Background: Heart failure (HF) is highly prevalent among older adults and is associated with high costs. Although serum total nonesterified fatty acids (NEFAs) have been positively associated with HF risk, the contribution of each individual NEFA to HF risk has not been examined. Objective: The aim of this study was to examine the association of individual fasting NEFAs with HF risk in older adults. Methods: In this prospective cohort study of older adults, we measured 35 individual NEFAs in 2,140 participants of the Cardiovascular Health Study using gas chromatography. HF was ascertained using review of medical records by an endpoint committee. Results: The mean age was 77.7 ± 4.4 years, and 38.8% were male. During a median follow-up of 9.7 (maximum 19.0) years, 655 new cases of HF occurred. In a multivariable Cox regression model controlling for demographic and anthropometric variables, field center, education, serum albumin, glomerular filtration rate, physical activity, alcohol consumption, smoking, hormone replacement therapy, unintentional weight loss, and all other measured NEFAs, we observed inverse associations (HR [95% CI] per standard deviation) of nonesterified pentadecanoic (15:0) (0.73 [0.57–0.94]), γ-linolenic acid (GLA) (0.87 [0.75–1.00]), and docosahexaenoic acid (DHA) (0.73 [0.61–0.88]) acids with HF, and positive associations of nonesterified stearic (18:0) (1.30 [1.04–1.63]) and nervonic (24:1n-9) (1.17 [1.06–1.29]) acids with HF. Conclusion: Our data are consistent with a higher risk of HF with nonesterified stearic and nervonic acids and a lower risk with nonesterified 15:0, GLA, and DHA in older adults. If confirmed in other studies, specific NEFAs may provide new targets for HF prevention.
With improved life expectancy and a growing aging US population, heart failure (HF) remains a major public health issue . Approximately 5 million Americans live with HF, and each year, nearly 1 million Americans are diagnosed with HF . The prevalence of HF is expected to continue to rise, partly due to improved survival after myocardial infarction, a shift in demographics to an older population, and an increase in the prevalence of type 2 diabetes (T2D) and obesity (two important risk factors for HF [2-7]). Furthermore, HF remains one of the leading causes of office visits among older adults and is associated with high mortality [8-10]. This underscores the need for identification of novel and modifiable risk factors that can be used to improve prevention of HF. Nonesterified fatty acids (NEFAs) are by-products of lipolysis  and can impair insulin signaling in skeletal muscle, promote hepatic gluconeogenesis, and reduce insulin secretion [12, 13], with subsequent development of T2D. NEFAs can also impair endothelial function  and might contribute to ischemic cardiomyopathy. Our group  has previously shown that serum total NEFAs were associated with a higher risk of T2D  and HF . Because total serum NEFAs comprise >30 individual fatty acids with various carbon lengths and degrees of unsaturation, it is unclear whether the observed elevated risk of HF with serum total NEFAs results from equal contribution of each individual NEFA or from a subset of individual NEFAs. For example, myocardium preferentially uses long-chain esterified fatty acids/NEFAs to meet energy demand via beta-oxidation . Animal experiments have shown that beta-oxidation of long-chain monounsaturated fatty acids can cause cardiotoxicity . In humans, phospholipid source of nervonic acid (24:1n-9) has been positively associated with cardiovascular risk factors and with a higher risk of HF , while very long-chain saturated fatty acids (20:0, 22:0, and 24:0) have been associated with a lower risk of HF . No previous study has examined the association of nonesterified nervonic acid or other individual NEFAs with HF. Identification of specific NEFAs that play a major role in the development of HF could provide potential new targets for future dietary and/or pharmacologic interventions. Hence, we sought to comprehensively examine this gap using an agnostic approach to test the primary hypothesis that some, but not all, individual NEFAs are associated with HF risk in older adults. In a secondary analysis, we tested the hypothesis that the association of individual NEFAs with HF is stronger when restricted to the first 5 years of follow-up, as previously reported between serum total NEFAs and incidence of diabetes and HF in the same cohort [15, 17].
