Circulating Sphingolipids and Incident CKD

Visual Abstract

Abstract

Key Points

• Higher levels of sphingomyelin (SM) carrying fatty acid 16:0 were associated with a 69% increased risk of CKD.

• Higher levels of SM carrying fatty acid 24:0 were associated with 29% lower risk of rapid decline in kidney function and 33% higher eGFR.

• These findings support future efforts to explore pathways that explain how circulating SMs influence risk of kidney outcomes.

Background

Few studies have assessed whether ceramides (Cer) and sphingomyelin (SM) species are associated with kidney health in community-based studies. We investigated associations of eight Cer and SM species with incident CKD (eGFR <60 ml/min per 1.73 m²) and other markers of kidney health (i.e., rapid decline in kidney function, eGFR, urine albumin-to-creatinine ratio) in a large cohort of American Indians.

Methods

We included participants from the Strong Heart Family Study, a prospective cohort study of risk factors for cardiometabolic diseases. We used generalized estimating equations to examine associations of Cer-16, Cer-20, Cer-22, Cer-24, SM-16, SM-20, SM-22, and SM-24, with kidney health.

Results

In total, 95 participants had CKD at baseline, 79 participants developed CKD during a mean follow-up of 5.4 years, 2167 participants remained free of CKD, and 270 participants experienced rapid decline in eGFR of >3 ml/min per 1.73 m² per year. After multivariable adjustment, higher levels of SM-16 were associated with greater risk of CKD (relative risk [RR], 1.69; 95% confidence interval [CI], 1.24 to 2.23), while higher levels of SM-24 were associated with lower risk of rapid decline in kidney function (RR, 0.71; 95% CI, 0.58 to 0.87). Higher levels of circulating SM-24 were also associated with higher eGFR (RR, 1.33; 95% CI, 0.47 to 2.18). Cer-16, Cer-20, Cer-22, Cer-24, and SM-20 were not associated with kidney health.

Conclusions

Associations of Cer and SMs with kidney health differ based on the length of the acylated saturated fatty acid attached to the SM.

Introduction

CKD is a leading cause of mortality among American Indians (AIs).¹ The high burden of CKD among AIs is attributable, at least in part, to the high prevalence of diabetes in the population. AIs are 2.5 times as likely to have diagnosed diabetes and twice as likely have ESKD than non-Hispanic White participants of similar age.^2,3 Identifying potentially modifiable biologic mechanisms that influence development of CKD in AIs is essential to improve the health of this population.

Sphingolipids are a class of lipids with an acylated fatty acid attached to sphingoïd base (sphingosine) backbone. For most sphingolipids, the acylated fatty acid is saturated and varies in number of carbons. Sphingosine N-acylated with a fatty acid forms ceramides (Cer), and head groups attached to Cer form more complex sphingolipids; for sphingomyelins (SMs), phosphoryl choline is attached to the Cer.⁴ Previous work demonstrates the role of Cer and SMs in several biologic processes related to CKD in cell and animal studies, including inflammation, oxidative stress, apoptosis, and necrosis in podocytes and proximal tubular cells.⁵ For instance, previous work in animal and cell studies indicates that Cer-16 promotes apoptosis, while Cer-20 and Cer-22 inhibit apoptosis.^6,7 As apoptosis plays a critical role in the pathogenesis of kidney disease, this finding suggests that the length of the acylated fatty acid in the sphingolipid may be an important factor in kidney health.

Recent studies in humans suggest that distinct fatty acids impart specific biologic activity to a given lipid, and the magnitude and direction of associations of circulating Cer and SMs with health outcomes vary by the length of the acylated fatty acid in the sphingolipid. For example, in a large population-based cohort of elderly adults, there were differences between Cer and SMs carrying palmitic acid (Cer-16 and SM-16 [16 carbons and 0 double bonds]) versus longer chain fatty acids, arachidic acid (Cer-20, SM-20), behenic acid (Cer-22, SM-22), lignoceric acid (Cer-24, SM-24), on health outcomes. Higher levels of circulating Cer-16 and SM-16 were associated with increased risks of atrial fibrillation, heart failure, and mortality,^8–10 while higher levels of Cer and SMs carrying Cer-20, Cer-22, Cer-24, SM-20, SM-22, and SM-24 were associated with decreased risks.^8,9 Few studies have assessed associations of Cer and SMs with kidney outcomes,^11,12 and to our knowledge, none have focused on nonclinical population-based studies.

In this article, we assessed associations of the eight sphingolipid species (Cer-16, Cer-20, Cer-22 Cer, 24, SM-16, SM-20, SM-22, SM-24) shown to be associated with risk of cardiovascular diseases and mortality in previous studies with kidney health (i.e., incident CKD, rapid decline in kidney function, eGFR, urine albumin-to-creatinine ratio [ACR]) in a large cohort of AIs. We hypothesized that higher levels of Cer-16 and SM-16 are associated with decreased kidney health, while higher levels of Cer-20, Cer-22, Cer-24, SM-20, SM-22, and SM-24 are associated with increased kidney health. In exploratory analyses, we assessed associations of seven other Cer and sphingolipid species with kidney health.

