Abstract
OBJECTIVE
Limited data exist on the relation between long-term variability in blood lipid fractions and incident heart failure (HF) in the setting of type 2 diabetes mellitus (T2DM).
RESEARCH DESIGN AND METHODS
Among 9,443 participants with T2DM from the Action to Control Cardiovascular Risk in Diabetes (ACCORD) study, with lipid measurements available at six time points (baseline, 4, 8, 12, 24, and 36 months), we assessed variability in total cholesterol (TC), LDL cholesterol, HDL cholesterol, and triglycerides (TG) across visits, using coefficient of variation (CV), SD, and variability independent of the mean. Cox proportional hazards models were employed to estimate adjusted hazard ratios (HRs) for incident HF.
RESULTS
During a median follow-up of 5.0 years, 345 participants developed HF. Participants in the highest quartile of CV of TC had a 68% higher relative risk of HF compared with those in the lowest quartile (adjusted HR [aHR] 1.68, 95% CI 1.22–2.30). Similarly, those in the highest quartile of LDL cholesterol CV had a 76% higher relative risk (aHR 1.76, 95% CI 1.27–2.42) of HF, while those in the highest quartile of HDL cholesterol CV had a 53% higher risk (aHR 1.53, 95% CI 1.13–2.06). For TG CV, participants in the highest quartile had a 49% higher risk of HF compared with the lowest quartile (aHR 1.49, 95% CI 1.09–2.04). Similar patterns were observed for other variability metrics.
CONCLUSIONS
Increased variability in TC, LDL cholesterol, HDL cholesterol, or TG is independently associated with a higher HF risk among individuals with T2DM.
Graphical Abstract
Introduction
Heart failure (HF) is a growing public health concern, particularly among individuals with type 2 diabetes mellitus (T2DM), who face a substantially higher risk of developing HF compared with the general population (1–3). Among the numerous risk factors for HF, dyslipidemia has been postulated as a potentially significant contributor to myocardial dysfunction in the setting of diabetes (4–6). Elevated levels of lipid fractions, including total cholesterol (TC), LDL cholesterol, and triglycerides (TG), as well as low levels of HDL cholesterol, have been associated with adverse cardiovascular outcomes, including HF (7,8). However, most of the evidence linking lipid fractions to HF has been derived from studies measuring lipid levels at a single point in time (4–6), which may not fully capture the dynamic nature of lipid metabolism over time or the true cardiovascular risk conferred by fluctuations in lipid levels over time.
Recent studies have highlighted the potential importance of visit-to-visit variability (VVV) in lipid fractions as a marker of underlying metabolic instability or residual cardiovascular risk not captured by static measurements (9–17). Indeed, emerging evidence has suggested that greater VVV in lipid fractions is associated with an increased risk of adverse cardiovascular outcomes, such as coronary artery disease and stroke, independent of mean lipid levels (9–17). However, the relation between VVV of lipid fractions and the risk of HF remains understudied, especially in individuals with T2DM, who are at heightened risk for both dyslipidemia and HF, because of their unique metabolic and vascular profiles (3).
Understanding the association between lipid variability and the HF risk may add to our current understanding of lipid-related metabolic dysregulation and cardiovascular risk, particularly in the context of T2DM. While traditional lipid measures have offered valuable insights (4–6), the incorporation of VVV has the potential to further our understanding of metabolic dysregulation, and help refine risk stratification and treatment strategies.
This study aimed to evaluate the associations between VVV of lipid fractions and incident HF using data from the Action to Control Cardiovascular Risk in Diabetes (ACCORD) study. The ACCORD study offers a unique opportunity to investigate this association in a well-characterized cohort of adults with T2DM, where lipid levels were measured repeatedly over time (18). We hypothesized that greater VVV in lipid fractions (TC, LDL cholesterol, HDL cholesterol, and TG) would be associated with a higher risk of HF, independent of average lipid levels and other traditional cardiovascular risk factors.
