Skip main navigation

Ceramide Scores Predict Cardiovascular Risk in the Community

Originally published, Thrombosis, and Vascular Biology. 2021;41:1558–1569



Cardiovascular disease remains a leading cause of mortality worldwide. Ceramide scores have been associated with adverse outcomes in patients with established coronary artery disease. The prognostic value of ceramide score has not been assessed in the general population. We tested the hypothesis that ceramide scores are associated with major adverse cardiac events (MACE) in a community-based cohort with average coronary artery disease burden at enrollment.

Approach and results:

In a prospective community-based cohort, we performed passive follow-up using a record linkage system to ascertain the composite outcome of MACE, defined as acute myocardial infarction, coronary revascularization (bypass grafting or percutaneous intervention), stroke, or death. Ceramides were analyzed as log-transformed continuous variables, ratios or scores, and quartiles with adjustment for confounders. We analyzed 1131 subjects, 52% females, mean age±(SD) 64±9 years. After a median follow-up of 13.3 years (Q1, 12.7; Q3, 14.4), 486 patients experienced a MACE: myocardial infarction (80), coronary artery bypass surgery (34), percutaneous coronary intervention (62), stroke (94), and all-cause death (362). Ceramide ratios were significantly associated with MACE independently of LDL-c (low-density lipoprotein cholesterol) and conventional coronary artery disease risk factors. Those in the highest quartile of ceramide score had nearly 1.5-fold risk of MACE, hazard ratio, 1.47 (95% CI, 1.12–1.92). There was a dose-response association across quartiles of ceramide ratios and MACE.


Elevated ceramide score is a robust predictor of cardiovascular disease and MACE in the community. The risk conferred by the ceramide score has a dose-response behavior and is independent of conventional risk factors.


  • Ceramide score:

  • Robust biomarker for atherosclerotic risk for primary prevention.

  • Reproducible and modifiable with interventions: diet, aerobic exercise, and lipid-lowering agents.

  • Assesses response to primary intervention and motivates patients.

  • No radiation, cost-effective.

  • Results are easy to interpret by providers.


Cardiovascular disease (CVD) remains a leading cause of mortality worldwide with >17.8 million deaths per year.1 Accordingly, improved prevention of cardiovascular morbidity and mortality would have a significant impact on public health. Traditional lipid biomarkers, such as HDL-c (high-density lipoprotein cholesterol), LDL-c (low-density lipoprotein cholesterol), and triglycerides, have been extensively used to evaluate the risk for atherosclerotic CVD for several decades.2–5 HDL-c and LDL-c, along with other constituents of the Framingham score, predict ≈75% of the risk6 and thus they are important decision-making instruments for prevention and treatment.7 However, the need for additional predictive biomarkers is underlined by the very substantial residual risk that exists above the standard clinical and biochemical risk predictors.8

Ceramides are main constituents of the sphingolipid cell signaling pathways, and their plasma levels have been associated with several metabolic and cardiovascular conditions.9 Clinical studies with ceramides initially focused mainly on patients with known coronary artery disease (CAD) to identify patients with residual risk for secondary prevention. For example, in a European cohort of patients with stable ischemic heart disease and acute coronary syndromes ceramide ratio (d18:1/16:0)/(d18:1/24:0) improved prediction of cardiovascular events by 8.2%%.10 The Mayo Clinic has validated this approach in an angiography cohort and thus has implemented the ceramide score in clinical practice as the first lipidomic biomarker for residual atherosclerotic risk.11 Data on primary prevention and ceramides have also been reported but not specifically in a community cohort.12,13 Moreover, to date, ceramide scores have not been explored for primary prevention. The current study assesses the ceramide score in the community in a cohort of subjects initially enrolled for assessment of the prevalence of asymptomatic left ventricular dysfunction.14 Our investigations focus on primary prevention and the role of ceramide as lipidomic biomarkers of atherosclerotic risk in a cohort recruited from the general population who, at the time of enrollment, had no clinically known atherosclerotic disease. We hypothesized that similar to prior studies with secondary prevention, ceramides would be predictive of cardiovascular adverse end points in the general population.


All data and supporting materials have been provided with the published article. The authors declare that all supporting data are available within the article (and in the Data Supplement).

We performed a passive follow-up of a prospective community-based cohort using a record linkage system to determine the composite outcome of major adverse cardiac events (MACE), defined as acute myocardial infarction, coronary revascularization (coronary artery bypass grafting or percutaneous coronary intervention), stroke, or death. Ceramides were analyzed as continuous variables, ratios, and quartiles adjusting for potential confounders.

Patients and Methods

The authors of the current article will not make their data, analytic methods, and study materials available to other researchers. Study approval was issued by the Mayo Foundation and Olmsted Medical Center review board. Informed consent was obtained from all patients. Using the Rochester Epidemiology Project (REP), individuals older than 45 years were selected for participation in the study. As outlined in previous studies, this database is centered in Olmsted County, Minnesota.15,16 The REP cohort is predominately White middle class. Further specifics of the population are described in past studies.17,18 The REP has been successful in mapping out a variety of complex disease associations in the community settings. The REP framework was used to randomly select a fraction of 7% within each sex and age stratum of county residents 45 years of age or older. Of the 4203 selected residents, 2042 (47%) unrelated men and women were enrolled. Between the years 1997 and 2000, participants were recruited and data were passively collected from the REP, as previously described. After 4 years, all participants were invited to return for a second examination, and 1402 participated between the years 2001 and 2004. For this study, we first identified the 1402 individuals who were evaluated at the second examination. We excluded subjects who experienced MACE before the second examination (N=78) and those with unavailable samples for analysis (N=193). A total of 1131 subjects were included in the final analysis (Figure 1). Each of the enrolled participants underwent a physical examination; blood work that included kidney function, lipid testing, and neurohormonal biomarkers of cardiovascular damage/distress between September 1, 2001, and December 30, 2004. As previously described, measurements of body mass index (calculated as the weight in kilograms divided by the height in meters squared), blood pressure, height, and weight, were obtained during the initial data collection period. All participants completed medical surveys and questionnaires following initial data collection. Method for ceramide testing and algorithm for ceramide score has been described previously.11,19

Figure 1.

Figure 1. Cohort selection–a flow chart of patients included in our study. MACE indicates major adverse cardiac events.