Methods

Setting and Study Population

The Strong Heart Family Study (SHFS) is a family-based longitudinal study of risk factors for CVD in 12 AI communities in Arizona, North Dakota, South Dakota, and Oklahoma. Participants completed two examinations over 8 years, a baseline examination in 2001–2003 and a follow-up examination in 2006–2009. Previous publications have described the study design in detail.¹³ In total, 2780 AIs from 92 large families completed the baseline examination, and 91% of the participants also completed follow-up. The Institutional Review Board from each Indian Health Service region and all 12 communities approved the study, and written informed consent was obtained from all participants. All procedures were in accordance with the Declaration of Helsinki of 1975.

In total, 2713 participants had available plasma sphingolipid measures from baseline. For this analysis, we excluded participants who did not participate in the follow-up examination (n=347). Participants who did not participate in follow-up were slightly older (mean age 46 [20] versus 40 [16], more likely to be male [50% versus 39%], and have prevalent CVD [11% versus 5%] and diabetes [23% versus 18%]) when compared with participants who completed follow-up (Supplemental Table 1). For analyses that examined the association of plasma Cer and SM species with incident CKD and with change in eGFR, we excluded participants who were missing eGFR at baseline (n=28) or follow-up (n=43), and in analyses of incident CKD only, those with an eGFR <60 ml/min per 1.73 m² at baseline (n=49). For analyses with the outcome change in ACR, we excluded participants missing measures of urinary albumin or creatinine at baseline (n=42) or follow-up (n=87). Sample sizes were n=2246 for the incident CKD analyses, n=2295 for the change in eGFR analyses, and n=2237 for the change ACR analyses.

Data Collection

Each examination included a standardized interview, physical examination, and blood draw.^13,14 Information on demographic and other health factors were collected at the personal interviews. Measures of adiposity were collected during the physical examination; participants were instructed to wear lightweight clothing and no shoes for these assessments. Body mass index (BMI) was calculated as body weight divided by height-squared (kg/m²). Waist circumference was measured at the umbilicus, while the participant was in a supine position. ACCUSPLIT AE120 pedometers (Livermore, CA) were used to estimate ambulatory activity over a 1-week period.¹⁵ Blood samples were collected after a 12-hour overnight fast and were stored at −80°C. Plasma glucose was measured using enzymatic methods, and serum creatinine was measured using the picric acid method.^13,14 Urine samples were collected at the beginning of each examination, processed, and stored at −80°C. Albumin and creatinine were measured using nephelometric immunochemistry and alkaline picrate methods, respectively.^13,14

Measurement of Sphingolipids

Plasma SMs and Cer containing distinct saturated fatty acids acylated to the sphingoid backbone were measured using stored plasma samples by liquid chromatography tandem mass spectroscopy. Details of the laboratory methods and quality control procedures have been reported in detail previously.¹⁶ In total, 22 sphingolipid species were measured. For the current investigation, we restricted analyses to the 15 sphingolipid species with coefficient of variation ≤20%. This included eight Cer and SM species of primary (hypothesis-driven) interest: Cer-16, Cer-20, Cer-22, Cer-24, SM-16, SM-20, SM-22, and SM-24. For these analyses, Cer-24 comprised a composite concentration computed as the sum of the concentrations of two species of Cer with 24:0 having the distinct “d18:1” and “d18:2” sphingoid backbones. Other plasma sphingolipids assessed in secondary (nonhypothesis driven) analyses included one ceramide (Cer-18), two SMs (SM-14, and SM-18), three hexosyl-Cer (HexCer; HexCer-16, HexCer-22, and HexCer-24), and one lactosyl-Cer (lactosyl-ceramide-16). As described previously, Cer and SM species of interest were moderately to strongly correlated in the SHFS¹⁷ (Supplemental Table 2).

Ascertainment of Kidney Outcomes

Kidney function was assessed at the baseline and follow-up study examinations. eGFR was calculated using the 2021 CKD Epidemiology Collaboration creatinine equation. Incident CKD was defined as eGFR <60 ml/min per 1.73 m² at follow-up. Rapid decline in kidney function was defined as decline in eGFR of >3 ml/min per 1.73 m² per year (yes/no). We also assessed eGFR continuously to better detect small, but potentially important differences over time.^18–30 As recent studies indicate that ACR is strongly associated with CKD progression,^31–33 we examined change in urine ACR in secondary analyses; this analysis included all participants with available ACR to examine change over the full spectrum of urine albumin excretion.^31,34,35

Statistical Analyses

Generalized estimating equations with an independence working correlation and robust standard errors were used to examine associations of each circulating Cer and SM species with incident CKD and rapid decline in kidney function. For the continuous outcomes (i.e., change in eGFR, ACR), we used linear models with the identity link. A Bonferroni correction was used to address multiple comparisons. The significance level was set as two-tailed α=0.002 (0.05 divided by 24 based on three primary kidney outcomes and eight primary sphingolipid species). To reduce skewness and influence of potential outliers on risk estimates, Cer and SM concentrations were log-transformed for all analyses; results were scaled to one SD of log-transformed concentrations. ACR was also log-transformed due to skew, and the results are presented as geometric mean ratios.

