Sphingolipids and Mortality in Hemodialysis

CKD is associated with profound alterations in lipid metabolism and accelerated cardiovascular disease, with patients requiring KRT exhibiting the highest risk of cardiovascular death. However, traditional cholesterol-centric paradigms for predicting and reducing cardiovascular disease have not clearly translated to the kidney failure population. Multiple randomized trials have shown that statin therapy does not reliably reduce cardiovascular events in kidney failure, highlighting that residual risk in dialysis may be driven by nontraditional biology and requires novel treatment approaches.¹

Sphingolipids are a diverse group of bioactive lipids that have emerged as promising biomarkers and plausible mediators of cardiovascular risk across the entire spectrum of cardiovascular-kidney-metabolic syndrome. Sphingolipids, including ceramides and sphingomyelins, are composed of a sphingoid base with a varying head group linked to a fatty acid chain and play essential roles in plasma membrane structure, cell signaling, and apoptosis.² However, unlike cholesterol, sphingolipids are not appreciably absorbed from the diet but are endogenously synthesized and recycled. De novo synthesis relies on a common biosynthetic pathway to generate ceramide, which can be further converted into complex sphingolipids, such as sphingomyelins and glycosphingolipids.³ The de novo pathway not only sustains basal sphingolipid synthesis but also links nutrient availability to lipid signaling and cellular homeostasis, with direct implications for physiology and disease. Large bodies of preclinical and observational evidence link circulating ceramides to many metabolic disorders, including diabetes, insulin resistance, cardiomyopathy, atherosclerosis, and steatohepatitis. In particular, longitudinal studies implicate ceramides with long acyl chains (C14:0-C18:0) in the pathogenesis of cardiovascular disease, whereas very long–chain ceramides (C20:0-C24:0) seem to be more benign.⁴ Nevertheless, mechanistic studies in murine models lacking ceramide synthase isoforms with distinct acyl-chain length specificity have produced ambiguous results regarding which ceramide species are cardiotoxic. Deficiency of CER5, which, together with CER6, makes C14:0-C16:0 ceramides, attenuates lipotoxicity and cardiomyopathy in mice.² However, overexpression of hepatic CER2 to enhance very long–chain ceramide levels also induces mitochondrial dysfunction and apoptosis, suggesting the biologic effects of any given ceramide species depend on the ratio of long- to very long–chain ceramides and subcellular localization.^2,5

In this issue of JASN, Lidgard et al. take an important step by asking whether the association between serum sphingolipid concentrations and all-cause and cardiovascular mortality persists in patients with kidney failure on maintenance hemodialysis.⁶ In a secondary analysis of the Hemodialysis (HEMO) Study, the authors used data and baseline serum from 927 patients who were receiving hemodialysis to measure a targeted panel of 16 long-chain (C14:0-C18:0) and very long–chain (C20:0-C24:1) sphingolipids by liquid chromatography–tandem mass spectrometry. After multivariable adjustment for baseline covariates using Cox proportional hazards models, the authors report an approximate two-fold higher all-cause mortality per doubling of serum long-chain ceramides or sphingomyelin concentrations. These associations were primarily driven by the associations of sphingolipids with cardiovascular death. Conversely, an inverse association was found between saturated, very long–chain ceramide and sphingomyelin concentrations and cardiovascular death. The study raises the possibility of targeting sphingolipid biosynthesis or clearance in this vulnerable kidney failure population with a high risk of cardiovascular events and mortality.

The investigators should be commended for this important work. The major strengths include systematic analysis in a large, well-adjudicated dialysis cohort—a group that has been traditionally underrepresented in cardiovascular trials—and the use of state-of-the-art analytic strategies for sphingolipid quantification. Limitations inherent to this study are no different from those of other observational studies examining the association of lipoprotein levels with cardiovascular outcomes. Importantly, the investigators were unable to adjust for residual kidney function or concomitant statin use, two covariates inversely correlated with sphingolipid concentrations.^7,8 In addition, adjusting for highly correlated sphingolipids with opposing cardiovascular risk (e.g., long-chain species adjusted for their very long–chain counterparts) may inflate effect estimates when these lipid species are not mutually independent. The measured sphingolipids are also time-varying variables with dynamic levels. In the absence of repeated measurements in Cox models, using baseline values alone may not accurately reflect the residual risk associated with changes in time-varying lipids due to diet or clearance during high-flux hemodialysis. Although the data source is from a clinical trial, the observational nature of the secondary analysis precludes causal inference.