Statistical Analyses

Participants were characterized by their demographics, clinical, and laboratory data including ceramides. Ceramides along with other continuous variables including body mass index, biomarkers, and laboratory measurements were summarized by median and interquartile range. Age and blood pressure were approximately normally distributed and summarized by mean and SD. Binary and multicategory variables were described by counts and percentages.

Combined (stroke/myocardial infarction [MI] and MACE) and individual (stroke, MI, coronary artery bypass grafting, percutaneous coronary intervention, and death) end points were analyzed as time-to-event outcomes. The overall rate of events at 5 and 10 years was estimated via Kaplan-Meier method. To assess the effects of ceramides on each outcome, Cox proportional hazards models were fit after log-transforming individual ceramides and their ratios to better approximate normality. Hazard ratios (HR), both unadjusted and adjusted for important covariates, were estimated and are presented with corresponding 95% CI. Adjustment for traditional CAD risk factors was accomplished by fitting age, sex, body mass index, hypertension, CAD, glomerular filtration rate, ApoB, Apo A1, LDL-C, and total cholesterol as covariates together with each ceramide. Atherosclerotic cardiovascular disease (ASCVD)–adjusted models included a categorical risk variable with cutoffs recommended by ASCVD guidelines. Survival C statistics were calculated to assess discrimination ability and compared between models using bootstrapping with 1000 samples to estimate (with 95% CI) the improvement of each adjusted model’s C statistic over that of the baseline model. The integrated discrimination index (IDI) and relative IDI at 5 years were also calculated as a way to assess improvement in discrimination. These estimates are reported with bootstrap 95% CIs and the bootstrap samples were used to test differences in IDI between models.

Several Kaplan-Meier plots of survivorship from stroke/MI were produced. Each set of curves was accompanied by HR estimates obtained by fitting a Cox model with the relevant ceramide variable as a 4-level factor split by its quartiles as well as relevant adjustment variables. Both unadjusted and adjusted Kaplan-Meier curves are shown, the adjusted curves were generated using a method of direct adjustment.

P<0.05 were considered statistically significant. All analyses were conducted in SAS version 9.4 (Cary, NC).


Baseline Characteristics

We included 1131 subjects in our analysis. The main clinical and biochemical characteristics of the cohort are listed in Table 1. The mean age was 64 years old, and 52.2% of the subjects were female. The majority of subjects were white (97.5%) with a slightly elevated body mass index. Of the patients included, 40.8% carried a diagnosis of hypertension, 11.2% had diabetes, and 6.6% were current cigarette smokers. A family history of CAD in first-degree relatives was present in 551 (59.1%) of patients with available information (n=933). The lipid profile at the time of enrollment showed a mean total cholesterol of 199 mg/dL, with a mean HDL-c 48 mg/dL, mean triglycerides of 128 mg/dL, and a mean calculated LDL-c of 119 mg/dL. Of the subjects included, 29% were on statin therapy at the time of inclusion. The mean fasting glucose was 96 mg/dL. The median ceramide score was 3. A score of 0 to 2 conferring low cardiovascular risk prediction was found in 42.9% of subjects; a ceramide score of 3 to 6 suggesting moderate risk was found in 40.2% of cohort. Increased risk was identified in 13.8% of subjects as described by a ceramide score of 7 to 9, whereas 3% of patients were deemed at very high risk by a ceramide score of 10 to 12 (Table 1).

Table 1. Baseline Demographics

OverallWith calculated ASCVD
(N=1131)[n miss](N=852)[n miss]
ASCVD, %, n (%)[279]
 Low (<7)339 (39.8%)339 (39.8%)
 Mid (≥7–<10)104 (12.2%)104 (12.2%)
 High(≥10)409 (48.0%)409 (48.0%)
 Age, y, mean±SD64±963±8
 Women, n (%)590 (52.2%)431 (50.6%)
 Race: White, n (%)1103 (97.5%)835 (98.0%)
 BMI, kg/m2, median (IQR)28 (25–31)28 (25–31)
 Hypertension, n (%)462 (40.8%)352 (41.3%)
 Hypertension drug, n (%)476 (42.1%)367 (43.1%)
 Family history of premature coronary artery disease, N (%)551 (59.1%)[198]417 (60.3%)[160]
 Ever smoked, n (%)522 (48.3%)[50]419 (49.2%)
 Current smoker, n (%)71 (6.6%)53 (6.2%)
 Statin at enrollment, n (%)310 (29.0%)[63]238 (29.5%)[46]
 Calculated GFR (MDRD), median (IQR)75 (67–87)[11]76 (69–87)[7]
 Coronary artery disease, n (%)129 (11.4%)92 (10.8%)
 Diabetes, n (%)109 (11.2%)[155]93 (10.9%)
 Systolic BP, mm Hg, mean±SD126±19[2]124±18
 ApoB, mg/dL105 (91–120)105 (91–120)
 Apo A1, mg/dL157 (139–180)156 (139–179)
 ApoB/A1 ratio0.70 (0.50–0.80)0.70 (0.60–0.80)
 Free fatty acids0.56 (0.39–0.78)[5]0.55 (0.38–0.75)[3]
 Cholesterol (total)199 (175–222)[5]197 (174–221)
 HDL (cholesterol)48 (38–59)[5]47 (38–59)
 LDL (cholesterol)119 (99.2–141)[22]118 (98.7–140)[12]
 CRP0.10 (0.10–0.30)[5]0.14 (0.07–0.32)
 Glucose96.0 (90.0–104)[5]96.0 (90.0–104)
 Triglyceride128 (96.0–178)[5]129 (96.0–180)
Ceramides, median (IQR)
 Ceramide (16:0), μmol/L0.26 (0.23–0.31)[3]0.26 (0.22–0.30)[3]
 Ceramide (18:0), μmol/L0.09 (0.07–0.12)[16]0.09 (0.07–0.12)[14]
 Ceramide (24:1), μmol/L1.14 (0.94–1.35)[7]1.13 (0.92–1.33)[7]
 Ceramide (24:0), μmol/L3.56 (3.00–4.33)[3]3.50 (3.00–4.40)[3]
 Ceramide ratio (16:0)/(24:0)0.07 (0.06–0.09)[3]0.07 (0.06–0.09)[3]
 Ceramide ratio (18:0)/(24:0)0.03 (0.02–0.03)[16]0.03 (0.02–0.03)[14]
 Ceramide ratio (24:1)/(24:0)0.32 (0.26–0.38)[7]0.32 (0.26–0.38)[7]
 Ceramide score (0–12)3 (1–5)[3]3 (1–5)[3]
 Ceramide score category, n (%)[3][3]
 0–2484 (42.9%)386 (45.5%)
 3–6454 (40.2%)339 (39.9%)
 7–9156 (13.8%)104 (12.2%)
 10–1234 (3.0%)20 (2.4%)

ASCVD indicates atherosclerotic cardiovascular disease; BMI, body mass index; BP, blood pressure; CRP, C reactive protein; GFR, glomerular filtration rate; HDL, high-density lipoprotein; IQR, interquartile range; and LDL low-density lipoprotein.

