Abstract
Heart failure (HF) is among the leading diagnoses for those admitted to the hospital over 65 years old in high-income countries. While there is strong evidence for the use of pharmacological interventions in the treatment of HF with reduced ejection fraction (HFrEF), there is limited evidence for a similar approach to decreasing morbidity and mortality of HF with preserved ejection fraction (HFpEF). This discrepancy highlights the importance of lifestyle change (i.e. diet) for prevention of HFpEF. Given the paucity of data on dietary predictors of HFpEF and the recent changes in diagnostic criteria, we set out to assess the associations of dietary and demographic predictors with HFpEF in the Coronary Artery Risk Development in Young Adults (CARDIA) cohort. We found that males in their fifties compared to age-matched females had worse measures of diastolic function (e’ lateral: 8.47 ± 2.28 vs. 8.98 ± 2.49, p < .001) and myocardial shortening (i.e. GLS: −15.91 ± 2.73 vs −16.98 ± 3.1, p < .001). Each one point of GLS increase was associated with 12 % increase in risk of HFpEF, while HDL intake was found to be protective against HFpEF. We also found that higher dietary HDL intake when individuals were in their fifties was associated with higher (i.e. better) measures of both e’ lateral and e’ septal velocities. Our data indicate that GLS appears to be a robust predictor of HFpEF and is influenced by dietary behaviors across the lifespan that affect BMI in males and hypertension in females.
1. Introduction
Heart failure with preserved ejection fraction (HFpEF) accounts for more than half of all HF cases and its prevalence continues to rise as our population ages [1]. While there is strong evidence for the use of pharmacological interventions in the treatment of HF with reduced ejection fraction (HFrEF), there is currently limited evidence for the use of pharmacological treatments for decreasing the morbidity and mortality of HFpEF [2]. This discrepancy highlights the importance of health behavior change, including diet, for the primary, secondary, and tertiary prevention of HFpEF [3].
Given that hypertension is a major risk factor for HF, one commonly recommended dietary treatment for HFpEF is the Dietary Approaches to Stop Hypertension (DASH) diet [2]. Although some evidence exists showing the effectiveness of the DASH diet in both the prevention and treatment of heart failure, clinical trials are still lacking and data on its benefits in HFpEF are limited [4]. Several mechanisms have been proposed to explain the beneficial effects of DASH diet on HF, which extend beyond hypertension management. The increased consumption of key micronutrients promoted by the DASH diet (including potassium, magnesium, and calcium) may increase insulin sensitivity [5] and exercise capacity [6] while decreasing endothelial dysfunction [6], metabolic risk [7], and low-density lipoprotein cholesterol levels [8]. Why then isn't the cardiovascular burden lessening worldwide but rather is increasing? Our argument is that the DASH diet is a radical shift that most individuals consuming a Western-type diet consider a challenge. Given that the implementation of large-scale dietary change is a major public health task, there exists a knowledge gap on specific macro- and micronutrients that can feasibly be manipulated to prevent or treat HFpEF, as this would represent a manageable and sustainable lifestyle change.
Reducing dietary sodium is a common prescription for individuals with HF, although the evidence supporting this recommendation is limited [9]. In a 2018 Cochrane systematic review of randomized controlled trials that assessed the effects of reduced sodium intake on HF, no high-quality evidence was found to suggest that reduced sodium intake can improve clinical outcomes [10]. There is very limited data available exploring the effects of sodium consumption on individuals with HFpEF, specifically [11]. In contrast to sodium, magnesium does appear to help prevent and potentially treat HF via its flow-mediated vasodilatory [12] and anti-inflammatory effects [13]. In a large cohort study, low magnesium intake (< 2.3 mg/kg) was associated with an increased risk of HF hospitalizations among African Americans [14]. Similar findings were observed in another cohort study of over 80,000 post-menopausal women [15]. However, in their sub-cohort analysis, while low magnesium intake was associated with an increased risk of incident HFrEF, it was not predictive of HFpEF. On the contrary, research in animal models demonstrated that magnesium supplementation can reverse cardiac diastolic dysfunction and therefore potentially treat HFpEF [16]. A potential explanation for the lack of an association between magnesium and HFpEF in humans is the inaccurate classification of individuals with HFpEF compared to HFrEF. While HFpEF was previously diagnosed as HF with LVEF >50 % and signs and symptoms of HF, diagnosis now also requires evidence of increased left ventricular filling pressures. See Fig. 1 for variables used to assess diastolic dysfunction in this study [2,17].
Fig. 1.
Variables used to assess diastolic dysfunction in this study and key cutoff values.
While micronutrients, such as magnesium, may aid in the prevention and treatment of HFpEF, macronutrients, including HDL and fructose, may also play a significant role. Recent work in mice showed that reconstituted Milano HDL can reverse the pathological cardiac remodelling seen in HFpEF [18] suggesting that diets high in HDL may be protective against HFpEF, although data in humans is lacking to support this association. Recent work in rats has shown that short-term (4 weeks) moderate increases in dietary fructose can cause diastolic dysfunction [19]. Our recently published longitudinal cohort study showed that dietary fructose intake was associated with increased CVD risk by mid-life [20], and hypertensive target organ damage [21], although potential associations between high fructose consumption and HFpEF, specifically, have yet to be explored.
Given the limited evidence supporting pharmacological treatments for reducing morbidity and mortality in HFpEF, the paucity of data on dietary predictors of HFpEF, and the recent changes in diagnostic criteria, we set out to assess the potential associations of dietary predictors of HFpEF using the most up-to-date diagnostic criteria. The primary aim of this study was to explore dietary predictors of HFpEF, specifically magnesium, sodium, fructose, LDL, and HDL, among middle-aged adults in the Coronary Artery Risk Development in Young Adults (CARDIA) study. The CARDIA study is a prospective longitudinal cohort study that was initiated in 1985 to examine lifestyle and demographic determinants of clinical and subclinical cardiovascular disease. To our knowledge, no previous studies have investigated the role of diet on HFpEF within the CARDIA cohort. Given the emerging interest in using GLS for the diagnosis of HFpEF, our secondary aim was to explore the predictive potential of GLS for HFpEF. Our tertiary aim was to examine the associations between dietary factors and key echocardiographic variables related to LV diastolic dysfunction.