Materials and Methods
We used data collected on participants of the Cardiovascular Health Study (CHS), which is a prospective cohort consisting of 5,888 men and women aged 65 years and older who were selected for recruitment from four US communities (Forsyth County, NC, USA; Washington County, MD, USA; Sacramento County, CA, USA; and Pittsburgh, PA, USA). A description of design, methods, and procedures in the CHS has been previously published . In brief, 5,201 adults aged ≥65 years were enrolled between 1989 and 1990. In addition, a supplemental cohort of 687 predominantly African American adults were recruited between 1992 and 1993 from three of the same communities (excepting Washington County). Of the 4,413 participants who attended the 1996–1997 visit (year 9), the current study focused on 2,402 CHS participants who underwent an oral glucose tolerance test at the 1996–1997 examination to characterize both individual fasting NEFAs and post-OGTT NEFAs. Of the 2,402 people with OGTT, we selected 2,145 participants with unthawed fasting and post-OGTT serum samples due to sensitivity of individual NEFAs to thaw-freeze cycles. After excluding specimens that were oxidized or hemolyzed (n = 5), fasting NEFA measurements were available on 2,140. In this report, we focused on fasting NEFAs, and we excluded participants with prevalent HF (n = 137), resulting in a final analysis sample of 2,003 participants. The institutional review board of each participating center approved the study, and all participants gave informed written consent to participate in the study. A separate exemption approval to use de-identified archived serum samples to measure individual NEFAs was obtained from the Tufts University/Tufts Medical Center Institutional Review Board.
Serum lipids were extracted using previously described techniques [23, 24], and the NEFA subfraction was separated via solid-phase chromatography using an aminopropyl column , followed by methylation to form fatty acid methyl esters [24, 26]. We used an Autosystem XL gas chromatograph (Perkin Elmer, Boston, MA, USA) equipped with a 100-m × 0.25-mm capillary column (HP INNOWax, Agilent Technologies, CA, USA) to separate the fatty acid methyl esters. We identified 35 fatty acids using authenticated standards (NuCheck Prep, MN, USA). Inter-assay coefficients of variation were 0.5–4.3% for fatty acids with concentrations >5 mol%; 1.8–7.1% for those present at concentrations between 1 and 5 mol%; and 2.8–11.1% for those present at concentrations <1 mol%. All measurements for NEFAs were completed at the Jean Mayer USDA Human Nutrition Research Center on Aging, Tufts University, Boston, MA.
HF events were adjudicated by the CHS Events Committee, as previously described [27, 28]. In brief, HF validation required symptoms (shortness of breath, fatigue, orthopnea, and paroxysmal nocturnal dyspnea); clinical signs including pulmonary rales, peripheral edema, gallop rhythm, and displaced left ventricular apical impulse; pulmonary edema and increasing cardiomegaly on chest X-rays; and use of diuretics, digitalis, or vasodilators for HF treatment. Incident HF was ascertained upon review of pertinent data on hospitalization or outpatient visits such as medical history, physical examination, report of chest X-ray, and medications. HF subtype (reduced or preserved ejection fraction) was based on left ventricular ejection fraction (<45 and ≥45%, respectively).
We collected self-reported information on age, sex, ethnicity, educational attainment, smoking status, unintentional weight loss >10 pounds, health status, hormone replacement therapy (HRT), and alcohol consumption. Leisure time activity (kcal/week) was assessed with a modified Minnesota Leisure-Time Activities questionnaire . Weight, waist circumference, and height were measured using standardized protocols. Serum albumin was measured using standard methods. Hypertension was defined as systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or physician diagnosis of hypertension plus treatment with blood pressure lowering medications. T2D was defined if a participant was using insulin or oral hypoglycemic agents, or had a fasting glucose level of ≥7 mmol/L (126 mg/dL) or a nonfasting glucose level of ≥11.1 mmol/L (200 mg/dL). The glomerular filtration rate was estimated based on cystatin C, as previously described .