Three models were fit to examine associations of Cer and SM species with each outcome of interest. The first model (minimally adjusted model) included age (linear), sex (male/female), and field site (Arizona/Oklahoma/Dakotas). The second model (multivariable-adjusted model) additionally adjusted for education (linear), smoking (never/former/current), ambulatory activity (linear), BMI (linear), waist circumference (linear), prevalent diabetes (yes/no), duration of diabetes (continuous), prevalent CVD (yes/no), use of hypertension agents (yes/no), and use of insulin or other diabetes medications (yes/no). Owing to correlation across plasma sphingolipid species, a third model (fully adjusted model/primary model) additionally adjusted for one of the other sphingolipid species as done in previous studies^{10,17,36–38}: analyses of Cer-20, Cer-22, or Cer-24 included Cer-16; analyses of SM-20, SM-22, or SM-24 included SM-16; analyses of Cer-16 included Cer-22; analyses of SM-16 included SM-22.

It is possible that prevalent diabetes and CVD may be in the causal pathway of the Cer and SM species with risk of kidney damage (e.g., circulating Cer or SM species may influence risk of diabetes which then influence risk of developing CKD), and the models that adjust for prevalent diabetes and CVD may be overadjusted. As such, we repeated all analyses of the Cer and SM species with incident CKD, rapid decline in kidney function, or eGFR without adjustment for diabetes or CVD (but including all other model 3 covariates) in exploratory analyses. In addition, in exploratory analyses, we examined whether associations of each primary sphingolipid with incident CKD, rapid decline in kidney function, or eGFR differed by age, sex, BMI, or diabetes status. A multiplicative term was created individually for sex, age, BMI, and diabetes status, with each sphingolipid, and included in separate fully adjusted models (model 3). Likelihood ratio tests were used to evaluate the statistical significance of each interaction term. In addition, we repeated all analyses stratified by prevalent albuminuria (urine ACR ≥30 mg/g at baseline) to better understand if observed associations differed among participants with or without albuminuria at baseline. We also ran exploratory analyses that assessed associations of the Cer and SMs of interest with: (1) incident albuminuria (urine ACR ≥30 mg/g at follow-up); (2) incident CKD restricted to participants with mild loss of kidney function (i.e., eGFR <90 ml/min per 1.73 m² at baseline) to better understand if the magnitude of associations differed by underlying kidney health; and (3) a composite incident CKD outcome in which participants were classified as developing CKD if either eGFR <60 ml/min per 1.73 m² or ACR ≥30 mg/g at follow-up.

Multiple imputations were used (20 replicates) to address occasional missing values for smoking (n=3), duration of diabetes (n=32), and physical activity (n=203) using information on age, sex, site, BMI, and waist circumference. Imputations were executed in the multiple imputation by chained equations package in R using the Fully Conditional Expectation method with predictive mean matching methods, as described previously.^16,39,40 Models were clustered by family to account for the familial nature of the SHFS. The imputed datasets were uploaded into STATA version 16.0 (Stata Corp, College Station, TX) for analyses.

Results

In total, 95 participants had CKD at baseline, 79 participants developed CKD during follow-up, and 2167 participants remained free of CKD throughout the study (Figure 1). Baseline characteristics of study participants according to select kidney phenotypes are presented in Table 1. Men comprised about 39% of each analytic cohort, and the average age at baseline was approximately 40 years (range, 14–89 years). Cardiometabolic risk factors, including diabetes and albuminuria, were common.

Figure 1.

Flow diagram of study participant. ACR, albumin-to-creatinine ratio.

Table 1.

Baseline characteristics of Strong Heart Family Study participants according to kidney phenotypes for each analysis

Characteristic	Included in Incident CKD Analysis	Included in Rapid Decline in Renal Function and eGFR (Continuous) Analyses	Included in UACR Analysis
n	2246	2295	2237
Age, yr	39.4 (16.2)	39.9 (16.6)	39.7 (16.4)
Male, %	39.4	39.1	39.4
BMI, kg/m2	31.3 (7.6)	31.4 (7.6)	31.3 (7.6)
Waist circumference, cm	102.1 (18.3)	102.2 (18.3)	102.1(18.3)
Education, yr	12.2 (2.3)	12.2 (2.3)	12.3 (2.3)
Smoking, %
Never	40.1	40.5	40.5
Former	22.8	22.9	22.6
Current	37.0	36.5	36.9
Ambulatory activity, steps/d	6032 (3956)	5959 (3968)	5993 (3960)
Prevalent CVD, %	3.8	4.6	4.2
Prevalent diabetes, %	17.3	18.1	17.7
Duration of diabetes, yr	7.4 (8.4)	7.9 (8.6)	7.7 (8.6)
Insulin or hypoglycemic use, %	12.4	13 0.2	12.7
Antihypertensive medication use, %	11.1	12.2	11.8
Prevalent albuminuria, %	13.5	14.4	13.9
eGFR, ml/min per 1.73 m2	109.4 (17.4)	108.0 (20.0)	108.5 (18.9)

BMI, body mass index; CVD, cardiovascular disease; UACR, urine albumin-to-creatine ratio.