End Points


Subjects were followed over a median of 13.3 (Q1, 12.7; Q3, 14.4) years. Overall, 486 patients developed MACE. Kaplan-Meier–estimated rates of MACE were 13.6% (95% CI, 11.6%–15.6%) and 29.2% (95% CI, 26.5%–31.8%) at 5 and 10 years, respectively. By the end of the follow-up, a personal history of CAD had developed in 11.4% of the cohort. Ceramide score had an HR of 1.16 (1.05–1.28, P=0.005) when adjusted for ASCVD, and 1.12 (1.02–1.23, P=0.018) when adjusted for traditional CAD risk factors.

MI and Stroke

Cox proportional analysis showed that ceramide score did not correlate strongly with myocardial infarction (MI) as an individual adverse end-point after adjustment for 10-year pooled cohort risk ASCVD score (HR, 1.28 [95% CI, 0.99–1.66], P=0.063). Ceramide score did however reach statistical significance as a predictor for stroke, with ASCVD-adjusted HR 1.32 (95% CI, 1.04–1.66) and P=0.021 (Table I in the Data Supplement). Association of ceramide score with the combined stroke/MI outcome was significant as well, HR similar to that of the individual stroke outcome with an ASCVD-adjusted value of 1.31 (95% CI, 1.09–1.57) and P=0.004 (Table 2). Kaplan-Meier curves of survivorship from combined stroke and MI were divided according to ceramide ratio18:0/24:0, ceramide ratio 24:1/24:0, and ceramide score quartiles. These were plotted with and without adjustment for ASCVD. Corresponding Cox analyses showed statistically significant HRs for ceramide ratio 18:0/24:0, ceramide ratio 24:1/24:0, and ceramide score upper quartiles after adjustment for guideline-endorsed ASCVD risk categories, each referencing lowest quartiles. The HR (95% CI) estimates indicated decreased survival free of stroke and MI associated with elevated ceramides: 3.00 (1.17–7.68), 2.93 (1.52–5.67), and 2.63 (1.36–5.09) for the upper quartiles of ceramide ratio 18:0/24:0, ceramide ratio 24:1/24:0, and ceramide score, respectively (Figure 2). These indications remained true whether adjustment was made for decision cut points or risk categories as recommended by the ASCVD guidelines (Figure 2). Additionally, there was a dose-response behavior of the ceramide score distribution with MACE (Figure 3).

Table 2. Combined Stroke and Myocardial Infarction Outcomes

Outcome: stroke/MI
HR (95% CI)P valueHR (95% CI)P valueC statistic (95% CI)
Baseline: ASCVD
 Low (<7.5)Ref0.63 (0.59–0.68)
 Mid (7.5–10)2.24 (1.06–4.76)0.0360.63 (0.59–0.68)
 High (≥10)3.37 (2.07–5.48)<0.0010.63 (0.59–0.68)
Cer (16:0)1.10 (0.91–1.34)0.3331.08 (0.89–1.31)0.4520.64 (0.59–0.69)
Log[Cer (18:0)]1.21 (1.00–1.47)0.0501.18 (0.97–1.42)0.0920.66 (0.61–0.71)
Log[Cer (24:0)]0.85 (0.71–1.02)0.0840.83 (0.69–1.00)0.0450.66 (0.60–0.71)
Cer (24:1)1.25 (1.06–1.48)0.0091.19 (1.00–1.41)0.0450.66 (0.61–0.71)
Log[Cer (16:0)/(24:0)]1.26 (1.05–1.52)0.0131.29 (1.07–1.55)0.0090.67 (0.62–0.72)
Log[Cer (18:0)/(24:0)]1.33 (1.10–1.61)0.0031.32 (1.09–1.59)0.0040.67 (0.62–0.72)
Log[Cer (24:1)/(24:0)]1.41 (1.17–1.69)<0.0011.37 (1.13–1.64)0.0010.68 (0.63–0.72)
Cer score1.36 (1.14–1.64)<0.0011.31 (1.09–1.57)0.0040.67 (0.63–0.72)

ASCVD indicates atherosclerotic cardiovascular disease; Cer, ceramides; HR, hazard ratio; and MI, myocardial infarction.

* Models were adjusted for ASCVD with guideline-endorsed cutoffs.

Figure 2.

Figure 2. Survival by Kaplan-Meier curves of for ceramides (Cer; 18:0)/(24:0; A), (24:1)/(24:0; B), and Cer score (C) by atherosclerotic cardiovascular disease cutoff point quartiles. HR indicates hazard ratio; and MI myocardial infarction.

Figure 3.

Figure 3. Dose-response curve of ceramide ratios and score with major adverse cardiac events (MACE). Forrest plot of ceramide ratios and score by quartiles in relationship with MACE expressed as hazard ratio.


At follow-up, there were 108 deaths among those with nonmissing risk factors and 58 for those with nonmissing ASCVD score. When analyzed for mortality, ceramide score did not show a statistically significant correlation (data not shown). Ceramide ratio16:0/24:0 was significant in the unadjusted and ASCVD-adjusted models but lost significance when adjusted for conventional risk factors.