2. Methods
2.1. Study sample
All data were obtained from the CARDIA Study Data Coordinating Center (University of Alabama) using the process outlined at https://www.cardia.dopm.uab.edu/publications-2/publications-documents. Black and white, men and women (n = 5114) ages 18 to 30 years were recruited from four centers (Birmingham, AL; Chicago, IL; Minneapolis, MN; and Oakland, CA). In addition to baseline data, eight follow-ups have been conducted to date (years 2, 5, 7, 10, 15, 20, 25, and 30) with retention rates (among survivors) of 91 %, 86 %, 81 %, 79 %, 74 %, 72 %, 72 %, and 71 %, respectively.
In the present study, we used two subsets of participants to achieve our primary, secondary, and tertiary aims. For our primary and secondary aims, we analyzed the data from participants who had completed the dietary questionnaire at year 20 and speckle tracking echocardiography (STE) scans at year 30 by which measurements of GLS were obtained. For our tertiary aim, we analyzed the data from participants who had completed the dietary questionnaire at year 20 and Doppler echocardiography at year 30. We excluded participants who had an EF <50 % given that we wanted to compare individuals with HFpEF to individuals without heart failure. All variables had <5 % missing data and no imputations were completed.
2.2. Study measures
Dietary intake was assessed at year 0, 7, and 20 using the CARDIA Diet History questionnaire which was developed and validated for the CARDIA study [22]. The questionnaire was interviewer-administered and included a food frequency questionnaire of 1609 food items. The reference time for recall was the past month and interviewers used follow-up questions to deduce serving sizes and food frequency. All outliers were included in the participants' data records and no imputations were performed for missing data. From this raw data, daily nutrient scores were created (e.g. daily magnesium intake in milligrams, daily LDL intake in mg/day). To allow for more accurate comparisons across participants with varying total calorie consumption, we expressed sucrose and fructose as a percentage of total calories. We converted their gram values by multiplying by 4 (calories per gram of sugar), dividing by the total calorie intake, and then multiplying by 100.
Covariates included in our models included baseline age and BMI, total physical activity at year 20, and years of regular cigarette use at year 30. Participation in physical activity was assessed using the validated CARDIA Physical Activity History questionnaire [23]. Participants self-reported their engagement in various moderate (e.g. waling, golfing, gardening) and vigorous (e.g. running, racquet sports, job-related lifting) activities over the past year. A computer algorithm then calculated a “physical activity score” for each activity a participant engaged in based on the intensity of the activity (in METs), months of participation in that activity, and a weighting factor based on whether participants engaged in a minimum weekly duration threshold (ranging from 2 to 5 h per week). The “total activity score” was the sum of individual activity scores (e.g. say a participant who walked, played tennis, and weight lifted). A total activity score of 300 corresponded with the current physical activity guidelines of 150 min of moderate physical activity per week. With regards to smoking habits, participants were asked to report “altogether, how many years did [they] smoke cigarettes regularly?”, where “regularly” was defined as 5 cigarettes per week, almost every week.
Doppler echocardiography was conducted by trained technicians with the Artida cardiac ultrasound scanner (Toshiba Medical Systems, Otawara, Japan) using a standardized protocol. The LVEF variable used in this study was measured using 2D-guided M-Mode with a four-chamber view. Diastolic dysfunction parameters e’(septal) and e’(lateral) were measured using recordings of transmitral flow. LAVi was calculated by dividing the 2D 4-chamber left atrial volume by body surface area. STE images for myocardial strain, including GLS, were analyzed using a 16-segment model with Wall Motion 2D Tracking software (Toshiba Medical Systems). The GLS variable used in this study was measured from a 3D, 4-chamber view. The CARDIA scanning protocol can be found at http://cardia2.dopm.uab.edu/). An overview of main study variables and their associated timepoints are represented in Fig. 2.
Fig. 2.
Measurement timepoints for main study variables.
2.3. Statistical analysis
Statistical Package for the Social Sciences (SPSS) version 29.0.2.0 was used for all analyses. The Kolmogorov-Smirnov test showed that our continuous variables were not normally distributed. A correlation matrix with Spearman's rho for non-parametric data was used to assess for collinearity among predictor variables and covariates. While originally considered for inclusion in our models, BMI at year 20, BMI at year 30, triglyceride intake at year 30, and potassium intake at year 20 were excluded due multicollinearity. To achieve our primary and secondary aims, three hierarchical binary logistic regressions were completed to explore dietary and myocardial strain predictors of HFpEF defined as (A) LVEF ≥50 % and e’ (septal) < 7 cm/s, and (B) LVEF ≥50 % and e’ (lateral) < 10 cm/s, and (C) HFpEF as LVEF ≥50 % AND LAVi> > 34 mL/m2. To achieve our tertiary aim, four hierarchical multiple linear regressions were then completed to explore associations between dietary predictors and global longitudinal strain, e’ (septal), e’(lateral), and LAVi.
3. Results
Participant characteristics are reported in Table 1. The median age of study participants included in this study at the year 30 (Y30) follow-up is 56-year-old, which is when HFpEF assessment was completed for the purposes of the present study.
Table 1.