We used generalized additive models to test for departures from linearity for the association of each individual NEFA with incident HF. Only one individual NEFA (20:5n-3) showed evidence of nonlinearity in a model without adjustment for the other NEFAs; however, in the model with adjustment for other individual NEFAs, there was little evidence of meaningful nonlinearity. Therefore, we present only the estimates from the linear model for simplicity and clarity. We used Cox proportional hazards regression that included all 35 NEFAs to estimate hazard ratios with 95% confidence intervals for incident HF associated with a 1 standard deviation (SD) increment in each NEFA. Time at risk was calculated as the interval in days from the date of the 1996–1997 visit to the earliest of date of incident HF, date of death, date of loss to follow-up, or administrative censoring date (June 2015). Model 1 adjusted for age, sex, race, field center, and all other NEFAs. Model 2 controlled for all variables in model 1 plus education, serum albumin, BMI, waist circumference, estimated glomerular filtration rate using cystatin, physical activity, alcohol, smoking status, HRT, and unintentional weight loss. We also examined additional control for potential intermediate factors including prevalent diabetes, coronary artery disease, atrial fibrillation, hypertension, and use of blood pressure medications using model 2. We found no departure from proportional hazards; however, because our previous analysis of total NEFAs and HF showed a stronger association within the first 5 years of follow-up , we conducted a secondary analysis in the present study limiting follow-up to the first 5 years. Additional secondary analyses included association of fasting serum total NEFAs measured on samples collected during 1996–1997 study visit with HF risk. Finally, we conducted sensitivity analyses excluding 370 people with prevalent coronary artery disease and excluding the first 2 years of follow-up.
The mean age of the 2,003 study participants at baseline (1996–1997) was 77.7 ± 4.4 years, 38.8% were male, and 14.2% were black. Table 1 presents baseline characteristics by quartiles of nonesterified nervonic acid. Higher serum concentration of nervonic acid was associated with older age, female sex, high blood pressure, and high alcohol consumption (Table 1). Means (SD) of the most abundant individual NEFAs (µmol/L) were 150.0 (62.9) for 18:1n-9 (oleic); 124.0 (44.5) for 16:0 (palmitic); 78.7 (32.5) for 18:2n-6 (linoleic); and 60.3 (17.2) for 18:0 (stearic) (see online suppl. Table 1; for all online suppl. material, see www.karger.com/doi/10.1159/000513917). During a median follow-up of 9.4 years, 655 (32.7%) participants developed HF, including 152 cases with reduced and 250 with preserved left ventricular ejection fraction. In a multivariable-adjusted Cox regression model that included all 35 NEFAs, each SD increment of nonesterified stearic acid (18:0) was associated with a 30% higher risk of HF (95% CI: 4–63%), adjusting for age, sex, race, field center, education, serum albumin, BMI, waist circumference, glomerular filtration rate based on cystatin C, leisure time physical activity, alcohol, smoking status, HRT, unintentional weight loss, and all other NEFAs (Table 2); corresponding values for nonesterified nervonic acid (24:1n-9) was 17% (95% CI: 6–29%), Table 2. In contrast, nonesterified pentadecanoic acid (15:0), γ-linolenic acid (GLA) (18:3n-6), and docosahexaenoic acid (DHA) (22:6n-3) were associated with a 27% (95% CI: 6–43%), 13% (95% CI: 0–25%), and 27% (95% CI: 12–39%) lower risk of HF, respectively, in a fully adjusted model (Table 2). Additional adjustment for potential mediators (T2D, hypertension- and blood pressure-lowering medications, CHD, and atrial fibrillation) did not materially alter the results (HR [95% CI] per SD increment: 0.78 (0.60–1.10) for 15:0; 1.30 (1.04–1.63) for 18:0; 1.16 (1.05–1.28) for 24:1n-9; 0.86 (0.74–1.00) for 18:3n-6; and 0.72 (0.60–0.87) for 22:6n-3). In a secondary analysis restricted to the first 5 years of follow-up, we observed associations of nonesterified pentadecanoic acid (15:0) (HR per SD: 0.55 [95% CI: 0.36–0.85]) and nonesterified nervonic acid (HR per SD: 1.24 [95% CI: 1.08–1.42]) with HF in a multivariable adjusted model (online suppl. Table 2). Additionally, we observed associations of individual NEFAs with HF that were not apparent with the