Incident CKD

During a mean (SD) follow-up of 5.4 (1.1) years (range, 2.8–8.5 years; interquartile range, 4.6–6.2 years), 79 participants developed CKD. Participants with higher levels of circulating SM-16 had a higher risk of CKD after adjustment for age, sex, and field site, education, smoking, physical activity, BMI, waist circumference, prevalent diabetes, duration of diabetes, prevalent CVD, use of hypertension or diabetes medications, and SM-22 (relative risk [RR], 1.69; 95% confidence interval [CI], 1.24 to 2.23). On the other hand, data suggested that higher levels of circulating SM-22 and SM-24 may be associated with a lower risk of incident CKD, but these risk estimates did not reach statistical significance after correction for multiple testing. Cer-16, Cer-20, Cer-22, Cer-24, and SM-20 were not associated with risk of CKD (Table 2).

Table 2.

Risk estimates^a,b (95% confidence interval) for associations of circulating ceramide and sphingomyelin species carrying saturated fatty acids with kidney outcomes

Ceramide or Sphingomyelin Species	Model 1	P Value	Model 2	P Value	Model 3	P Value
Incident CKDa
Cer-16	1.34 (1.03 to 1.73)	0.03	1.29 (1.00 to 1.66)	0.05	1.24 (0.89 to 1.73)	0.20
Cer-20	1.06 (0.75 to 1.50)	0.73	1.07 (0.76 to 1.51)	0.69	0.82 (0.51 to 1.32)	0.41
Cer-22	1.28 (0.95 to 1.72)	0.11	1.22 (0.91 to 1.63)	0.18	1.05 (0.71 to 1.55)	0.81
Cer-24	1.32 (1.01 to 1.72)	0.04	1.26 (0.96 to 1.63)	0.09	1.11 (0.77 to 1.60)	0.58
SM-16	1.14 (0.89 to 1.47)	0.28	1.29 (1.01 to 1.65)	0.04	1.69 (1.24 to 2.32)c	0.001c
SM-20	0.94 (0.71 to 1.24)	0.65	1.08 (0.81 to 1.42)	0.61	0.82 (0.56 to 1.20)	0.30
SM-22	0.89 (0.67 to 1.18)	0.41	0.97 (0.74 to 1.27)	0.83	0.67 (0.47 to 0.95)	0.03
SM-24	0.89 (0.70 to 1.13)	0.34	0.97 (0.76 to 1.23)	0.80	0.67 (0.48 to 0.93)	0.02
Rapid decline in kidney functiona
Cer-16	1.12 (0.95 to 1.33)	0.17	1.07 (0.90 to 1.26)	0.43	1.00 (0.81 to 1.24)	0.99
Cer-20	1.00 (0.84 to 1.19)	1.00	1.00 (0.83 to 1.20)	0.96	0.90 (0.73 to 1.12)	0.36
Cer-22	1.13 (0.96 to 1.33)	0.15	1.10 (0.93 to 1.29)	0.26	1.10 (0.89 to 1.35)	0.38
Cer-24	1.08 (0.92 to 1.26)	0.35	1.04 (0.89 to 1.22)	0.62	0.99 (0.82 to 1.18)	0.88
SM-16	1.07 (0.91 to 1.26)	0.42	1.11 (0.95 to 1.30)	0.20	1.21 (1.00 to 1.46)	0.05
SM-20	0.91 (0.77 to 1.07)	0.26	1.00 (0.84 to 1.19)	0.99	0.87 (0.71 to 1.08)	0.21
SM-22	0.94 (0.80 to 1.09)	0.39	1.01 (0.86 to 1.17)	0.93	0.88 (0.74 to 1.06)	0.18
SM-24	0.85 (0.73 to 0.99)	0.04	0.91 (0.78 to 1.06)	0.23	0.71 (0.58 to 0.87)c	0.0008c
eGFRb
Cer-16	−0.26 (−0.88 to 0.36)	0.41	−0.16 (−0.75 to 0.43)	0.59	0.50 (−0.19 to 1.19)	0.15
Cer-20	−0.07 (−0.77 to 0.62)	0.84	−0.18 (−0.89 to 0.53)	0.62	−0.14 (−1.11 to 0.84)	0.78
Cer-22	−0.63 (−1.32 to 0.06)	0.07	−0.57 (−1.21 to 0.07)	0.08	−0.92 (−1.70 to −0.15)	0.02
Cer-24	−0.21 (−0.89 to 0.47)	0.54	−0.10 (−0.74 to 0.53)	0.75	0.01 (−0.79 to 0.80)	0.98
SM-16	−0.08 (−0.70 to 0.54)	0.80	−0.23 (−0.85 to 0.39)	0.46	−0.40 (−1.16 to 0.37)	0.31
SM-20	0.36 (−0.37 to 1.09)	0.33	−0.10 (−0.84 to 0.64)	0.79	0.11 (−0.81 to 1.03)	0.81
SM-22	0.33 (−0.37 to 1.03)	0.36	−0.03 (−0.71 to 0.65)	0.93	0.24 (−0.61 to 1.09)	0.58
SM-24	0.85 (0.17 to 1.52)	0.02	0.49 (−0.15 to 1.14)	0.13	1.33 (0.47 to 2.18)c	0.002c