Participant characteristics.
| Continuous variables | N | Median (SD) |
|---|---|---|
| Age Y0 | 2380 | 26 (4) |
| BMI Y0 | 2370 | 23.25 (4.52) |
| % calories from sucrose Y20 | 2380 | 7.44 (3.87) |
| % calories from fructose Y20 | 2380 | 4.56 (3.44) |
| Sodium (mg) Y20 | 2380 | 3080.10 (1812.47) |
| Magnesium (mg) Y20 | 2379 | 325.87 (184.81) |
| Total physical activity score Y20 | 2361 | 276.00 (276.51) |
| LDL (mg/dL) Y30 | 2296 | 109.00 (32.92) |
| HDL (mg/dL) Y30 | 2343 | 57.00 (18.99) |
| Years of regular cigarette use Y30 | 2349 | 0.00 (12.62) |
| e’ lateral (cm/s) Y30 | 2293 | 11.43 (2.73) |
| e’ septal (cm/s) Y30 | 2290 | 8.77 (2.40) |
| LAVi (mL/m2) Y30 | 2380 | 24.32 (7.93) |
| GLS (%) Y30 | 1554 | −16.54 (2.98) |
| Categorical variables | N | Frequency (%) |
|---|---|---|
| Race, Black | 1068 | 44.9 |
| Race, White | 1305 | 54.8 |
| Race, Hispanic | 7 | 0.3 |
| BP > 130/80 mmHg Y0 | 294 | 12.4 |
| BP > 130/80 mmHg Y20 | 142 | 6.0 |
* Values are medians (SD) for continuous variables and numbers (%) for categorical variables.
*“Regular” cigarette use was defined as 5 cigarettes or more per week, almost every week.
Table 2 presents dietary, demographic, and myocardial strain predictors of HFpEF. A hierarchical approach was used to identify key predictors and potential protective factors against HFpEF. Model 1 includes the baseline (i.e. measured at Y0, when the participants had a median age of 26) characteristics age, sex, race, BMI, and BP > 130/80 mmHg. In model 2, variables measured at Y20 and Y30 of follow up were added – BP > 130/80 mmHg at Y20, fructose intake at Y20, sodium intake at Y20, LDL intake at Y30, GLS at Y30, and years of regular cigarette use at Y30. Y20 and Y30 timepoints denote 20 and 30 years later, respectively, from baseline (i.e Y0), with median age of the participants being 46 and 56, respectively. Models 3, 4, and 5 introduced potential protective factors: HDL at Y30, magnesium at Y20, and physical activity level at Y20, respectively.
Table 2.
Hierarchical binary logistic regressions with demographic, dietary, and myocardial strain predictors of HFpEF.
| Model 1 |
Model 2 |
Model 3 |
Model 4 |
Model 5 |
||||||
|---|---|---|---|---|---|---|---|---|---|---|
| OR [CI] | p | OR [CI] | p | OR [CI] | p | OR [CI] | p | OR [CI] | p | |
| A. HFpEF as LVEF ≥ 50 % AND e’ (septal) < 7 cm/s (N = 1162 w/o HF, N = 248 w/ HFpEF) | ||||||||||
| obsHBP Y0 | 1.205 [0.786–1.846] | 0.393 | 1.129 [0.728–1.750] | 0.588 | 1.132 [0.729–1.756] | 0.581 | 1.134 [0.731–1.76] | 0.575 | 1.134 [0.73–1.762] | 0.576 |
| BMI Y0 | 1.044 [1.010–1.078] | 0.010 | 1.046 [1.012–1.081] | 0.008 | 1.041 [1.007–1.077] | 0.018 | 1.041 [1.007–1.077] | 0.018 | 1.041 [1.007–1.077] | 0.018 |
| obsHBP Y20 | 1.945 [1.111–3.405] | 0.020 | 1.856 [1.054–3.267] | 0.032 | 1.855 [1.054–3.265] | 0.032 | 1.855 [1.054–3.265] | 0.032 | ||
| % cal fructose Y20 | 0.999 [0.955–1.046] | 0.982 | 0.993 [0.949–1.04] | 0.776 | 0.992 [0.948–1.039] | 0.744 | 0.992 [0.948–1.039] | 0.745 | ||
| Years of cig use Y30 | 0.997 [0.986–1.009] | 0.654 | 0.997 [0.985–1.008] | 0.569 | 0.997 [0.986–1.008] | 0.590 | 0.997 [0.985–1.008] | 0.591 | ||
| GLS Y30 | 1.117 [1.063–1.175] | <0.001 | 1.114 [1.060–1.171] | <0.001 | 1.114 [1.06–1.171] | <0.001 | 1.114 [1.06–1.171] | <0.001 | ||
| HDL Y30 | 0.989 [0.980–0.998] | 0.013 | 0.989 [0.98–0.997] | 0.012 | 0.989 [0.98–0.998] | 0.013 | ||||
| Magnesium Y20 | 1 [0.999–1.001] | 0.753 | 1.000 [0.999–1.001] | 0.759 | ||||||
| B. HFpEF as LVEF ≥ 50 % AND e’ (lateral) < 10 cm/s (N = 1068 w/o HF, N = 345 w/ HFpEF) | ||||||||||
| obsHBP Y0 | 1.263 [0.86–1.855] | 0.234 | 1.172 [0.787–1.746] | 0.435 | 1.173 [0.787–1.747] | 0.433 | 1.163 [0.78–1.733] | 0.459 | 1.156 [0.775–1.724] | 0.478 |
| BMI Y0 | 1.02 [0.99–1.051] | 0.198 | 1.02 [0.989–1.052] | 0.207 | 1.016 [0.985–1.049] | 0.313 | 1.015 [0.984–1.048] | 0.342 | 1.015 [0.984–1.047] | 0.350 |
| obsHBP Y20 | 2.064 [1.215–3.508] | 0.007 | 2.002 [1.174–3.414] | 0.011 | 2.008 [1.177–3.426] | 0.010 | 2.011 [1.178–3.43] | 0.010 | ||
| % cal fructose Y20 | 0.967 [0.927–1.008] | 0.112 | 0.961 [0.921–1.003] | 0.068 | 0.965 [0.924–1.007] | 0.098 | 0.965 [0.925–1.007] | 0.103 | ||
| Years of cig use Y30 | 1.012 [1.002–1.021] | 0.020 | 1.011 [1.001–1.021] | 0.025 | 1.011 [1.001–1.021] | 0.034 | 1.01 [1.001–1.020] | 0.038 | ||
| GLS Y30 | 1.122 [1.073–1.174] | <0.001 | 1.12 [1.071–1.171] | <0.001 | 1.12 [1.071–1.171] | <0.001 | 1.12 [1.072–1.171] | <0.001 | ||
| HDL Y30 | 0.992 [0.984–0.999] | 0.029 | 0.992 [0.984–1.000] | 0.038 | 0.992 [0.984–1.000] | 0.044 | ||||
| Magnesium Y20 | 0.999 [0.998–1.000] | 0.234 | 0.999 [0.998–1.000] | 0.303 | ||||||
| C. HFpEF as LVEF ≥ 50 % AND LAVi> > 34 mL/m2 (N = 1294 w/o HF, N = 151 w/ HFpEF) | ||||||||||