longer follow-up period (online suppl. Table 2). Exclusion of the first 2 years of follow-up, or of 370 participants with prevalent coronary artery disease did not alter the conclusions. For example, for individual DHA (22:6n-3), (HR [95% CI]) per SD increment went from 0.73 (0.61–0.88) in main analysis to 0.73 (0.60–0.89) after exclusion of the first 2 years of follow-up and 0.74 (0.60–0.92) after exclusion of prevalent CAD. Corresponding values for stearic acid (18:0) were 1.30 (1.04–1.63), 1.35 (1.06–1.71), and 1.34 (1.03–1.74). In exploratory analyses, fasting serum total NEFAs (per SD increment) measured contemporaneously with individual NEFAs was not associated with HF risk (multivariable adjusted HR: 1.00 [95% CI: 0.91–1.09]). Last, examination of HF subtypes showed similar direction of effect sizes, albeit lack of precision due to limited cases of HF with preserved (n = 250) and reduced (n = 152) ejection fraction (online suppl. Table 3).
In this large prospective study of US older adults, we found a higher risk of HF with 2 individual NEFAs (stearic and nervonic acids) and a lower risk with 3 NEFAs (pentadecanoic acid, GLA, and DHA) in a multivariable adjusted model. Of these 5 NEFAs associated with HF, only 2 (pentadecanoic and nervonic acids) showed associations when follow-up was restricted to the first 5 years of the study in a secondary analysis.
To the best of our knowledge, this is the first study to comprehensively examine the association of individual NEFAs with HF risk in a large prospective cohort of older adults. These findings extend the interpretation of our previous work reporting a positive association of serum total NEFAs with HF risk  in the same cohort of older adults by suggesting that evaluation of the association of serum total NEFAs with HF risk may depend on the relative composition of individual NEFAs and direction of those associations (i.e., lower HF risk with 15:0, 18:3n-6, and 22:6n-3 vs. higher risk with 24:1n-9 and 18:0 NEFAs). Furthermore, assessment of individual NEFAs can help identify a subclass of NEFAs associated with higher versus lower risk of HF and inform future dietary and/or pharmacological approaches to alter serum concentrations of those specific NEFAs for HF benefit. Although we observed new individual NEFAs associated with HF when analysis was restricted to the first 5 years of follow-up, these findings are only exploratory and merit confirmation in future studies with adequate data to evaluate the impact of updating NEFAs over time on HF outcome. If confirmed in future studies, these findings might provide an opportunity to develop (i) new pharmacologic interventions that can alter the concentration of nervonic acid, for example, or (ii) new dietary recommendations to shift the balance between protective (i.e., DHA, pentadecanoic, and GLAs) and deleterious (i.e., stearic and nervonic acids) NEFAs to protect against HF.
In two large prospective cohorts, phospholipid very long-chain monounsaturated fatty acids were positively associated with cardiovascular risk factors and with a higher incidence of HF ; comparing the highest to the lowest quintile of plasma phospholipid nervonic acid (24:1n-9), multivariable adjusted hazard ratios were 1.75 (95% CI: 1.23–2.50) in the CHS and 1.92 (95% CI: 1.22–3.03) in the Atherosclerosis Risk in Communities (ARIC) Study . It is important to acknowledge that we cannot directly compare associations of phospholipid fraction with NEFAs as the former tends to reflect longer term intake, while NEFAs reflect a balance between lipolytic factors (i.e., catecholamines) and inhibitors of lipolysis (i.e., insulin or adiponectin). Nonetheless, it is important to note that nervonic acid from two different compartments (phospholipid vs. nonesterified fractions) showed a similar association with HF. Of note is that very long-chain fatty acids including nervonic acid (24:1n-9) are predominantly oxidized in peroxisomes, instead of mitochondria, and a lack of membrane-transporting enzymes for very long-chain fatty acids in peroxisomes [31, 32] can lead to accumulation of reactive oxygen species and other metabolites from beta-oxidation with subsequent apoptosis [33, 34]. Lower risk of HF observed with nonesterified DHA is consistent with observational and clinical trial data relating RBC membrane, phospholipid marine omega-3 fatty acids , or eicosapentaenoic acid/DHA supplements with HF .