CKD defined as eGFR <60 ml/min per 1.73 m²; rapid decline in kidney function defined as >3 ml/min per 1.73 m² decline per year in eGFR. In total, 79 of 2246 participants developed incident CKD; 270 of 2295 participants experienced rapid decline in kidney function; model 1 adjusted for age, sex, and site. Model 2 additionally adjusted for education, smoking, ambulatory activity, body mass index, waist circumference, prevalent cardiovascular disease, prevalent diabetes, duration of diabetes, use of antihypertensive medications, and use of insulin or oral diabetes medications. Model 3 additionally adjusted for one of the other sphingolipid species: exposures ceramide-20, ceramide-22, and ceramide-24 adjusted for ceramide-16; exposure ceramide-16 adjusted for ceramide-22; exposures sphingomyelin-20, sphingomyelin-22, and sphingomyelin-24 adjusted for sphingomyelin-16; exposure sphingomyelin-16 adjusted for sphingomyelin-22. Cer-16, Cer-20, Cer-22, Cer-24: ceramides with palmitic, arachidic, behenic, lignoceric acid, respectively; CVD, cardiovascular disease; SM-16, SM-20, SM-22-, SM-24: sphingomyelins with palmitic, arachidic, behenic, and lignoceric acid, respectively.

^aRelative risk (95% confidence interval) for eGFR (modeled continuously) reported as ml/min per 1.73 m² difference in eGFR per SD difference in log-transformed sphingolipid species concentration.

^bβ (95% confidence interval) for eGFR (modeled continuously) reported as ml/min per 1.73 m² difference in eGFR per SD difference in log-transformed sphingolipid species concentration.

^cStatistically significant after the Bonferroni correction for multiple comparisons.

Rapid Decline in Kidney Function

In total, 270 participants experienced rapid decline in kidney function (i.e., decline in eGFR of >3 ml/min per 1.73 m² per year) during follow-up (mean [interquartile range] follow-up: 5.4 [5.2–6.2] years). Participants with higher circulating levels of SM-24 had a 29% lower risk of experiencing rapid decline in kidney function (RR, 0.71; 95% CI, 0.58 to 0.87). Similar to findings for incident CKD, Cer-16, Cer-20, Cer-22, Cer-24, SM-20, and SM-22 were not associated with decline in kidney function (Table 2). There was also no association of SM-16 with decline in kidney function after correction for multiple testing (P = 0.05).

eGFR

The median change in eGFR between baseline and follow-up was −1.7 ml/min per 1.73 m². Consistent with the findings for incident CKD and rapid decline in kidney function, higher levels of circulating SM-24 were positively associated with eGFR (when eGFR was assessed continuously). For every one SD difference in log-transformed SM-24, eGFR was 1.33 ml/min per 1.73 m² higher (95% CI, 0.47 to 2.18). Cer-16, Cer-20, Cer-22, Cer-24, SM-16, SM-20, and SM-22 were not associated with eGFR modeled continuously (Table 2).

Secondary exposures of interest (i.e., Cer-14, Cer-18, SM-14, SM-18, GC-16, GC-22, GC-24, LC-16, and LC-24) were not associated with incident CKD, rapid decline in kidney function, or eGFR (continuous) after correction for multiple testing (Table 3). The median change in ACR between baseline and follow-up was 0.26 mg/g (median change of 4.5% in ACR between baseline and follow-up), and we did not observe associations of any measured circulating Cer or SMs with ACR (Supplemental Table 3).

Table 3.

Risk estimates^a,b (95% confidence interval) for associations of circulating ceramide and sphingomyelin species with kidney outcomes (secondary exposures of interest)

Ceramide or Sphingomyelin Species	Model 1	P Value	Model 2	P Value
Incident CKD
Cer-14	0.94 (0.70 to 1.26)	0.67	1.02 (0.72 to 1.44)	0.92
Cer-18	1.18 (0.84 to 1.66)	0.34	1.21 (0.86 to 1.70)	0.27
SM-14	1.42 (1.09 to 1.86)	0.01	1.54 (1.14 to 2.08)	0.005
SM-18	0.94 (0.70 to 1.26)	0.68	1.08 (0.82 to 1.44)	0.58
GC-16	0.99 (0.79 to 1.25)	0.96	1.19 (0.95 to 1.49)	0.13
GC-22	0.75 (0.58 to 0.96)	0.02	0.90 (0.69 to 1.17)	0.43
GC-24	0.82 (0.66 to 1.04)	0.10	0.96 (0.75 to 1.23)	0.76
LC-16	0.83 (0.64 to 1.08)	0.16	1.00 (0.78 to 1.28)	1.00
LC-24	0.90 (0.71 to 1.15)	0.40	0.89 (0.67 to 1.17)	0.39
Rapid decline in kidney function
Cer-14	1.02 (0.90 to 1.15)	0.80	1.04 (0.91 to 1.20)	0.57
Cer-18	1.05 (0.89 to 1.23)	0.56	1.04 (0.88 to 1.24)	0.64
SM-14	0.97 (0.82 to 1.15)	0.76	1.01 (0.86 to 1.20)	0.89
SM-18	0.95 (0.82 to 1.12)	0.55	1.03 (0.88 to 1.21)	0.71
GC-16	1.03 (0.90 to 1.17)	0.68	1.12 (0.98 to 1.28)	0.11
GC-22	0.94 (0.83 to 1.08)	0.39	1.07 (0.92 to 1.24)	0.38
GC-24	0.89 (0.77 to 1.02)	0.10	0.97 (0.83 to 1.13)	0.68
LC-16	1.00 (0.87 to 1.16)	0.96	1.11 (0.96 to 1.30)	0.17
LC-24	0.96 (0.81 to 1.13)	0.61	0.94 (0.79 to 1.12)	0.52
eGFR
Cer-14	0.28 (−0.26 to 0.83)	0.31	0.19 (−0.37 to 0.74)	0.51
Cer-18	0.23 (−0.44 to 0.89)	0.50	0.04 (−0.66 to 0.75)	0.90
SM-14	0.15 (−0.51 to 0.81)	0.65	−0.15 (−0.84 to 0.54)	0.67
SM-18	0.62 (−0.01 to 1.26)	0.05	0.15 (−0.53 to 0.82)	0.67
GC-16	0.43 (−0.05 to 0.92)	0.08	−0.01 (−0.56 to 0.55)	0.98
GC-22	0.33 (−0.35 to 1.01)	0.34	−0.19 (−0.89 to 0.52)	0.61
GC-24	0.76 (0.18 to 1.35)	0.01	0.40 (−0.20 to 1.01)	0.19
LC-16	0.19 (−0.36 to 0.75)	0.49	−0.21 (−0.79 to 0.37)	0.48
LC-24	0.11 (−0.40 to 0.63)	0.67	0.19 (−0.31 to 0.69)	0.46