Overall, the ceramide score predicted MACE and mortality in both the ASCVD-adjusted and unadjusted models with the adjusted HR (95% CI) of 1.16 (1.05–1.28; Table 3). When analyzed separately, ceramide score predicted stroke with HR (95% CI) of 1.32 (1.04–1.66) and MI with HR (95% CI) of 1.28 (0.99–1.66) in the ASCVD-adjusted model; however, the score failed to predict individual outcomes, such as MI, death, and need for revascularization (coronary artery bypass grafting and percutaneous coronary intervention; Table I in the Data Supplement).

Table 3. MACE and Death Adjusted for ASCVD

Outcome: MACE/death
HR (95% CI)P valueHR (95% CI)P valueC statistic (95% CI)
Baseline: ASCVD
Low (<7.5)Ref0.60 (0.57–0.62)
Mid (7.5–10)1.44 (0.95–2.17)0.0870.60 (0.57–0.62)
High (≥10)2.33 (1.84–2.96)<0.0010.60 (0.57–0.62)
Cer (16:0)1.11 (0.99–1.23)0.0691.08 (0.97–1.21)0.1450.61 (0.57–0.64)
Log[Cer (18:0)]1.06 (0.95–1.18)0.2961.04 (0.93–1.15)0.4110.60 (0.57–0.63)
Log[Cer (24:0)]0.93 (0.84–1.03)0.1830.91 (0.82–1.01)0.0840.61 (0.58–0.64)
Cer (24:1)1.12 (1.01–1.23)0.0261.08 (0.97–1.19)0.1560.61 (0.58–0.64)
Log[Cer (16:0)/(24:0)]1.18 (1.06–1.31)0.0021.19 (1.07–1.32)0.0010.62 (0.59–0.65)
Log[Cer (18:0)/(24:0)]1.12 (1.01–1.24)0.0331.12 (1.01–1.24)0.0400.61 (0.58–0.64)
Log[Cer (24:1)/(24:0)]1.21 (1.09–1.35)<0.0011.19 (1.07–1.32)0.0010.62 (0.59–0.65)
Cer Score1.20 (1.08–1.33)<0.0011.16 (1.05–1.28)0.0050.62 (0.59–0.65)

ASCVD indicates atherosclerotic cardiovascular disease; Cer, ceramides; HR, hazard ratio; and MACE, major adverse cardiac events.

* Models were adjusted for ASCVD with guideline-endorsed cutoffs.

Additionally, our analyses identified ceramide ratio 24:1/24:0 as a marker of high risk in some patients otherwise classified low risk by the ASCVD score (Figure 4). Conversely, we show that some patients with higher risk by ASCVD score have low ceramide ratio 24:1/24:0, suggesting that these patients might not be at high risk (Figure 4).

Figure 4.

Figure 4. Survival by Kaplan-Meier curves of for ceramides (Cer; 18:0)/(24:0; A), (24:1)/(24:0; B), and Cer score (C) by atherosclerotic cardiovascular disease (ASCVD) risk categories. HR indicates hazard ratio; and MI, myocardial infarction.

Performance of Ceramide Score in CV Risk Prediction

We assessed the ceramide score for CV risk prediction and found hazard ratio of 1.31 (1.09–1.57) for the stroke/MI outcome after adjustment for ASCVD. The model had a C statistic of 0.67 (0.63–0.72; Table II in the Data Supplement).

Performance of Ceramide Score in Comparison With ASCVD Score

When assessing the performance of ceramides in predicting events compared with the ASCVD score by use of the relative IDI index, all ceramide ratios and the ceramide score showed statistically significant value added over ASCVD for stroke/MI (P≤0.03, Table 5) Ceramide score also improved discrimination over ASCVD for prediction of MACE/death with a relative IDI of 17.2% (1.0%–53.5%; Table 5).

Table 4. C Statistic for Cer Ratios and Score

Outcome: MACE/death
C statistic increase from baseline model*IDIRelative IDI
with 95% CIP valuewith 95% CIP valuewith 95% CI
Cer (16:0)0.009 (−0.003 to 0.021)0.1320.001 (0.000 to 0.004)0.1065.0 (−0.3 to 25.1)
Log[Cer (18:0)]0.005 (−0.006 to 0.016)0.3740.000 (0.000 to 0.003)0.2202.4 (−0.9 to 16.1)
Log[Cer (24:0)]0.010 (−0.002 to 0.023)0.1150.002 (0.000 to 0.007)0.0519.3 (0.0 to 38.6)
Cer (24:1)0.011 (−0.003 to 0.024)0.1280.000 (0.000 to 0.003)0.2912.2 (−1.7 to 18.5)
Log[Cer (16:0)/(24:0)]0.023 (0.010 to 0.037)0.0010.004 (0.001 to 0.011)0.00124.9 (3.0 to 68.0)
Log[Cer (18:0)/(24:0)]0.015 (0.001 to 0.029)0.0410.002 (0.000 to 0.007)0.0619.8 (−0.1 to 42.4)
Log[Cer (24:1)/(24:0)]0.021 (0.008 to 0.034)0.0020.004 (0.000 to 0.011)0.00422.4 (1.7 to 66.8)
Cer score0.022 (0.006 to 0.037)0.0050.003 (0.000 to 0.009)0.00617.2 (1.0 to 53.5)

ASCVD indicates atherosclerotic cardiovascular disease; Cer, ceramides; IDI, integrated discrimination improvement; and MACE, major adverse cardiac events.

* Baseline model C statistic=0.60 (0.57 to 0.62); Adjustment factor: ASCVD with guideline-endorsed cutoffs.

† 5 y follow-up.

Table 5. Relative IDI Index for ASCVD and Cer Score

Outcome: Stroke/MI
C statistic increase from baseline model*IDIRelative IDI
With 95% CIP valueWith 95% CIP valueWith 95% CI
Cer (16:0)0.010(−0.010 to 0.030)0.3200.001 (0.000 to 0.005)0.1097.2 (−0.5 to 36.8)
Log[Cer (18:0)]0.026 (−0.003 to 0.056)0.0760.001 (−0.001 to 0.007)0.2617.3 (−4.9 to 58.3)
Log[Cer (24:0)]0.026 (0.001 to 0.050)0.0430.004 (0.000 to 0.014)0.04426.1 (0.0 to 105)
Cer (24:1)0.028 (0.003 to 0.053)0.0280.002 (0.000 to 0.009)0.18911.0 (−2.9 to 69.7)
Log[Cer (16:0)/(24:0)]0.040 (0.014 to 0.067)0.0030.005 (0.000 to 0.015)0.00736.7 (1.5 to 119)
Log[Cer (18:0)/(24:0)]0.042 (0.014 to 0.014)0.0040.004 (0.000 to 0.016)0.03431.5 (0.0 to 127)
Log[Cer (24:1)/(24:0)]0.047 (0.018 to 0.075)0.0010.006 (0.001 to 0.018)0.00546.3 (2.4 to 146)
Cer score0.045 (0.016 to 0.073)0.0020.006 (0.001 to 0.016)0.00441.0 (2.4 to 127)

ASCVD indicates atherosclerotic cardiovascular disease; Cer, ceramides; IDI, integrated discrimination improvement; and MI, myocardial infarction.