Research Design and Methods
Study Design
We performed a secondary analysis of data from the ACCORD study, a large, multicenter, randomized controlled trial conducted between January 2001 and October 2005 at 77 sites across the U.S. and Canada (18). The trial enrolled 10,251 participants aged 40 to 79 years who either had established cardiovascular disease (CVD) or were aged 55 to 79 years with additional risk factors, including albuminuria, atherosclerosis, left ventricular hypertrophy, or at least two other CVD risk factors. Participants were randomly assigned to an intensive glucose-lowering intervention targeting an HbA1c level below 6%, or to standard therapy aiming for an HbA1c between 7.0 and 7.9%. ACCORD had a 2 × 2 factorial design study, including a lipid trial and blood pressure trial. A total of 5,518 participants were enrolled in the ACCORD lipid trial, which involved randomization to either simvastatin plus fenofibrate or simvastatin plus placebo (19). Meanwhile, 4,733 participants were assigned to the ACCORD blood pressure (BP) trial, where they were randomized to either an intensive BP control group with a systolic BP target of under 120 mmHg or a standard BP group with a target of under 140 mmHg. The full rationale and design of the ACCORD trial have been extensively described in prior publications (18,19).
For this analysis, we excluded participants with HF at baseline (n = 495) or missing data on levels of lipid fractions (n = 313). Supplementary Fig. 1 summarizes the study exclusion process. The study protocol was approved by the Institutional Review Board at all participating centers, and all participants provided written informed consent.
Assessment of Lipid Fractions Variability
A fasting plasma lipid profile was obtained at the ACCORD central laboratory during the baseline visit and subsequently at 4, 8, 12, 24, and 36 months following randomization. Levels of TC, LDL cholesterol, HDL cholesterol, and TG were measured using standardized enzymatic methods, as previously described (18–20).
To assess long-term variability in lipid fractions, three metrics were employed: 1) the coefficient of variation (CV) across the six measurement points; 2) intraindividual SD; and 3) variability independent of the mean (VIM), calculated using the formula 100*SD/meanβ, where β represents the regression coefficient derived from the natural logarithm of SD as a function of the natural logarithm of the mean lipid level. Multiple variability measures were included to comprehensively capture the range of lipid fluctuations.
Ascertainment of Incident HF Events
Incident HF events were identified during regular clinic visits every 4 months. At each visit, participants were asked about emergency room visits, hospitalizations, and any medical procedures that occurred since their last visit. If participants missed a clinic visit, study staff reached out by phone to collect information about potential events. The HF events were classified as hospitalizations due to HF or deaths attributed to “congestive heart failure,” with diagnosis supported by clinical and radiologic findings. These events were reviewed and confirmed by an expert adjudication committee. Participants were followed from the start of the study until the occurrence of HF, death, or study completion in June 2009 (18).
Covariates
The covariates, selected based on their relation with lipid variability and HF, included the following variables collected at baseline: age, sex, race, the eight-treatment assignment of the original ACCORD trial (including the assignment to the simvastatin and fenofibrate arms), smoking status, alcohol consumption, BMI, BP, diabetes duration, medical history, use of antihypertensive medications, history of atherosclerotic cardiovascular disease (ASCVD) (defined as a history of myocardial infarction, angina, stroke, or any coronary, carotid, or peripheral revascularization procedure), and estimated glomerular filtration rate (eGFR) calculated using the Modification of Diet in Renal Disease (MDRD) equation (21). Additionally, the following covariates were considered in the assessment of lipid variability associations: average systolic BP and average HbA1c, which were both calculated across visits that occurred at 4, 8, 12, 16, 20, 24, 28, 32, and 36 months.
Statistical Analyses
Participants were compared across quartiles of the CV for TC, LDL cholesterol, HDL cholesterol, and TG using the ANOVA or Kruskal-Wallis test for continuous variables, and the χ2 test for categorical variables.