CKD defined as eGFR <60 ml/min per 1.73 m²; rapid decline in kidney function defined as >3 ml/min per 1.73 m² decline per year in eGFR. In total, 79 of 2246 participants developed incident CKD; 270 of 2295 participants experienced rapid decline in kidney function; model 1 adjusted for age, sex, and site. Model 2 additionally adjusted for education, smoking, ambulatory activity, body mass index, waist circumference, prevalent cardiovascular disease, prevalent diabetes, duration of diabetes, use of antihypertensive medications, and use of insulin or oral diabetes medications. Cer-14, Cer-18: ceramides with myristic or stearic acid respectively; CVD, cardiovascular diseases; GC-16, GC-22, GC-24: glucosyl-ceramides with palmitic, behenic, or lignoceric acid, respectively; LC-16, LC-24: lactosyl-ceramide with palmitic or lignoceric acid, respectively; SM-14, SM-18: sphingomyelins with myristic or stearic acid, respectively.

^aRelative risk (95% confidence interval) for eGFR (modeled continuously) reported as ml/min per 1.73 m² difference in eGFR per SD difference in log-transformed sphingolipid species concentration.

^bβ (95% confidence interval) for eGFR (modeled continuously) reported as ml/min per 1.73 m² difference in eGFR per SD difference in log-transformed sphingolipid species concentration.

Sensitivity Analyses

Removing prevalent diabetes and CVD as covariates in the primary model (model 3) did not materially change reported risk estimates. Of the 79 participants who developed CKD during follow-up, 36 had albuminuria at baseline. Exploratory analyses indicated that associations of SM-16 with incident CKD were primarily driven by participants with albuminuria (RR, 2.41; 95% CI, 1.57 to 3.70); there was no association of SM-16 with incident CKD among participants without albuminuria at baseline (Supplemental Table 4). Data suggested that among participants with albuminuria at baseline, SM-16 was associated with a lower risk of rapid decline in kidney function, but these risk estimates did not reach statistical significance after correction for multiple testing. Similarly, the magnitude of the association of SM-16, SM-22, and SM-24 with incident CKD was greatest among participants with eGFR <90 ml/min per 1.73 m² at baseline (Supplemental Table 5).

There were no statistically significant interactions of the primary sphingolipid species of interest with age (modeled linearly), sex, BMI (modeled linearly), or diabetes (yes/no) when assessing risk of CKD, rapid decline in kidney function, or eGFR (smallest P for interaction = 0.05). We also did not observe associations of the sphingolipids of interest with incident albuminuria (Supplemental Table 6). Similarly, we did not observe an association of the sphingolipids with the composite outcome (i.e., eGFR <60 ml/min per 1.73 m² or ACR ≥30 mg/g at follow-up; Supplemental Table 7) after correction for multiple comparisons.

Discussion

In this large, longitudinal family-study of cardiometabolic health in AIs, higher circulating levels of SM-16 were positively associated with development of CKD. On the other hand, higher circulating levels of SM-24 were associated with a lower risk of rapid decline in kidney function and higher eGFR. Furthermore, associations of the circulating SMs with incident CKD were strongest among participants with early signs of kidney disease at baseline (eGFR <90 ml/min per 1.73 m² or urine ACR ≥30 mg/g). Other circulating SM (i.e., SM-20) and Cer (i.e., Cer-16, Cer-20, Cer-22, and Cer-24) species were not associated with incident CKD or other kidney outcomes. Overall, these findings suggest that the length of the acylated saturated fatty acid attached to circulating SMs may influence associations of SM with kidney outcomes.