* Baseline model C statistic=0.63 (0.59 to 0.68); adjustment factor: ASCVD with guideline-endorsed cutoffs.

† 5 y follow-up.


Our data provide several novel insights. This study assesses the role of ceramide score for primary prevention of atherosclerotic CVD in a community cohort. We show that ceramide score is a robust approach to defining residual risk that can be used by providers in the general population without known CAD. These data augment the primary prevention trials previously reported in research populations12,13 and suggest a role for ceramide scores in the evaluation of the large number of patients in our communities.

Ceramides were first introduced clinically as biomarkers for atherosclerosis in the LURIC trial (Ludwigshafen risk and cardiovascular health).20 In that study, ceramides 16:0, 18:0, and 24:1 were associated with CV risk independently of traditional lipid biomarkers HDL-c and LDL-c. Subsequently, ceramides 16:0, 18:0, and 24:1 were found to be associated with negative cardiovascular outcomes in the FINRISK 2002, Corogene, and PREDIMED trials (Prevencion con Dieta Mediterranea).10,13,21

The mechanisms involving ceramide association with CVD are uncertain, but likely include their interdigitation with the lipid, atherosclerotic, and inflammatory pathways.22 Structurally ceramides are important components of membrane stability; they also act as second messengers in membrane signaling pathways involved in inflammation.23 It is clear that ceramides accumulate in atherosclerotic plaques and stimulate the atherosclerosis process.24 In animal models of atherosclerosis, inhibition of ceramide synthesis is diminished in areas of atherosclerotic plaque25 leading to speculation that enzymes in the ceramide synthesis pathways might serve as suitable targets for drug development.26

Over time, it became clear that individual ceramides, ceramide ratios, and scores bear important prognostic value in patients with known CAD.6,10,13,21,23 Distinct ceramides target different components of the lipidomic and inflammatory cascade and combining them into a score optimizes their prognostic potential and facilitates provider use and simplicity of interpretation. This rationale eventually led to the use of ceramide score to identify and risk-stratify patients with established, residual, or suspected ischemic heart disease who might require more aggressive medical therapy. The ceramide score has worked extremely well in the secondary prevention setting.11 The present investigation deployed the clinically established ceramide score in the general population in the community. We now show that ceramide score as previously described, and recently implemented into clinical practice at our institution,11 is a reliable instrument that can be used for risk stratification of patients in the community for primary prevention purposes. The ceramide scores were nearly statistically significantly correlated with MI and were significantly associated with stroke in the prevalence of asymptomatic left ventricular dysfunction cohort even after adjustment for the ASCVD score which is the current US gold standard, guideline-endorsed instrument for assessing cardiovascular atherosclerotic risk in the general population.27 The lack of statistical significance for MI may reflect a lack of power for that outcome. However, it may also reflect the associations of the score with intermediate outcomes such as coronary artery bypass grafting and percutaneous coronary intervention as might be expected in a population closely surveilled by a facility with an aggressive outreach program, such as the Mayo Clinic. The frequency of stroke and MI events may well have been even greater in the absence of such surveillance. We show that the ceramide score identifies subjects at cardiovascular risk, but does not define the specific event that will occur. It thus appears that the ceramide score is more of a marker of ASCVD than of a specific outcome as is the case with other lipidomic biomarkers. This finding is consistent with reports that the target of ASCVD cannot be predicted in a prehoc manner.7 An important finding of our study is a statistically significant ceramide score hazard ratio of 1.31 (1.09–1.57) for the stroke/MI outcome after controlling for ASCVD. Our model has a C statistic of 0.67 (0.63–0.72) which is acceptable for a multifactorial condition such as CAD (Table II in the Data Supplement). Additionally, this model showed that ceramide score and ceramide ratios have statistically significant value when added over ASCVD for stroke/MI (P≤0.03, Table 5). Ceramide score improved discrimination over ASCVD for prediction of MACE/death with a relative IDI of 17.2% (1.0%–53.5%; Table 5). There is some criticism regarding the accuracy of ASCVD in predicting cardiovascular events.8,28,29 We show an increased risk of stroke and MI of subjects with elevated ceramide score in a dose-dependent manner, which maintained statistical significance even in the ASCVD-adjusted model with HR 2.63 (95% CI, 1.36–5.09) for the fourth quartile (Figure 2C). These results highlight the prognostic information provided by a simple, cost-effective serum biomarker in the general population for diagnosis, assessment, and risk stratification of atherosclerotic CVD. Depending on the degree of risk and treatment targets, these data may also have important therapeutic implications for those patients who are at higher risk.

Another interesting finding of our analyses is the identification of ceramide 24:1/24:0 as a marker of high risk in patients otherwise deemed low risk by the ASCVD score (Figure 4). We further analyzed this category of patients and found no significant differences in regards to traditional risk factors for CAD except for age (data not shown). This finding might have important implications in clinical practice. It may be that these patients with higher ceramide ratio 24:1/24:0 and low ASCVD score are often missed and might benefit from more individualized, in-depth testing and more aggressive treatment for primary prevention. Conversely, our analyses show that there is a particular category of patients with higher risk by ASCVD score and low ceramide ratio 24:1/24:0, suggesting that these patients might do as well with less aggressive medical intervention. This category of patients had no particular risk factor identified when compared to the rest of the cohort. This is in agreement with some suggestions that the ASCVD score might overestimate risk,29 questioning the accuracy of this scoring system, further underlying the importance of more precise and robust tools to assess CVD for primary prevention.

