motivation for this study was derived from recently published experimental data showing that excess phosphate stimulates β-hydroxy-β-methylglutaryl-CoA reductase activity through aberrant activation of sterol regulatory element-binding protein and cleavage-activating protein–sterol regulatory element-binding protein 2 signaling (¹⁰). In in vivo and ex vivo experiments, this was shown to lead to severe atheromatous lesions in apoE knockout mice due to enhanced cholesterol synthesis in vascular smooth muscle cells and macrophages. Building on these intriguing data, Massy and colleagues conducted a post hoc analysis of A Study to Evaluate the Use of Rosuvastatin in Subjects on Regular Hemodialysis: An Assessment of Survival and Cardiovascular Events (AURORA) trial (⁶) to examine whether serum phosphate trajectories modified the association of statin therapy with major adverse cardiovascular events and all-cause mortality. The investigators further examined whether the findings could be replicated in the Deutsche Diabetes Dialyze Studie (4D) trial (⁷).

The crux of the study findings was that the association of statin therapy with major adverse cardiac events (MACE) and all-cause death appeared to differ depending on time-dependent serum phosphate concentrations: the lower the serum phosphate, the greater the protective association of statin therapy for MACE and all-cause mortality. It is important to note, however, the statistical test of interaction used to determine whether the risk of MACE and mortality differed by serum phosphate concentration was only significant for the latter outcome. In secondary analyses stratified by cardiovascular versus noncardiovascular death, there were no significant differences in the association of statin therapy with either outcome as a function of time-dependent serum phosphate (predefined P value for interaction >0.05 for both). When analyses were repeated using baseline serum phosphate concentrations analyzed in tertiles, the findings were directionally similar (assignment to rosuvastatin therapy was associated with numerically lower risk of MACE and all-cause death in the lowest tertile of serum phosphate as compared with the higher two tertiles), although the test for statistical interaction was NS for either outcome. Results of the latter analyses did not differ when adjusting for time-dependent LDL concentrations. Notably, the findings were unable to be replicated in the 4D study, which showed no differences in the association of assignment to atorvastatin therapy with MACE or mortality by serum phosphate concentrations, whether modeled on a continuous scale (time-dependent serum phosphate) or categorically in tertiles (baseline serum phosphate).

There are a number of strengths to this study, including leveraging the rich database for two landmark studies (AURORA and 4D), a sophisticated set of analyses, and the attempt to replicate the findings of the AURORA study in the 4D study. There are also weaknesses that should be considered when interpreting the result. First and foremost is the limitation of using serum phosphate concentrations to infer any causal associations between excess serum phosphate and statin efficacy. Serum phosphate concentrations represent an integrated measure of multiple different inputs and outputs, including nutrition, bidirectional flux between soft tissue and bone, and elimination with dialysis, among other factors. The issue of nutrition is particularly relevant because higher serum phosphate is often a good marker for individuals who demonstrate poor adherence with all recommended treatments, including consuming a low-phosphate diet. Thus, statins may be less efficacious in patients with higher serum phosphate because they are less adherent to recommended medical care in general, and not because of anything related to serum phosphate itself. Arguing against this explanation is the finding that serum LDL concentrations decreased to a similar extent in AURORA participants randomized to statin therapy, irrespective of whether they were in higher or lower phosphate trajectories, suggesting both groups demonstrated good adherence in taking the prescribed statin in the intervention arm. Nonetheless, this does not preclude the possibility that those in the higher phosphate group did not adhere to other therapeutic recommendations directly relevant to cardiovascular health, such as fluid restriction, sodium intake, or diabetes management. Second, it is distinctly odd that the greater efficacy of statin therapy associated with lower serum phosphate in lowering the risk of death was only really observed for noncardiovascular disease deaths. If the hypothesis is that lower serum phosphate allows for statins to work more effectively in lowering LDL concentrations, the opposite finding would have been expected. Finally, not only were the findings not able to be replicated in the 4D study, they were in some cases in the opposite direction. For example, in analyses stratified by time of follow-up (before versus after 2.5 years), higher serum phosphate concentrations appeared to be associated with greater, not lower, efficacy of statins for MACE that occurred after 2.5 years of follow-up in participants of the 4D study.

What do the findings of this study tell us? Ultimately, very little about any potential modulatory effects of serum phosphate itself on the efficacy of statins in humans. Nonetheless, the findings are provocative and provide justification to conduct further mechanistic studies as to whether serum phosphate directly or indirectly impedes statin efficacy. Should the evidence base become robust enough to support such a conclusion, this would set the stage for clinical trials testing whether a combination of phosphate lowering and statin therapy may be necessary to realize the salutary effects of statin therapy for cardiovascular outcomes in patients with kidney failure (although it should be noted that sevelamer reduces both serum phosphate and cholesterol but has never been shown to confer lower risk of death when compared with other, largely calcium-based phosphate binders or placebo). Even if serum phosphate does not directly modulate the efficacy of statin therapy, it is conceivable that it could still serve as a marker of some other factor (e.g., nutrition, adherence with medical therapy, altered mineral metabolism) that can help discriminate which patients may best benefit from prescription of statins and which may not. Given the exceptionally high rates of cardiovascular disease events and mortality in this population, efforts to determine whether statin therapy may still have a place in primary or secondary prevention of cardiovascular risk, vis-à-vis serum phosphate concentrations or another marker, are well worth the effort.

Disclosures

O.M. Gutiérrez reports receiving grant funding and honoraria from Akebia and Amgen; receiving honoraria from Ardelyx, AstraZeneca, and Reata; serving as an associate editor of CJASN; receiving grant funding from GlaxoSmithKline; and serving on the data monitoring committee for QED.

Funding

This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases grant K24DK116180.