At the same time, the most intriguing finding from this work by Lidgard et al. is its consistency with prior literature. The roughly two-fold risk of cardiovascular death associated with long-chain ceramides and sphingomyelins in hemodialysis mirrors what has been reported in nondialysis CKD and broader cardiovascular cohorts.⁴ This raises the question whether long-chain sphingolipids are a causal mediator of cardiovascular death in kidney failure or rather a consistent manifestation of excess fatty acid flux that persists across the spectrum of CKD?⁷ Accumulating evidence in the area of metabolic syndrome supports the theory that elevated ceramides may represent, at least in part, an adaptive response that shunts excess free fatty acids into sphingolipid pools when triglyceride storage and oxidative capacity are exceeded.² Mechanistically, several sphingolipid functions support the plausibility of this model: (1) sphingomyelins complex with cholesterol in plasma membranes to facilitate fatty acid diffusion; (2) ceramides promote the insertion of fatty acid transporters at the cell surface to facilitate downstream esterification and metabolism; and (3) ceramides can reduce mitochondrial efficiency, thereby increasing reliance on fatty acid oxidation. Together, these observations suggest that higher circulating ceramides in dialysis could reflect an attempt to buffer the saturated fatty acid overload associated with progressive CKD while simultaneously marking (and potentially contributing to) the lipotoxic state.

Ceramides can also function as potent second messengers; regulating essential cellular processes, including apoptosis, differentiation, proliferation, and senescence, and may contribute to vascular inflammation in cardiovascular-kidney-metabolic syndrome (reviewed in Baek et al.).³ Thus, it is enticing to envision novel therapeutic approaches that target sphingolipid synthesis or degradation. However, lessons from prior experiences with statin therapy in CKD should urge the field to proceed with caution. Multiple studies have demonstrated that statins and proprotein convertase subtilisin/kexin type 9 inhibitors can effectively reduce circulating long-chain ceramides by 25%–40%.⁸ Furthermore, hepatic silencing of CER2, which synthesizes C24:0 and C24:1 ceramides, in mice has been shown to reduce plasma levels of these ceramide species, suggesting that circulating ceramide levels are governed by liver metabolism.⁵ Yet, the dialysis paradox remains: Statins do not reduce cardiovascular events in hemodialysis, suggesting that lowering circulating ceramides may not translate into improved cardiovascular outcomes in the dialysis population.

Although this does not invalidate sphingolipids as prognostic biomarkers, Lidgard et al.’s findings raise several unanswered questions about the cardiovascular pathogenesis of sphingolipids in CKD. Lipidomic studies in diabetic or CKD mice show opposing concentrations of ceramides and sphingomyelin species in plasma, kidney, and macrophage samples, suggesting that regulation may be tissue specific.^9,10 This raises the question: Do circulating sphingolipids accurately reflect the sphingolipid composition of cardiac or vascular tissue in CKD? In addition, what level of reduction in circulating ceramide is needed to lower cardiovascular risk? Finally, what effect does residual kidney function have on sphingolipid metabolism and clearance? Future research should focus on longitudinal sphingolipid profiling with precise measurement of residual kidney function and on investigating whether modifying ceramide synthase balance or ceramide localization affects cardiovascular phenotypes in CKD/kidney failure model systems.

In summary, the work by Lidgard et al. takes an important step forward by identifying a sphingolipid signature associated with cardiovascular death in kidney failure. Clinically, sphingolipids may be most immediately useful for risk stratification and trial enrichment, whereas therapeutic translation will require evidence that modifying specific ceramide pools can overcome the inherent cardiovascular risk associated with long-term dialysis.

Supplementary Material

jasn-37-1121-s001.pdf ^{(1.4MB, pdf)}

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/JSN/F787.

Author Contributions

Writing – original draft: Keith L. Saum.

Writing – review & editing: Subramaniam Pennathur, Keith L. Saum.

Funding

S. Pennathur: National Institute of Diabetes and Digestive and Kidney Diseases (U54DK137314 and P30DK89503) and Juvenile Diabetes Research Foundation (5-COE-2019-861-S-B). K.L. Saum: National Institute of Diabetes and Digestive and Kidney Diseases (TL1DK136046 and U2CDK129445) and American Heart Association (26CDA1601970).