At the Mayo Clinic, the atherosclerotic ceramide score has been implemented for clinical use11 to risk-stratify patients with known or suspected residual CAD and identify those at higher risk for more aggressive management. Based on the results of the current investigation, we can extend the use of ceramide score for assessment of atherosclerotic CVD risk in the general population for primary prevention purposes. Previously 2 studies identified distinct ceramides as negative predictors of cardiovascular morbidity and mortality in the general population12,13; the specific ratio of ceramide 18:1/18:0 was associated with incident MACE with a hazard ratio of 1.21, whereas C24:0/C16:0 was associated with CAD, heart failure, and all-cause mortality at 6-year follow-up. Our study demonstrates that ceramide score is a more robust, accurate, and cost-effective instrument that can be used for risk stratification and primary prevention. This is particularly important for asymptomatic patients with no known personal history of CAD or risk factors, who are at intermediate risk due to the presence of other comorbidities, family history, or the presence of other risk enhancers. Currently, these patients are recommended additional risk stratification such as computer tomography coronary calcium scoring.7 Computer tomography coronary calcium scoring has been extensively used for asymptomatic patients at intermediate risk; however, the Agatston method which is widely used, has some limitations, such as reproducibility, the presence of distinct calcium densities in particular patient populations, and increase calcium density with statin treatment.30–32 Several studies are supportive of the fact that ceramide scores are reproducible11 and modifiable with diet,21 aerobic exercise,33 and statin therapy,34,35 conferring this lipidomic biomarker an attractive role not only in risk stratification but also in assessing the response to lifestyle changes or pharmacological interventions. Recently, Hilvo et al36 have validated a new model of prediction involving ceramide score and 2 additional phospholipids with improved prediction accuracy. It would be useful to further assess such a combination approach of ceramide score and phospholipids in our community cohort as well, which will likely improve further the prediction accuracy.

This study was performed using a predominately rural white population located in Olmsted County, Minnesota, in the Midwest. Given the homogeneity and predominance of White population, it may be difficult to use this study to completely generalize the results to the entire US population. However, this study assesses the use of ceramide score for primary prevention and to identify ceramide score as a robust serum biomarker that predicts cardiovascular morbidity in the community. Although ceramides have been studied for primary prevention previously,12,13 none of these investigations assessed the ceramide score which is currently used clinically for secondary prevention11 or compared these biomarkers with guideline-endorsed estimators of risk. In our cohort, the ceramide score did not correlate with mortality likely due to the low number of events; additionally, these patients were followed closely and risk factors were addressed aggressively. The C statistic found when assessing the association between the ASCVD score and ceramide score is reasonable for a multifactorial condition, such as CAD; additionally, there are studies suggesting the overestimation of CVD risk by ASCVD which may also play a role. Prior studies that led to the development of the ASCVD score were retrospective and were performed in subjects who were assessed and treated several decades ago per guidelines which are currently obsolete,37–39 whereas our subjects included in the current investigation were enrolled prospectively, followed, and treated in conformity with more contemporary guidelines.16


This current investigation examines the role of ceramide scores in a cohort of subjects from the community with average burden of CAD. This investigation identifies the ceramide risk score as a robust lipidomic biomarker that can be used for primary prevention and can be applied particularly in patients at intermediate risk. This has important consequences for risk stratification and therapeutic intensity options, as well as a reasonable tool to assess response to intervention. Specific subgroups of ceramides may identify patients at higher risk who may be overlooked by the 10-year pooled cohort score and conversely, a particular category of high-risk patients defined by the ASCVD score may be treated too aggressively; however, further studies are warranted.

Nonstandard Abbreviations and Acronyms


atherosclerotic cardiovascular disease


coronary artery disease


cardiovascular disease


high-density lipoprotein cholesterol


hazard ratio


integrated discrimination index


low-density lipoprotein cholesterol


Ludwigshafen Risk and Cardiovascular Health


major adverse cardiac events


myocardial infarction


Prevencion con Dieta Mediterranea


Rochester Epidemiological Project

Disclosures A.S. Jaffe has presently or in the past has consulted for most of the major diagnostic companies but not the company who makes the assays used in this analysis. The other authors report no conflicts.


The Data Supplement is available with this article at

For Sources of Funding and Disclosures, see page 1568.

Correspondence to: Allan S. Jaffe, MD, Cardiovascular Division, Gonda 5, Mayo Clinic, 200 First St SW, Rochester, MN. Email