A handful of previous epidemiologic studies have linked circulating sphingolipid species to kidney outcomes in clinical studies comprised of participants with diabetes (type 1 or type 2), but to the best of our knowledge, this is the first study to assess associations of Cer and SM species with kidney outcomes in population-based community-dwelling adults. Our results are consistent with findings from an untargeted lipidomic analysis focused on kidney impairment among individuals with type 1 diabetes. In that study, higher levels of SM-24 were associated with both a lower risk of worsening of albuminuria and also an aggregate end point of ≥30% decline in eGFR, incident ESKD, and all-cause mortality.⁴¹ Our results also complement findings from an analysis of a subset of participants (n approximately 500) from the Diabetes Control and Complications Trial that suggested that higher levels of Cer-20 and Cer-24 were associated with a lower risk of overt nephropathy among participants with type 1 diabetes.⁴² Our results support findings from a study of Singaporean adults with type 2 diabetes that reported higher levels of Cer-24 lowered risk of rapid decline in kidney function.¹¹ On the other hand, our findings are discordant with results from a large meta-analysis of three studies from Southeast Asia that assessed the associations of 553 metabolites with kidney outcomes. In that study, only Cer-20 (but not Cer-16 or Cer-22) was associated with eGFR slope in fully adjusted models.¹² Reasons for the conflicting findings may be due to underlying differences in kidney health among participants with and without diabetes or other demographic or social factors across study populations (e.g., differences in age, health behaviors, or morbidity).

Several biologic mechanisms implicate circulating sphingolipids in the pathogenesis of kidney disease. Most published mechanistic studies have focused on Cer, sphingosine-1 phosphate, or glycosphingolipids rather than SMs.⁴³ However, Cer can be converted to SMs through SM synthases, and it is plausible that the biologic processes that may explain observed associations of SMs with kidney health are likely similar to that of Cer and kidney health.⁴ It is well-established that circulating sphingolipids are involved in apoptosis,^6,7 atherosclerosis,^44–46 inflammation,⁴⁷ immune response,⁴⁸ mitochondrial dysfunction,^49–53 and oxidative stress.⁴⁷ Studies in animals and cell culture consistently show that Cer-16 promotes apoptosis, while Cer-20 and Cer-22 inhibit apoptosis.^6,7 In addition, accumulation of sphingolipids has been shown to impair both mitochondrial respiratory chain activity and activation of protein kinase B.⁵⁴ In diabetic mice, increasing Cer levels (and the resulting mitochondrial dysfunction) has been shown to induce podocyte damage—a key component of the glomerular filtration barrier.⁴³ Furthermore, some studies show that increasing levels of Cer result in kidney failure in rodents.^55–57

Previously published studies have reported differential associations of plasma phospholipid saturated fatty acids of different chain length with risk of diabetes, a major risk factor for CKD.^58–60 In those studies, circulating levels of 16:0 have been associated with a higher risk of diabetes, while very-long-chain saturated fatty acids 20:0, 22:0, and 24:0 have been associated with a lower risk of diabetes. Because plasma phospholipids contain sphingolipids, especially SMs, it is possible that sphingolipids may be a contributing driver of the association of plasma phospholipid saturated fatty acids with diabetes risk. To the best of our knowledge, no studies have examined the relationship of very-long-chain saturated fatty acids with kidney outcomes.

In our analyses, the magnitude of associations of SM-16 with incident CKD and rapid decline in kidney function was highest among participants with albuminuria at baseline. Similarly, the relationships of SM-16, SM-22, and SM-24 with incident CKD were strongest among participants with evidence of kidney damage at baseline (i.e., eGFR <90 ml/min per 1.73 m²). This may suggest that participants with underlying kidney injury or dysfunction at baseline may be more susceptible to negative effects of SM-16 (and protective effects of SM-22 and SM-24) on risk of CKD compared with participants with no albuminuria of with eGFR ≥90 ml/min per 1.73 m² at baseline.

These findings are supported by previous work that demonstrates that individuals with albuminuria are more likely to have impairment in fatty acid synthesis and oxidation and that CKD in participants with albuminuria may be clinically distinct from individuals without albuminuria.⁶¹

Because circulating levels of sphingolipids are potentially modifiable by diet^62,63 and/or pharmaceutical agents, the findings herein may be clinically significant. Previous work has shown that very-long-chain saturated fatty acids 24:0 can be derived from intake of nuts and peanut butter⁶⁴; as circulating fatty acids comprise sphingolipids, consumption of these foods may influence levels of sphingolipids with saturated fatty acid 24:0. In addition, known drug therapies, such as Fenretinide and Fingolimod, affect levels of total circulating sphingolipids through inhibition of Cer synthesis (e.g., dihydroceramide desaturase, an enzyme involved in the conversion of dihydroceramide to Cer,⁶⁵ and serine palmitoyl transferase, a key enzyme in sphingolipid metabolism^66–68). Although these drugs do not target specific species of Cer and SMs, the findings support future work to develop agents that target specific Cer synthases and sphingolipid species.⁶⁹

This analysis has many strengths. The SHFS is the largest study of cardiometabolic health among AIs in the United States; given the high prevalence of diabetes and cardiovascular diseases in the population, a better understanding of whether novel lipids are related to kidney health is needed. The SHFS has detailed data on many cardiometabolic risk factors, including demographic, behavioral, and clinical risk factors. This increased our ability to adjust for potential confounders in the analysis. Furthermore, measurement error and bias were minimized since the SHFS uses data collection instruments with known reliability and validity.