Incidence rates per 1,000 person-years were determined by dividing the cumulative number of HF events by the total at-risk person-years, which were calculated as the sum of the follow-up time contributed by each participant from study baseline to either the date of HF diagnosis, death, loss to follow-up, or end of the study period, whichever came first. Cox proportional hazards regression was employed to calculate adjusted hazard ratios (HRs) and 95% CIs for incident HF. For all lipid fractions, each variability metric was modeled both as a continuous variable and in quartiles, with the lowest quartile serving as the reference.
The regression models to evaluate the association between lipid variability and HF risk were constructed in three stages: 1) the first model adjusted for age (continuous), sex, race/ethnicity (categorical), and randomization arm (categorical) (model 1); 2) the second model included all variables from model 1, with additional adjustments for BMI (continuous), current smoking status (categorical), alcohol consumption (categorical), use of antihypertensive medications (categorical), eGFR, duration of diabetes, average HbA1c (continuous), average systolic BP (continuous), and history of ASCVD (categorical) (model 2); 3) the third model included all variables from model 2 plus the use of lipid-lowering therapies (statin, fibrates and other lipid-lowering therapies—categorical), and the average value of each of the lipid fraction being analyzed (continuous)—average TC for TC variability, average LDL cholesterol for LDL cholesterol variability, average HDL cholesterol for HDL cholesterol variability, or average TG for TG variability (model 3). We evaluated the proportional hazards assumption using Schoenfeld residuals for all Cox proportional hazards models. The global tests for the proportional hazards assumption did not indicate any violations for any of the analyses (P > 0.05 for all models), confirming that the assumption was met across all covariates included in the models. A two-sided P value < 0.05 was considered statistically significant, and all analyses were conducted using STATA version 14.2 (StataCorp LLC, College Station, TX).
Results
Characteristics of Study Participants
The study sample included 9,443 participants with a mean (SD) age of 62.7 (6.6) years, of whom 3,613 (38.3%) were women, and 5,896 (62.4%) were White adults. Table 1 presents the characteristics of participants across quartiles of CV of TC. Participants in the highest quartile of CV of TC were relatively younger and more frequently women compared with those in the lower quartiles. They also had lower average levels of HDL cholesterol along with higher LDL cholesterol and TG levels. Additionally, individuals in the highest quartile of TC variability exhibited lower rates of baseline ASCVD. The baseline characteristics of participants according to the incidence of HF and to the quartiles of CV of LDL cholesterol, HDL cholesterol, and TG are provided in Supplementary Tables 1–4.
Table 1.
Characteristics of participants by TC variability