  • 1. Collaborators GBDCoD. Global, regional, and national age-sex-specific mortality for 282 causes of death in 195 countries and territories, 1980-2017: a systematic analysis for the Global Burden of Disease Study 2017.Lancet. 2018; 392:1736–1788.CrossrefMedlineGoogle Scholar
  • 2. Dawber TR, Kannel WB, Mcnamara PM. The prediction of coronary heart disease.Trans Assoc Life Insur Med Dir Am. 1964; 47:70–105.MedlineGoogle Scholar
  • 3. Kannel WB, Dawber TR, Friedman GD, Glennon WE, Mcnamara PM. Risk factors in coronary heart disease. An evaluation of several serum lipids as predictors of coronary heart disease; the Framingham study.Ann Intern Med. 1964; 61:888–899. doi: 10.7326/0003-4819-61-5-888CrossrefMedlineGoogle Scholar
  • 4. Lewington S, Whitlock G, Clarke R, Sherliker P, Emberson J, Halsey J, Qizilbash N, Peto R, Collins R; Prospective Studies Collaboration. Blood cholesterol and vascular mortality by age, sex, and blood pressure: a meta-analysis of individual data from 61 prospective studies with 55,000 vascular deaths.Lancet. 2007; 370:1829–1839. doi: 10.1016/S0140-6736(07)61778-4CrossrefMedlineGoogle Scholar
  • 5. Di Angelantonio E, Sarwar N, Perry P, Kaptoge S, Ray KK, Thompson A, Wood AM, Lewington S, Sattar N, Packard CJ, et al; Emerging Risk Factors Collaboration. Major lipids, apolipoproteins, and risk of vascular disease.JAMA. 2009; 302:1993–2000. doi: 10.1001/jama.2009.1619CrossrefMedlineGoogle Scholar
  • 6. Wilson PW, D’Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB. Prediction of coronary heart disease using risk factor categories.Circulation. 1998; 97:1837–1847. doi: 10.1161/01.cir.97.18.1837LinkGoogle Scholar
  • 7. Arnett DK, Blumenthal RS, Albert MA, Buroker AB, Goldberger ZD, Hahn EJ, Himmelfarb CD, Khera A, Lloyd-Jones D, McEvoy JW, et al. 2019 ACC/AHA guideline on the primary prevention of cardiovascular disease: executive summary: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines.J Am Coll Cardiol. 2019; 74:1376–1414. doi: 10.1016/j.jacc.2019.03.009CrossrefMedlineGoogle Scholar
  • 8. DeFilippis AP, Young R, Blaha MJ. Calibration and discrimination among multiple cardiovascular risk scores in a modern multiethnic cohort.Ann Intern Med. 2015; 163:68–69. doi: 10.7326/L15-5105-2CrossrefMedlineGoogle Scholar
  • 9. Borodzicz S, Czarzasta K, Kuch M, Cudnoch-Jedrzejewska A. Sphingolipids in cardiovascular diseases and metabolic disorders.Lipids Health Dis. 2015; 14:55. doi: 10.1186/s12944-015-0053-yCrossrefMedlineGoogle Scholar
  • 10. Laaksonen R, Ekroos K, Sysi-Aho M, Hilvo M, Vihervaara T, Kauhanen D, Suoniemi M, Hurme R, März W, Scharnagl H, et al. Plasma ceramides predict cardiovascular death in patients with stable coronary artery disease and acute coronary syndromes beyond LDL-cholesterol.Eur Heart J. 2016; 37:1967–1976. doi: 10.1093/eurheartj/ehw148CrossrefMedlineGoogle Scholar
  • 11. Meeusen JW, Donato LJ, Bryant SC, Baudhuin LM, Berger PB, Jaffe AS. Plasma ceramides.Arterioscler Thromb Vasc Biol. 2018; 38:1933–1939. doi: 10.1161/ATVBAHA.118.311199LinkGoogle Scholar
  • 12. Peterson LR, Xanthakis V, Duncan MS, Gross S, Friedrich N, Volzke H, Felix SB, Jiang H, Sidhu R, Nauck M, et al. Ceramide remodeling and risk of cardiovascular events and mortality.J Am Heart Assoc. 2018; 7:e007931.LinkGoogle Scholar
  • 13. Havulinna AS, Sysi-Aho M, Hilvo M, Kauhanen D, Hurme R, Ekroos K, Salomaa V, Laaksonen R. Circulating ceramides predict cardiovascular outcomes in the population-based FINRISK 2002 Cohort.Arterioscler Thromb Vasc Biol. 2016; 36:2424–2430. doi: 10.1161/ATVBAHA.116.307497LinkGoogle Scholar
  • 14. Ammar KA, Jacobsen SJ, Mahoney DW, Kors JA, Redfield MM, Burnett JC, Rodeheffer RJ. Prevalence and prognostic significance of heart failure stages: application of the American College of Cardiology/American Heart Association heart failure staging criteria in the community.Circulation. 2007; 115:1563–1570. doi: 10.1161/CIRCULATIONAHA.106.666818LinkGoogle Scholar
  • 15. St Sauver JL, Grossardt BR, Yawn BP, Melton LJ, Pankratz JJ, Brue SM, Rocca WA. Data resource profile: the Rochester Epidemiology Project (REP) medical records-linkage system.Int J Epidemiol. 2012; 41:1614–1624. doi: 10.1093/ije/dys195CrossrefMedlineGoogle Scholar
  • 16. Rocca WA, Yawn BP, St Sauver JL, Grossardt BR, Melton LJHistory of the Rochester Epidemiology Project: half a century of medical records linkage in a US population.Mayo Clin Proc. 2012; 87:1202–1213. doi: 10.1016/j.mayocp.2012.08.012CrossrefMedlineGoogle Scholar
  • 17. Redfield MM, Rodeheffer RJ, Jacobsen SJ, Mahoney DW, Bailey KR, Burnett JCPlasma brain natriuretic peptide concentration: impact of age and gender.J Am Coll Cardiol. 2002; 40:976–982. doi: 10.1016/s0735-1097(02)02059-4CrossrefMedlineGoogle Scholar
  • 18. Melton LJHistory of the Rochester Epidemiology Project.Mayo Clin Proc. 1996; 71:266–274. doi: 10.4065/71.3.266CrossrefMedlineGoogle Scholar
  • 19. Kauhanen D, Sysi-Aho M, Koistinen KM, Laaksonen R, Sinisalo J, Ekroos K. Development and validation of a high-throughput LC-MS/MS assay for routine measurement of molecular ceramides.Anal Bioanal Chem. 2016; 408:3475–3483. doi: 10.1007/s00216-016-9425-zCrossrefMedlineGoogle Scholar
  • 20. Sigruener A, Kleber ME, Heimerl S, Liebisch G, Schmitz G, Maerz W. Glycerophospholipid and sphingolipid species and mortality: the Ludwigshafen Risk and Cardiovascular Health (LURIC) study.PLoS One. 2014; 9:e85724. doi: 10.1371/journal.pone.0085724CrossrefMedlineGoogle Scholar