This study also has limitations. Although we considered several demographic and clinical factors that may affect both circulating Cer or SM levels and kidney outcomes, residual confounding due to unmeasured or poorly measured factors is possible. In addition, this was a targeted analysis of select Cer and SM species, and it is not possible to examine the associations of all sphingolipid species with kidney outcomes since these data were not collected as part of the laboratory assay. The SHFS is a cohort study, and eGFR and urine albumin-to-creatine ratio were assessed at two time points—which may not fully capture kidney health. Finally, the generalizability of findings to other populations—particularly populations with different sociocultural and socioeconomic environments, diet, metabolism, and/or genetics, is unknown.

As the global burden of CKD continues to grow, it is important to better understand the associations of sphingolipids with kidney-related outcomes to inform novel CKD prevention efforts. These findings support future efforts to explore pathways that explain how circulating SMs may influence risk of kidney outcomes in an AI population with a high burden of CKD and related complications.

Supplementary Material

kidney360-7-302-s001.pdf ^{(2.6MB, pdf)}

kidney360-7-302-s002.pdf ^{(218.2KB, pdf)}

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/KN9/B292.

Author Contributions

Conceptualization: Amanda M. Fretts, Andrew N. Hoofnagle, Irena B. King, Rozenn N. Lemaitre, David S. Siscovick, Colleen M. Sitlani, Jason G. Umans.

Data curation: Andrew N. Hoofnagle, Rozenn N. Lemaitre, Jason G. Umans.

Formal analysis: Amanda M. Fretts, Paul N. Jensen, Rozenn N. Lemaitre, Colleen M. Sitlani.

Funding acquisition: Rozenn N. Lemaitre.

Investigation: Amanda M. Fretts, Andrew N. Hoofnagle, Irena B. King, Rozenn N. Lemaitre, Benjamin Lidgard, Reya H. Mokiao, David S. Siscovick, Jason G. Umans.

Methodology: Amanda M. Fretts, Andrew N. Hoofnagle, Paul N. Jensen, Rozenn N. Lemaitre, Benjamin Lidgard, David S. Siscovick, Colleen M. Sitlani.

Project administration: Rozenn N. Lemaitre.

Resources: Jason G. Umans.

Supervision: Rozenn N. Lemaitre.

Validation: Amanda M. Fretts, Andrew N. Hoofnagle.

Visualization: Amanda M. Fretts, Benjamin Lidgard, David S. Siscovick.

Writing – original draft: Amanda M. Fretts.

Writing – review & editing: Amanda M. Fretts, Andrew N. Hoofnagle, Paul N. Jensen, Irena B. King, Rozenn N. Lemaitre, Benjamin Lidgard, Reya H. Mokiao, David S. Siscovick, Colleen M. Sitlani, Jason G. Umans.

Funding

R.N. Lemaitre: National Institute of Diabetes and Digestive and Kidney Diseases (R01DK103657). A.N. Hoofnagle: National Institute of Diabetes and Digestive and Kidney Diseases (P30 DK035816). This work was supported by National Heart, Lung, and Blood Institute (75N92019D00027, 75N92019D00028, 75N92019D00029, and 75N92019D00030).

Declarative Statements

This study includes clinical experimentation and received Institutional Review Board or Ethics Committee approval. All patients provided written informed consent. This study includes clinical experimentation and complies with the Declaration of Helsinki.

Data Availability Statements

Data belong to a third party, and authors are not authorized to share the data. Third Party: Tribes that participated in the research. Reason for Restriction: The authors are not authorized to share data described in this manuscript due to data sharing agreements between the authors and the tribes, and data presented in this manuscript will not be made publicly available. However, the SHFS welcomes collaboration, and interested investigators can submit a manuscript proposal to be reviewed and approved by the SHFS Presentations and Publications Committee. Details can be found at: https://strongheartstudy.org/Research/Papers-and-Abstracts.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/KN9/B293.

Supplemental Table 1. Baseline characteristics of participants who did not complete the follow-up examination (n=347).

Supplemental Table 2. Spearman correlation coefficients between the sphingolipid species in the SHFS.

Supplemental Table 3. Geometric mean ratios (95% CI) for associations of circulating Cer and SM species carrying saturated fatty acids with albumin-creatinine ratios (n=2295).

Supplemental Table 4. Risk estimates^a,b (95% CI) for associations of circulating Cer and SM species carrying saturated fatty acids with kidney outcomes stratified by albuminuria at baseline.

Supplemental Table 5. Relative risk (95% CI) for associations of circulating Cer and SM species carrying saturated fatty acids with incident CKD among those with eGFR <90 at baseline (n=392).

Supplemental Table 6. Relative risk (95% CI) for associations of circulating Cer and SM species carrying saturated fatty acids with incident albuminuria (n=2306).

Supplemental Table 7. Relative risk (95% CI) for associations of circulating Cer and SM species with incident CKD.