| CV of TC, quartiles | ||||||
|---|---|---|---|---|---|---|
| Characteristic | Total | <8.37 | 8.37–12.63 | 12.64–18.01 | >18.01 | P value |
| n | 9,443 | 2,361 | 2,361 | 2,361 | 2,360 | |
| At baseline | ||||||
| Age, mean (SD), years | 62.7 (6.6) | 63.0 (6.6) | 62.8 (6.6) | 62.5 (6.5) | 62.4 (6.5) | 0.002 |
| Sex | <0.001 | |||||
| Women | 3,613 (38.3) | 975 (41.3) | 841 (35.6) | 863 (36.6) | 934 (39.6) | |
| Men | 5,830 (61.7) | 1,386 (58.7) | 1,520 (64.4) | 1,498 (63.4) | 1,426 (60.4) | |
| Race and ethnicity | 0.028 | |||||
| White | 5,896 (62.4) | 1,471 (62.3) | 1,477 (62.6) | 1,488 (63.0) | 1,460 (61.9) | |
| Black | 1,768 (18.7) | 484 (20.5) | 452 (19.1) | 426 (18.0) | 406 (17.2) | |
| Hispanic | 680 (7.2) | 160 (6.8) | 157 (6.6) | 167 (7.1) | 196 (8.3) | |
| Other | 1,099 (11.6) | 246 (10.4) | 275 (11.6) | 280 (11.9) | 298 (12.6) | |
| Treatment arm | 0.28 | |||||
| Standard glycemic lowering | 4,730 (50.1) | 1,223 (51.8) | 1,174 (49.7) | 1,172 (49.6) | 1,161 (49.2) | |
| Intensive glycemic lowering | 4,713 (49.9) | 1,138 (48.2) | 1,187 (50.3) | 1,189 (50.4) | 1,199 (50.8) | |
| Current smoking | 1,308 (13.9) | 291 (12.3) | 323 (13.7) | 333 (14.1) | 361 (15.3) | 0.030 |
| Alcohol drinking | 2,302 (24.4) | 622 (26.3) | 574 (24.3) | 564 (23.9) | 542 (23.0) | 0.050 |
| Diabetes duration, median (IQR), years | 9.0 (5.0, 15.0) | 10.0 (5.0, 15.0) | 9.0 (5.0, 15.0) | 9.0 (5.0, 15.0) | 9.0 (5.0, 15.0) | 0.20 |
| BMI, mean (SD), kg/m2 | 32.2 (5.4) | 32.1 (5.4) | 32.2 (5.4) | 32.3 (5.3) | 32.1 (5.4) | 0.63 |
| eGFR, mean (SD), mL/min per 1.73 m2 | 91.5 (27.1) | 91.3 (24.8) | 91.6 (24.6) | 91.6 (25.5) | 91.5 (32.8) | 0.99 |
| Systolic BP, mean (SD), mm Hg | 136.3 (16.9) | 136.2 (16.5) | 135.1 (16.9) | 136.2 (16.8) | 137.8 (17.1) | <0.001 |
| Diastolic BP, mean (SD), mm Hg | 75.0 (10.5) | 74.6 (10.2) | 74.3 (10.6) | 75.2 (10.4) | 76.0 (10.8) | <0.001 |
| Use of BP-lowering drug | 7,841 (83.0) | 1,975 (83.7) | 1,980 (83.9) | 1,966 (83.3) | 1,920 (81.4) | 0.086 |
| History of ASCVD | 3,112 (33.0) | 763 (32.3) | 812 (34.4) | 787 (33.3) | 750 (31.8) | 0.23 |
| During variability assessment period | ||||||
| Average systolic BP, mean (SD), mm Hg | 129.4 (12.0) | 129.2 (11.7) | 129.0 (11.9) | 129.7 (12.1) | 129.7 (12.4) | 0.13 |
| Average diastolic BP, mean (SD), mm Hg | 70.3 (8.2) | 70.1 (8.1) | 70.0 (8.2) | 70.6 (8.0) | 70.7 (8.4) | 0.004 |
| Average HbA1c, mean (SD), % | 7.3 (0.8) | 7.3 (0.8) | 7.3 (0.8) | 7.3 (0.8) | 7.3 (0.9) | 0.032 |
| Average TC, mean (SD), mg/dL | 175.1 (35.2) | 173.0 (33.0) | 170.7 (33.2) | 174.2 (33.1) | 182.5 (39.9) | <0.001 |
| Average HDL cholesterol, mean (SD), mg/dL | 42.8 (11.2) | 45.4 (12.3) | 42.7 (10.9) | 41.9 (10.7) | 41.3 (10.3) | <0.001 |
| Average LDL cholesterol, mean (SD), mg/dL | 98.2 (27.0) | 97.6 (27.2) | 95.7 (25.7) | 97.8 (26.1) | 101.6 (28.5) | <0.001 |
| Average TG, median (IQR), mg/dL | 147.0 (104.2, 210.3) | 133.5 (95.0, 185.0) | 143.0 (100.7, 201.7) | 152.3 (109.7, 216.7) | 161.9 (114.7, 244.5) | <0.001 |
Data are n (%) unless otherwise indicated. IQR, interquartile range.