  • 21. Wang DD, Toledo E, Hruby A, Rosner BA, Willett WC, Sun Q, Razquin C, Zheng Y, Ruiz-Canela M, Guasch-Ferré M, et al. Plasma ceramides, mediterranean diet, and incident cardiovascular disease in the PREDIMED Trial (Prevención con Dieta Mediterránea).Circulation. 2017; 135:2028–2040. doi: 10.1161/CIRCULATIONAHA.116.024261LinkGoogle Scholar
  • 22. Meeusen JW, Donato LJ, Kopecky SL, Vasile VC, Jaffe AS, Laaksonen R. Ceramides improve atherosclerotic cardiovascular disease risk assessment beyond standard risk factors.Clin Chim Acta. 2020; 511:138–142. doi: 10.1016/j.cca.2020.10.005CrossrefMedlineGoogle Scholar
  • 23. Kurz J, Parnham MJ, Geisslinger G, Schiffmann S. Ceramides as novel disease biomarkers.Trends Mol Med. 2019; 25:20–32. doi: 10.1016/j.molmed.2018.10.009CrossrefMedlineGoogle Scholar
  • 24. Chatterjee SB, Dey S, Shi WY, Thomas K, Hutchins GM. Accumulation of glycosphingolipids in human atherosclerotic plaque and unaffected aorta tissues.Glycobiology. 1997; 7:57–65. doi: 10.1093/glycob/7.1.57CrossrefMedlineGoogle Scholar
  • 25. Park TS, Panek RL, Mueller SB, Hanselman JC, Rosebury WS, Robertson AW, Kindt EK, Homan R, Karathanasis SK, Rekhter MD. Inhibition of sphingomyelin synthesis reduces atherogenesis in apolipoprotein E-knockout mice.Circulation. 2004; 110:3465–3471. doi: 10.1161/01.CIR.0000148370.60535.22LinkGoogle Scholar
  • 26. Hojjati MR, Li Z, Zhou H, Tang S, Huan C, Ooi E, Lu S, Jiang XC. Effect of myriocin on plasma sphingolipid metabolism and atherosclerosis in apoE-deficient mice.J Biol Chem. 2005; 280:10284–10289. doi: 10.1074/jbc.M412348200CrossrefMedlineGoogle Scholar
  • 27. Grundy SM, Stone NJ, Bailey AL, Beam C, Birtcher KK, Blumenthal RS, Braun LT, de Ferranti S, Faiella-Tommasino J, Forman DE, et al. 2018 AHA/ACC/AACVPR/AAPA/ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNA guideline on the management of blood cholesterol: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines.Circulation. 2019; 139:e1082–e1143. doi: 10.1161/CIR.0000000000000625LinkGoogle Scholar
  • 28. DeFilippis AP, Blaha MJ. Predicted vs observed clinical event risk for cardiovascular disease.JAMA. 2015; 314:2082. doi: 10.1001/jama.2015.12916CrossrefMedlineGoogle Scholar
  • 29. Nanna MG, Peterson ED, Wojdyla D, Navar AM. The accuracy of cardiovascular pooled cohort risk estimates in U.S. Older Adults.J Gen Intern Med. 2020; 35:1701–1708. doi: 10.1007/s11606-019-05361-4CrossrefMedlineGoogle Scholar
  • 30. Blaha MJ, Mortensen MB, Kianoush S, Tota-Maharaj R, Cainzos-Achirica M. Coronary artery calcium scoring: is it time for a change in methodology?.JACC Cardiovasc Imaging. 2017; 10:923–937. doi: 10.1016/j.jcmg.2017.05.007CrossrefMedlineGoogle Scholar
  • 31. Forbang NI, McClelland RL, Remigio-Baker RA, Allison MA, Sandfort V, Michos ED, Thomas I, Rifkin DE, Criqui MH. Associations of cardiovascular disease risk factors with abdominal aortic calcium volume and density: The Multi-Ethnic Study of Atherosclerosis (MESA).Atherosclerosis. 2016; 255:54–58. doi: 10.1016/j.atherosclerosis.2016.10.036CrossrefMedlineGoogle Scholar
  • 32. Leber AW, Knez A, White CW, Becker A, von Ziegler F, Muehling O, Becker C, Reiser M, Steinbeck G, Boekstegers P. Composition of coronary atherosclerotic plaques in patients with acute myocardial infarction and stable angina pectoris determined by contrast-enhanced multislice computed tomography.Am J Cardiol. 2003; 91:714–718. doi: 10.1016/s0002-9149(02)03411-2CrossrefMedlineGoogle Scholar
  • 33. Bergman BC, Brozinick JT, Strauss A, Bacon S, Kerege A, Bui HH, Sanders P, Siddall P, Kuo MS, Perreault L. Serum sphingolipids: relationships to insulin sensitivity and changes with exercise in humans.Am J Physiol Endocrinol Metab. 2015; 309:E398–E408. doi: 10.1152/ajpendo.00134.2015CrossrefMedlineGoogle Scholar
  • 34. Ng TW, Ooi EM, Watts GF, Chan DC, Weir JM, Meikle PJ, Barrett PH. Dose-dependent effects of rosuvastatin on the plasma sphingolipidome and phospholipidome in the metabolic syndrome.J Clin Endocrinol Metab. 2014; 99:E2335–E2340. doi: 10.1210/jc.2014-1665CrossrefMedlineGoogle Scholar
  • 35. Tarasov K, Ekroos K, Suoniemi M, Kauhanen D, Sylvänne T, Hurme R, Gouni-Berthold I, Berthold HK, Kleber ME, Laaksonen R, et al. Molecular lipids identify cardiovascular risk and are efficiently lowered by simvastatin and PCSK9 deficiency.J Clin Endocrinol Metab. 2014; 99:E45–E52. doi: 10.1210/jc.2013-2559CrossrefMedlineGoogle Scholar
  • 36. Hilvo M, Meikle PJ, Pedersen ER, Tell GS, Dhar I, Brenner H, Schöttker B, Lääperi M, Kauhanen D, Koistinen KM, et al. Development and validation of a ceramide- and phospholipid-based cardiovascular risk estimation score for coronary artery disease patients.Eur Heart J. 2020; 41:371–380. doi: 10.1093/eurheartj/ehz387CrossrefMedlineGoogle Scholar
  • 37. The Atherosclerosis Risk in Communities (ARIC) Study: design and objectives. The ARIC investigators.Am J Epidemiol. 1989; 129:687–702.CrossrefMedlineGoogle Scholar
  • 38. Fried LP, Borhani NO, Enright P, Furberg CD, Gardin JM, Kronmal RA, Kuller LH, Manolio TA, Mittelmark MB, Newman A. The Cardiovascular Health Study: design and rationale.Ann Epidemiol. 1991; 1:263–276. doi: 10.1016/1047-2797(91)90005-wCrossrefMedlineGoogle Scholar
  • 39. Friedman GD, Cutter GR, Donahue RP, Hughes GH, Hulley SB, Jacobs DR, Liu K, Savage PJ. CARDIA: study design, recruitment, and some characteristics of the examined subjects.J Clin Epidemiol. 1988; 41:1105–1116. doi: 10.1016/0895-4356(88)90080-7CrossrefMedlineGoogle Scholar