| obsHBP Y0 | 0.737 [0.400–1.359] | 0.329 | 0.675 [0.363–1.253] | 0.213 | 0.664 [0.356–1.239] | 0.198 | 0.67 [0.359–1.252] | 0.209 | 0.703 [0.376–1.316] | 0.271 |
| BMI Y0 | 1.042 [1.003–1.082] | 0.035 | 1.040 [1.000–1.080] | 0.049 | 1.046 [1.006–1.087] | 0.022 | 1.049 [1.009–1.09] | 0.017 | 1.052 [1.011–1.094] | 0.012 |
| obsHBP Y20 | 2.155 [1.129–4.114] | 0.020 | 2.404 [1.254–4.610] | 0.008 | 2.406 [1.253–4.619] | 0.008 | 2.411 [1.253–4.639] | 0.008 | ||
| % cal fructose Y20 | 0.977 [0.922–1.034] | 0.421 | 0.986 [0.931–1.046] | 0.646 | 0.978 [0.922–1.038] | 0.460 | 0.974 [0.917–1.035] | 0.394 | ||
| Years of cig use Y30 | 0.992 [0.978–1.006] | 0.275 | 0.993 [0.979–1.008] | 0.350 | 0.994 [0.980–1.009] | 0.438 | 0.995 [0.981–1.010] | 0.519 | ||
| GLS Y30 | 1.028 [0.969–1.090] | 0.361 | 1.035 [0.975–1.099] | 0.261 | 1.036 [0.975–1.100] | 0.256 | 1.034 [0.974–1.099] | 0.275 | ||
| HDL Y30 | 1.017 [1.008–1.027] | <0.001 | 1.016 [1.007–1.026] | <0.001 | 1.016 [1.006–1.025] | 0.001 | ||||
| Magnesium Y20 | 1.001 [1.000–1.002] | 0.022 | 1.001 [1.000–1.002] | 0.100 | ||||||
Notes. ObsHBP = BP >130/80 mmHg at follow-up, irrespective of hypertension diagnosis. Model 1 adjusted for race, sex, and age. Models 2, 3, & 4 adjusted for race, sex, age, sodium at Y20, LDL at Y30, and years of regular cigarette use at Y30 (‘regular’ = 5 cigarettes/wk., almost every week). Model 5 adjusted for race, sex, age, sodium at Y20, LDL at Y30, years of regular cigarette use ay Y30, and physical activity level at Y20.
Statistically significant results are bolded for emphasis.
In regression A (HFpEF defined as LVEF ≥50 % and e’ (septal) < 7 cm/s) the incidence of HFpEF at Y30 of follow up was 17.6 %. Each one point increase in BMI at Y0 was associated with an increased risk of HFpEF from 4.1 to 4.6 %. In addition, a BP > 130/80 mmHg at Y20, irrespective of hypertension diagnosis, was associated with a 86–95 % increased risk of HFpEF 10 years later. Each one point increase in GLS (i.e. GLS becoming less negative by one point) measured at Y30 was associated with an increased risk of HFpEF by 11.4–11.7 %. HDL intake was associated with reduced risk of HFpEF when added in model 3 and remained significant after the adjustment for magnesium consumption and physical activity levels in models 4 and 5 with a decreased risk of 1.1 % per 1 mg/dL increase. In regression B (HFpEF defined as LVEF ≥50 % and e’ (lateral) < 10 cm/s), incidence was 24.4 %. BP > 130/80 mmHg at Y20 was associated with a 100–106 % increased risk of HFpEF 10 years later. Likewise, for every one point increase in GLS, the risk of HFpEF increased by 12 %. As in regression A, intake of HDL was associated with reduced risk of HFpEF, albeit only by 0.8 %. Unique to regression B, for every year of regular cigarette use, the risk of HFpEF increased by 1 %. In regression C (HFpEF defined as LVEF ≥50 % AND LAVi> > 34 mL/m2), incidence of HFpEF was 10.7 %. Each one point increase in BMI at Y0 was associated with an increased risk of HFpEF from 4 to 5 %. In addition, BP > 130/80 mmHg at Y20 was associated with a 115–141 % increased risk of HFpEF 10 years later. Similarly to regressions A and B, HDL was associated with reduced risk of HFpEF by ~2 % across models 3–5. Unique to regression C, for each point increase in magnesium, there was an increased risk of HFpEF by 0.1 %.
Dietary and demographic predictors of GLS are presented in Table 3. Hierarchical multiple linear regressions were conducted using the same five models. Female sex was associated with a more negative (improved) GLS, indicating better myocardial strain among females with a small effect size. BP > 130/80 mmHg at Y20 was directly correlated with GLS across all models with a small effect size.
Table 3.
Hierarchical multiple linear regression with dietary predictors of global longitudinal strain.
| β-weight (standardized) |
|||||
|---|---|---|---|---|---|
| Model | 1 | 2 | 3 | 4 | 5 |
| Female Sex | −0.171⁎⁎ | −0.177⁎⁎ | −0.157⁎⁎ | −0.157⁎⁎ | −0.156⁎⁎ |
| obsHBP Y0 | 0.024 | 0.019 | 0.019 | 0.019 | 0.019 |
| BMI Y0 | 0.013 | 0.011 | 0.006 | 0.006 | 0.006 |
| obsHBP Y20 | 0.065⁎ | 0.062⁎ | 0.062⁎ | 0.062⁎ | |
| % cal fructose Y20 | −0.017 | −0.023 | −0.023 | −0.023 | |
| Sodium Y20 | −0.018 | −0.019 | −0.02 | −0.02 | |
| Years of cig use Y30 | 0.045 | 0.043 | 0.043 | 0.043 | |
| LDL Y30 | 0.002 | −0.004 | −0.004 | −0.005 | |
| HDL Y30 | −0.051 | −0.051 | −0.051 | ||
| Magnesium Y20 | 0.002 | 0.001 | |||
Note. N= 1451. ObsHBP = BP >130/80 mmHg at follow-up, irrespective of hypertension diagnosis. Model 1 adjusted for race and age. Models 2, 3, & 4 adjusted for race, age, and years of regular cigarette use at Y30 (‘regular’ defined as 5 cigarettes per week, almost every week). Model 5 adjusted for race, age, years of regular cigarette use at Y30, and physical activity level at Y20.
p < .05.
p < .001.