The current study has some limitations including the observational study design which cannot preclude unmeasured or residual confounding as source of bias. Second, we only had one measurement of individual NEFAs at baseline and were not able to account for changes in serum NEFAs over time. Third, our cohort consisted of older adults with mean age of 77 years, thereby limiting generalizability of our findings to the general population. Fourth, we had limited statistical power and precision to evaluate associations of individual NEFAs with HF type (preserved vs. reduced ejection fraction). Fifth, we did not adjust for multiple comparisons, and our findings should be considered hypothesis-generating. Despite these limitations, our study has numerous strengths including the novel aspect of examining association of individual NEFAs with HF risk; availability of data on major confounding factors; relatively long follow-up in the CHS; multi-center design with participants from different geographic regions of the USA; a standardized approach to adjudicate HF by the CHS Endpoint Committee; and objective and accurate measurement of individual NEFAs with excellent coefficients of variation for individual NEFAs with high concentrations.
Given the current burden of HF in the USA, our data provide preliminary evidence for potential novel targets for pharmacological and/or dietary intervention to modify HF risk factors and with the ultimate goal to prevent development of HF. Thus, replication of our findings while addressing some of the shortcomings outlined above is warranted.
In conclusion, we have demonstrated in this large prospective study of older adults that individual NEFAs are differentially associated with HF risk. Specifically, we showed that higher serum levels of nonesterified DHA, GLA, and pentadecanoic acid were associated with lower incidence of HF risk, while higher levels of nonesterified stearic and nervonic acids were associated with a higher risk of HF.
We are indebted to the participants and staff of the Cardiovascular Health Study.
Statement of Ethics
The institutional review board of each participating center approved the study, and all participants gave informed written consent to participate in the study. A separate exemption approval to use de-identified archived serum samples to measure individ-ual NEFAs was obtained from the Tufts University/Tufts Medi-cal Center Institutional Review Board (BWH IRB protocol #2016P001518 [PI: Luc Djousse]).
Conflict of Interest Statement
The authors have no conflicts of interest to disclose. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Funding agencies did not have any role in the design and conduct of the study, and collection, management, analysis, and interpretation of the data.
This research was supported by contracts HHSN268201200036C, HHSN268200800007C, HHSN268201800001C, N01HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083, N01HC85086, and 75N92021D00006 and grants U01HL080295 and U01HL130114 from the National Heart, Lung, and Blood Institute (NHLBI), with additional contribution from the National Institute of Neurological Disorders and Stroke (NINDS). Additional support was provided by R01AG023629 and R01AG053325-01 from the National Institute on Aging (NIA). A full list of principal CHS investigators and institutions can be found at CHS-NHLBI.org. Brigham and Women’s Hospital (BWH) received NIH funding for this project.
Luc Djousse conceived, designed the study, secured funding, and drafted the manuscript. Mary Lou Biggs and Barbara Mc-Knight performed statistical analyses. Nirupa R. Matthan and Alice H. Lichtenstein performed nonesterified fatty acid measurements. Joachim H. Ix, Annette L. Fitzpatrick, Irena King, Rozenn N. Lemaitre, Jorge R. Kizer, and David S. Siscovick contributed to data analyses, results interpretation, critical review and editing of the manuscript. Kenneth Mukamal contributed to data collection, study design, data interpretation, and securing funding. All authors critically reviewed and approved the final version of the manuscript.