Variability of Lipid Fractions and Incidence of HF
Over a median follow-up period of 5.0 years (interquartile range 4.0–5.7), 345 participants developed incident HF events (incidence rate per 1,000 person-years: 7.5 [95% CI 6.8–8.4]). The cumulative incidence of HF was higher among participants with higher levels of variability for all four of the lipid fractions (Supplementary Fig. 2).
Total Cholesterol
The adjusted HRs for the variability metrics of all four lipid fractions are presented in Table 2 and Supplementary Tables 5 and 6. After multivariable adjustment, the HRs for incident HF per SD increase in the CV, intraindividual SD, and VIM for TC were 1.23 (95% CI 1.12–1.36), 1.21 (95% CI 1.09–1.34), and 1.24 (95% CI 1.12–1.36), respectively. Participants in the highest quartile of TC CV had a 68% higher risk of incident HF compared with those in the lowest quartile (HR 1.68, 95% CI 1.22–2.30). The HRs for incident HF associated with the highest quartiles of TC SD and VIM were 1.83 (95% CI 1.31–2.55) and 1.58 (95% CI 1.16–2.16), respectively.
Table 2.
Rates and HRs for incident HF by variability of lipid fractions
| Measure of variability | CV of lipid fractions, quartiles | P trend | Per 1-SD increment | |||
|---|---|---|---|---|---|---|
| CV of TC, % | <8.37 | 8.37–12.63 | 12.64–18.01 | >18.01 | ||
| No. of events/no. at risk | 64/2,361 | 82/2,361 | 94/2,361 | 105/2,360 | 345/9,443 | |
| Rate/1,000 person-years | 5.6 (4.3–7.1) | 7.1 (5.7–8.8) | 8.2 (6.7–10.0) | 9.4 (7.7–11.3) | 7.5 (6.8–8.4) | |
| Model 1 | Reference | 1.24 (0.89–1.72) | 1.45 (1.05–2.00)* | 1.75 (1.28–2.39)‡ | <0.001 | 1.25 (1.13–1.37)‡ |
| Model 2 | Reference | 1.20 (0.86–1.66) | 1.38 (1.00–1.91) | 1.70 (1.24–2.34)† | <0.001 | 1.24 (1.13–1.37)‡ |
| Model 3 | Reference | 1.19 (0.86–1.66) | 1.37 (0.99–1.90) | 1.68 (1.22–2.30)† | 0.001 | 1.23 (1.12–1.36)‡ |
| CV of LDL cholesterol, % | <12.55 | 12.55–18.90 | 18.91–27.16 | >27.16 | ||
| No. of events/no. at risk | 61/2,361 | 97/2,361 | 88/2,361 | 99/2,360 | 345/9,443 | |
| Rate/1,000 person-years | 5.3 (4.1–6.8) | 8.5 (6.9–10.3) | 7.6 (6.2–9.4) | 8.9 (7.3–10.8) | 7.5 (6.8–8.4) | |
| Model 1 | Reference | 1.57 (1.14–2.17)† | 1.42 (1.02–1.97)* | 1.74 (1.26–2.40† | 0.003 | 1.16 (1.04–1.28)† |
| Model 2 | Reference | 1.48 (1.07–2.04)* | 1.35 (0.97–1.88) | 1.75 (1.27–2.41)† | 0.002 | 1.16 (1.05–1.28)† |
| Model 3 | Reference | 1.48 (1.07–2.04)* | 1.36 (0.97–1.89) | 1.76 (1.27–2.42)† | 0.002 | 1.16 (1.05–1.29† |
| CV of HDL cholesterol, % | <6.72 | 6.72–9.48 | 9.49–13.12 | >13.12 | ||
| No. of events/no. at risk | 71/2,362 | 73/2,361 | 86/2,360 | 115/2,360 | 345/9,443 | |
| Rate/1,000 person-years | 6.2 (4.9–7.8) | 6.3 (5.0–8.0) | 7.5 (6.1–9.2) | 10.2 (8.5–12.2) | 7.5 (6.8–8.4) | |