Dietary and demographic predictors of e’ septal, e’ lateral velocities, and LAVi are presented in Table 4. HDL was directly correlated with e’ septal, e’ lateral velocities, and LAVi even after adjustment for magnesium intake and physical activity levels in models 4 and 5. LDL was also negatively correlated with e’ septal velocity and LAVi across models 2–5. Interestingly, females on average had a significantly lower (worse) e’ lateral velocity, although this was only statistically significant following the addition of HDL in model 3 and remained significant in models 4 and 5. Females had a higher (improved) e’ septal velocity that was statistically significant across all models. Females also had a lower (improved) LAVi, although this relationship was not significant with the addition of hypothesized harmful dietary predictors and observed high BP in model 2 and physical activity in model 5. Magnesium consumption was positively correlated with e’ lateral velocity and remained significant after adjusting for physical activity levels in model 5. Magnesium was also positively correlated with LAVi, but did not remain significant in model 5.
Table 4.
Hierarchical multiple linear regression with dietary predictors of e’ septal, e’ lateral, and LAVi.
| β-weight (standardized) |
|||||
|---|---|---|---|---|---|
| Model | 1 | 2 | 3 | 4 | 5 |
| Predicting e’ lateral (N = 2141) | |||||
| Female Sex | −0.038 | −0.043 | −0.081⁎⁎ | −0.081⁎⁎ | −0.084⁎⁎ |
| obsHBP Y0 | −0.056⁎ | −0.052⁎ | −0.051⁎ | −0.050⁎ | −0.051⁎ |
| BMI Y0 | −0.052⁎ | −0.048⁎ | −0.033 | −0.032 | −0.032 |
| obsHBP Y20 | −0.086⁎⁎ | −0.084⁎⁎ | −0.082⁎⁎ | −0.082⁎⁎ | |
| % cal fructose Y20 | 0.012 | 0.021 | 0.016 | 0.017 | |
| Sodium Y20 | −0.026 | −0.024 | −0.073⁎ | −0.075⁎ | |
| Years of cig use Y30 | −0.055⁎ | −0.053⁎ | −0.050⁎ | −0.051⁎ | |
| LDL Y30 | −0.031 | −0.022 | −0.023 | −0.022 | |
| HDL Y30 | 0.099⁎⁎ | 0.094⁎⁎ | 0.095⁎⁎ | ||
| Magnesium Y20 | 0.067⁎ | 0.071⁎ | |||
| Predicting e’ septal (N = 2136) | |||||
| Female Sex | 0.091⁎⁎ | 0.102⁎⁎ | 0.076⁎ | 0.076⁎ | 0.067⁎ |
| obsHBP Y0 | −0.045⁎ | −0.043 | −0.042 | −0.042 | −0.045⁎ |
| BMI Y0 | 0.012 | 0.012 | 0.023 | 0.023 | 0.021 |
| obsHBP Y20 | −0.080⁎⁎ | −0.078⁎⁎ | −0.078⁎⁎ | −0.078⁎⁎ | |
| % cal fructose Y20 | −0.005 | 0.001 | 0.002 | 0.003 | |
| Sodium Y20 | 0.025 | 0.026 | 0.035 | 0.03 | |
| Years of cig use Y30 | −0.015 | −0.014 | −0.015 | −0.018 | |
| LDL Y30 | −0.060⁎ | −0.053⁎ | −0.052⁎ | −0.051⁎ | |
| HDL Y30 | 0.070⁎ | 0.070⁎ | 0.074⁎ | ||
| Magnesium Y20 | −0.012 | 0.001 | |||
| Predicting LAVi (N = 2220) | |||||
| Female Sex | −0.044⁎ | −0.033 | −0.058⁎ | −0.059⁎ | −0.031 |
| obsHBP Y0 | −0.006 | −0.013 | −0.013 | −0.012 | −0.005 |
| BMI Y0 | 0.054⁎ | 0.047⁎ | 0.057⁎ | 0.059⁎ | 0.065⁎ |
| obsHBP Y20 | 0.032 | 0.034 | 0.037 | 0.036 | |
| % cal fructose Y20 | −0.016 | −0.011 | −0.019 | −0.022 | |
| Sodium Y20 | 0.032 | 0.033 | −0.037 | −0.024 | |
| Years of cig use Y30 | −0.038 | −0.037 | −0.032 | −0.025 | |
| LDL Y30 | −0.053⁎ | −0.046⁎ | −0.048⁎ | −0.052⁎ | |
| HDL Y30 | 0.067⁎ | 0.060⁎ | 0.050⁎ | ||
| Magnesium Y20 | 0.096⁎ | 0.059 | |||
Note. ObsHBP = BP >130/80 mmHg at follow-up, irrespective of hypertension diagnosis.
Model 1 adjusted for race and age. Models 2, 3, & 4 adjusted for race, age, and years of regular cigarette use at Y30 (‘regular’ defined as 5 cigarettes per week, almost every wk). Model 5 adjusted for race, age, years of regular cigarette use at Y30, and physical activity level at Y20.
p < .05.
p < .001.
4. Discussion
In this retrospective observational study, our primary aim was to identify key dietary risk and protective factors of HFpEF using hierarchical modeling among middle-aged adults from the CARDIA cohort. We found higher dietary HDL-C levels to be associated with a reduced risk of HFpEF (~1 %) at mid-life. Consistent across multiple HFpEF definitions, higher BMI and impaired GLS emerged as strong risk factors, with GLS associated with a ~ 12 % increased risk, reinforcing its diagnostic relevance—the focus of our secondary aim. Our tertiary aim was to explore the associations between dietary factors and echocardiographic markers of left ventricular diastolic dysfunction. Overall, we found HDL-C to have a small protective effect against HFpEF-related diastolic dysfunction.
Increased dietary HDL and lower LDL intake, especially in female participants, demonstrated protective effects, lowering HFpEF risk and improving indices of diastolic function. Increased Global longitudinal strain (GLS) was a consistent predictor of increased HFpEF risk across both sexes, supporting the role of strain imaging with speckle tracking echocardiography as an early marker of subclinical disease.
Although HDL-C was associated with only a 1 % decrease in risk of HFpEF, this finding remains noteworthy given the comprehensive nature of our models, which accounted for multiple dietary and demographic covariates including hypertension, BMI, smoking, and physical activity levels. Our findings align with research in ApoA-I Milano mouse models [18] but diverge from a recent multicohort study (n = 16,925) where higher HDL-C was not associated with a lower risk of HFpEF after adjusting for known risk factors, including age, sex, race, BMI, hypertension, estimated glomerular filtration rate, diabetes, smoking, LDL-C, and triglycerides [24]. In contrast, they found HDL particle concentration to be associated with a lower risk of HFpEF suggesting the complex role of HDL in HFpEF pathophysiology. The complex pathophysiology at play is further supported by the work of Sasko and colleagues which showed that HDL's protective antioxidant capacity is diminished in patients with HFpEF [25]. The protective role of HDL in HFpEF was further supported by our tertiary analyses showing that HDL is positively correlated with higher e’ septal and lateral velocities, whereby higher velocities indicate improved diastolic function.
Contrary to our hypothesis, fructose intake was not found to be a significant risk factor of HFpEF. This was further supported by our tertiary analyses, where fructose consumption was not associated with key diastolic function markers. These results contradict the findings in animal studies, where short-term high-fructose diets have been linked to diastolic dysfunction [19]. One possible explanation is that the CARDIA cohort's overall fructose intake was relatively low −4.56 % of total calories – which amounts to ~22.5 g, or approximately one can of soda or sweetened beverage. The studies in animals used at minimum 20 % of fructose as a feeding paradigm. It is also possible that other dietary or metabolic factors mitigate the impact of fructose in middle-aged adults [26,27]. Additionally, given the evidence that estrogen may be protective against fructose-induced cardiac hypertrophy [28,29] it is plausible that estrogen's cardioprotective effects attenuated the association between fructose intake and HFpEF risk in the female participants. As the CARDIA cohort continues to age, future follow-up data may help to clarify whether fructose becomes a more significant risk factor over time.
While magnesium did not emerge as a protective factor against HFpEF, it was positively associated with higher e’ lateral velocities and LAVi, indicating better diastolic function. The lack of a protective effect against HFpEF aligns with findings from the Women's Health Initiative cohort study [15], which, to our knowledge, remains the only study addressing the association between magnesium and HFpEF. Existing research has focused on magnesium in the context of tertiary prevention, where low magnesium levels have been linked to an increased risk of heart failure events in HFpEF patients [30]. Moreover, magnesium supplementation has been shown to reduce all-cause mortality in critically ill HFpEF patients [31]. Interestingly, our tertiary analyses suggest that magnesium may have a cardioprotective effect on diastolic function.
Observed mid-life elevated blood pressure at Y20 (i.e. BP > 130/80 mmHg), irrespective of hypertension diagnosis, emerged as a strong predictor of HFpEF in all three HFpEF definitions, approximately doubling the risk, even after adjusting for potential protective factors. This finding aligns with the well-established link between hypertension and HFpEF and supports the critical role of managing mid-life hypertension to prevent diastolic dysfunction and HFpEF development. We would like to highlight that our finding places the emphasis on management of hypertension and bringing the BP below 130/80 mmHg, as we considered only the BP reading >130/80 mmHg, irrespective of hypertension diagnosis and/or current antihypertensive medication regimen.
The heterogeneity of HFpEF pathogenesis complicates the ability of both clinicians and researchers to accurately diagnose HFpEF and identify all risk factors [32]. Current AHA/ACC/HFSA guidelines require impaired filling pressures to diagnose HFpEF [2], although the ASE/EACVI guidelines provide a wide range of markers that can be used to diagnose diastolic dysfunction [17]. Our findings suggest that utilizing multiple HFpEF operationalizations in research may provide a more comprehensive understanding of risk and protective factors across the diverse pathophysiological pathways of HFpEF.
5. Strengths and limitations
In this study, we relied solely on objective echocardiographic measures of diastolic dysfunction to define HFpEF. This approach was necessary because the CARDIA study did not collect detailed symptom data or distinguish between preserved and reduced ejection fraction in self-reported heart failure diagnoses. Given the relatively low prevalence of symptomatic HFpEF in our middle-aged cohort, using only objective measures also likely allowed us to identify more participants with probable HFpEF, thereby improving statistical power for our analyses. We also recognize that the 2016 ASE/EACVI guidelines recommend four echocardiographic variables for identifying diastolic dysfunction (septal e', lateral e', average E/e', and LAVi, and peak TR velocity), and consider diastolic dysfunction present if more than half of the available parameters meet abnormal thresholds. We operationalized HFpEF using single echocardiographic parameters. Given the heterogeneity of HFpEF pathophysiology, we believe this variable-specific approach to be a strength as it allowed us to explore whether dietary components were differentially associated with specific diastolic abnormalities, rather than limiting our study to one composite definition. We omitted E/e' or TR velocity parameters in our HFpEF operationalizations because these parameters yielded too few cases for a powered analysis. We recognize that there are limitations to using e’ septal and e’ lateral velocities. Velocities decrease with age, and are influenced by comorbidities including mitral annular calcification and conduction abnormalities. We believe that our younger population and large sample size helps to mitigate some of these influences.
6. Conclusion and clinical implications
This retrospective cohort study found that increased BMI in adolescence and blood pressure higher than 130/80 mmHg in mid-life (irrespective of hypertension diagnosis) are significant predictors of HFpEF development in individuals in their fifties and beyond. We also found that increasing HDL and lowering LDL intake, may provide protective effects, lowering HFpEF risk and improving indices of diastolic function. Finally, impaired GLS emerged as a consistent predictor of increased HFpEF risk, and was found to be more impaired in the male population of the CARDIA cohort and in those with BP >130/80 mmHg, while none of the dietary predictors we included in our analysis have shown any effect size (Fig. 3). Our findings highlight a useful role of strain imaging with speckle tracking echocardiography as an early marker of subclinical disease which, in conjunction with lifestyle and nutritional counseling, may play a pivotal role in mitigating the risk of worsening HFpEF in aging populations.
Fig. 3.
Dietary and demographic predictors of HFpEF across the lifespan. Increased BMI in adolescence and blood pressure higher than 130/80 mmHg in mid-life (irrespective of hypertension diagnosis) are strong predictors of HFpEF development.
CRediT authorship contribution statement
Meaghan Osborne: Writing – review & editing, Writing – original draft, Formal analysis, Data curation, Conceptualization. Charlotte Turner: Writing – review & editing, Writing – original draft. Shaun Cardozo: Writing – review & editing, Writing – original draft. Dragana Komnenov: Writing – review & editing, Writing – original draft, Validation, Supervision, Resources, Project administration, Methodology, Formal analysis, Data curation, Conceptualization.
Ethics approval
-
1.
Ethical approval for this study was obtained from the IRB Review Board Wayne State University School of Medicine, Protocol number 063319MP2X.
-
2.
The Research Materials Distribution Agreement was obtained by the National Institutes of Health, Heart, Lung and Blood Institute which provided the de-identified data from the CARDIA study participants
Sources of funding
None.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
None.
References
- 1.Teramoto K., Teng T.-H.K., Chandramouli C., Tromp J., Sakata Y., Lam C.S. 2022. Epidemiology and clinical features of heart failure with preserved ejection fraction. Cardiac failure review. 8e27. Doi. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Heidenreich P.A., Bozkurt B., Aguilar D., Allen L.A., Byun J.J., Colvin M.M., et al. 2022 AHA/ACC/HFSA guideline for the management of heart failure: executive summary: a report of the American College of Cardiology/American Heart Association joint committee on clinical practice guidelines. J. Am. Coll. Cardiol. 2022;79(17):1757–1780. doi: 10.1016/j.cardfail.2022.02.010. [DOI] [PubMed] [Google Scholar]
- 3.Cilia L., Saeed A., Ganga H.V., Wu W.-C. Heart failure with preserved ejection fraction: prevention and management. Am. J. Lifestyle Med. 2019;13(2):182–189. doi: 10.1177/1559827617695219. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Rifai L., Silver M.A. A review of the DASH diet as an optimal dietary plan for symptomatic heart failure. Prog. Cardiovasc. Dis. 2016;58(5):548–554. doi: 10.1016/j.pcad.2015.11.001. doi. [DOI] [PubMed] [Google Scholar]
- 5.Asemi Z., Samimi M., Tabassi Z., Sabihi S.-s., Esmaillzadeh A. A randomized controlled clinical trial investigating the effect of DASH diet on insulin resistance, inflammation, and oxidative stress in gestational diabetes. Nutrition. 2013;29(4):619–624. doi: 10.1016/j.nut.2012.11.020. [DOI] [PubMed] [Google Scholar]
- 6.Rifai L., Pisano C., Hayden J., Sulo S., Silver M.A. Baylor University medical center proceedings; 2015: Taylor & Francis. 2015. Impact of the DASH diet on endothelial function, exercise capacity, and quality of life in patients with heart failure; pp. 151–156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Mozaffarian D., Appel L.J., Van Horn L. Components of a cardioprotective diet: new insights. Circulation. 2011;123(24):2870–2891. doi: 10.1161/circulationaha.110.968735. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Obarzanek E., Sacks F.M., Vollmer W.M., Bray G.A., Miller E.R., 3rd, Lin P.H., et al. Effects on blood lipids of a blood pressure-lowering diet: the dietary approaches to stop hypertension (DASH) trial. Am. J. Clin. Nutr. 2001;74(1):80–89. doi: 10.1093/ajcn/74.1.80. [DOI] [PubMed] [Google Scholar]
- 9.Yancy C.W. Sodium restriction in heart failure: too much uncertainty—do the trials. JAMA Intern. Med. 2018;178(12):1700–1701. doi: 10.1001/jamainternmed.2018.4653. [DOI] [PubMed] [Google Scholar]
- 10.Mahtani K.R., Heneghan C., Onakpoya I., Tierney S., Aronson J.K., Roberts N., et al. Reduced salt intake for heart failure: a systematic review. JAMA Intern. Med. 2018;178(12):1693–1700. doi: 10.1001/jamainternmed.2018.4673. [DOI] [PubMed] [Google Scholar]
- 11.Hummel S.L., Seymour E.M., Brook R.D., Sheth S.S., Ghosh E., Zhu S., et al. Low-sodium DASH diet improves diastolic function and ventricular–arterial coupling in hypertensive heart failure with preserved ejection fraction. Circ. Heart Fail. 2013;6(6):1165–1171. doi: 10.1161/CIRCHEARTFAILURE.113.000481. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Mofrad M.D., Djafarian K., Mozaffari H., Shab-Bidar S. Effect of magnesium supplementation on endothelial function: a systematic review and meta-analysis of randomized controlled trials. Atherosclerosis. 2018;27398–105 doi: 10.1016/j.atherosclerosis.2018.04.020. [DOI] [PubMed] [Google Scholar]
- 13.Liu M., Dudley S.C., Jr. Magnesium, oxidative stress, inflammation, and cardiovascular disease. Antioxidants. 2020;9(10) doi: 10.3390/antiox9100907. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Taveira T.H., Ouellette D., Gulum A., Choudhary G., Eaton C.B., Liu S., et al. Relation of magnesium intake with cardiac function and heart failure hospitalizations in Black adults: the Jackson heart study. Circ. Heart Fail. 2016;9(4) doi: 10.1161/CIRCHEARTFAILURE.115.002698. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wu W.C., Huang M., Taveira T.H., Roberts M.B., Martin L.W., Wellenius G.A., et al. Relationship between dietary magnesium intake and incident heart failure among older women: the WHI. J. Am. Heart Assoc. 2020;9(7) doi: 10.1161/JAHA.119.013570. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Liu M., Jeong E.-M., Liu H., Xie A., So E.Y., Shi G., et al. Magnesium supplementation improves diabetic mitochondrial and cardiac diastolic function. JCI Insight. 2019;4(1) doi: 10.1172/jci.insight.123182. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Nagueh S.F., Smiseth O.A., Appleton C.P., Byrd B.F., Dokainish H., Edvardsen T., et al. Recommendations for the evaluation of left ventricular diastolic function by echocardiography: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. Eur. J. Echocardiogr. 2016;17(12):1321–1360. doi: 10.1016/j.echo.2016.01.011. [DOI] [PubMed] [Google Scholar]
- 18.Mishra M., Muthuramu I., Aboumsallem J.P., Kempen H., De Geest B. Reconstituted HDL (Milano) treatment efficaciously reverses heart failure with preserved ejection fraction in mice. Int. J. Mol. Sci. 2018;19(11) doi: 10.3390/ijms19113399. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Komnenov D., Levanovich P.E., Perecki N., Chung C.S., Rossi N.F. 2020. Aortic stiffness and diastolic dysfunction in Sprague Dawley rats consuming short-term fructose plus high salt diet. Integrated blood pressure control; pp. 111–124. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Osborne M., Bernard A., Falkowski E., Peterson D., Vavilikolanu A., Komnenov D. Longitudinal associations of dietary fructose, sodium, and potassium and psychological stress with vascular aging index and incident cardiovascular disease in the CARDIA cohort. Nutrients. 2023;16(1) doi: 10.3390/nu16010127. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Komnenov D., Al-Hadidi M., Ali H., Al-Jamal M., Salami K., Shelbaya S., et al. Dietary fructose and sodium consumed during early mid-life are associated with hypertensive end-organ damage by late mid-life in the CARDIA cohort. Nutrients. 2024;16(7) doi: 10.3390/nu16070913. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Van Horn R. Linda, PhD, Slattery R. Marty, PhD, Hilner M. Joan, Bragg R. Charlotte, MS, Jacobs D., Jr., Gernhofer R. Niki, MS, et al. The cardia dietary history: development, implementation, and evaluation. J. Am. Diet. Assoc. 1991;91(9):1104–1112. doi. [PubMed] [Google Scholar]
- 23.Jacobs D.R., Hahn L.P., Haskell W.L., Pirie P., Sidney S. Validity and reliability of short physical activity history: CARDIA and the Minnesota heart health program. J. Cardiopulm. Rehabil. Prev. 1989;9(11):448–459. doi: 10.1097/00008483-198911000-00003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Pandey A., Patel K.V., Segar M.W., Shapiro M.D., Ballantyne C.M., Virani S.S., et al. Association of high-density lipoprotein parameters and risk of heart failure: a multicohort analysis. Heart Failure. 2024;12(7):1242–1253. doi: 10.1016/j.jchf.2024.03.007. [DOI] [PubMed] [Google Scholar]
- 25.Sasko B., Kelesidis T., Kostin S., Scharow L., Mueller R., Jaensch M., et al. Reduced antioxidant high-density lipoprotein function in heart failure with preserved ejection fraction. Clin. Res. Cardiol. 2025:1–9. doi: 10.1007/s00392-024-02583-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Park J.H., Ku H.J., Kim J.K., Park J.-W., Lee J.H. Amelioration of high fructose-induced cardiac hypertrophy by naringin. Sci. Rep. 2018;8(1) doi: 10.1038/s41598-018-27788-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Alves R., Suehiro C.L., Oliveira F.G.d., Frantz E.D.C., Medeiros R.F.d., Vieira R.d.P., et al. Aerobic exercise modulates cardiac NAD (P) H oxidase and the NRF2/KEAP1 pathway in a mouse model of chronic fructose consumption. J. Appl. Physiol. 2020;128(1):59–69. doi: 10.1152/japplphysiol.00201.2019. [DOI] [PubMed] [Google Scholar]
- 28.Bundalo M., Zivkovic M., Tepavcevic S., Culafic T., Koricanac G., Stankovic A. Fructose-rich diet-induced changes in the expression of the renin angiotensin system molecules in the heart of ovariectomized female rats could be reversed by estradiol. Horm. Metab. Res. 2015;47(07):521–527. doi: 10.1055/s-0034-1394373. [DOI] [PubMed] [Google Scholar]
- 29.Romic S., Tepavcevic S., Zakula Z., Milosavljevic T., Stojiljkovic M., Zivkovic M., et al. Does oestradiol attenuate the damaging effects of a fructose-rich diet on cardiac Akt/endothelial nitric oxide synthase signalling? Br. J. Nutr. 2013;109(11):1940–1948. doi: 10.1017/S0007114512004114. [DOI] [PubMed] [Google Scholar]
- 30.Nishihara T., Yamamoto E., Sueta D., Fujisue K., Usuku H., Oike F., et al. Clinical significance of serum magnesium levels in patients with heart failure with preserved ejection fraction. Medicine. 2019;98(38) doi: 10.1097/MD.0000000000017069. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Song L., Ying J., Li M., Ying L., Zhao C. Propensity score matched cohort study on magnesium supplementation and mortality in critically ill patients with HFpEF. Sci. Rep. 2025;15(1):1944. doi: 10.1038/s41598-025-85931-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Roh J., Hill J.A., Singh A., Valero-Muñoz M., Sam F. Heart failure with preserved ejection fraction: heterogeneous syndrome, diverse preclinical models. Circ. Res. 2022;130(12):1906–1925. doi: 10.1161/CIRCRESAHA.122.320257. [DOI] [PMC free article] [PubMed] [Google Scholar]