| Model 1 | Reference | 1.02 (0.74–1.42) | 1.20 (0.88–1.65) | 1.64 (1.22–2.20)† | <0.001 | 1.16 (1.08–1.25)‡ |
| Model 2 | Reference | 0.99 (0.71–1.37) | 1.17 (0.85–1.61) | 1.57 (1.16–2.11)† | 0.001 | 1.20 (1.10–1.30)‡ |
| Model 3 | Reference | 0.97 (0.70–1.35) | 1.16 (0.84–1.60) | 1.53 (1.13–2.06)† | 0.002 | 1.14 (1.05–1.24)† |
| CV of TG Level, % | <18.40 | 18.40–26.10 | 26.11–35.91 | >35.91 | ||
| No. of events/no. at risk | 72/2,361 | 75/2,361 | 91/2,361 | 107/2,360 | 345/9,443 | |
| Rate/1,000 person-years | 6.3 (5.0–8.0) | 6.5 (5.2–8.1) | 7.9 (6.4–9.7) | 9.5 (7.8–11.4) | 7.5 (6.8–8.4) | |
| Model 1 | Reference | 1.00 (0.72–1.38) | 1.21 (0.89–1.65) | 1.55 (1.14–2.09)† | 0.002 | 1.16 (1.05–1.28)† |
| Model 2 | Reference | 1.03 (0.74–1.43) | 1.23 (0.90–1.69) | 1.65 (1.22–2.24)† | <0.001 | 1.18 (1.07–1.30)† |
| Model 3 | Reference | 1.02 (0.73–1.41) | 1.20 (0.88–1.64) | 1.49 (1.09–2.04)* | 0.007 | 1.12 (1.01–1.24)* |
Data are HRs (95% CI) unless otherwise indicated. Model 1 was adjusted for age, sex, race/ethnicity, and randomization arm; model 2 included model 1 plus BMI, current smoking status, alcohol consumption, use of antihypertensive medications, eGFR, duration of diabetes, average HbA1c, average systolic BP, and history of ASCVD; model 3 included model 2 plus the use of lipid-lowering therapies and the average value of the respective lipid fraction being analyzed—average TC for TC variability, average LDL cholesterol for LDL cholesterol variability, average HDL cholesterol for HDL cholesterol variability, or average TG for TG variability. *P < 0.05, †P < 0.01, ‡P < 0.001.
LDL Cholesterol
The adjusted HRs for incident HF per SD increase in the CV, SD, and VIM for LDL cholesterol were 1.16 (95% CI 1.05–1.29), 1.17 (95% CI 1.04–1.31), and 1.16 (95% CI 1.04–1.28), respectively. Participants in the highest quartile of LDL cholesterol CV had a 76% higher risk of HF compared with those in the lowest quartile (HR 1.76, 95% CI 1.27–2.42). The HRs for HF in the highest quartiles of LDL cholesterol SD and VIM were 1.38 (95% CI 0.98–1.93) and 1.68 (95% CI 1.22–2.31), respectively.
HDL Cholesterol
The HRs for incident HF per SD increase in the CV, SD, and VIM of HDL cholesterol were 1.14 (95% CI 1.05–1.24), 1.22 (95% CI 1.10–1.36), and 1.16 (95% CI 1.06–1.26), respectively. Comparing the highest to the lowest quartiles, the HRs for HF were 1.53 (95% CI 1.13–2.06) for CV, 1.71 (95% CI 1.24–2.34) for SD, and 1.70 (95% CI 1.26–2.30) for VIM of HDL cholesterol.
Triglycerides
The adjusted HRs for incident HF per SD increase in CV, SD, and VIM of TG were 1.12 (95% CI 1.01–1.24), 0.98 (95% CI 0.86–1.13), and 1.13 (95% CI 1.01–1.25), respectively. Comparing the top to the bottom quartiles, the HRs for HF were 1.49 (95% CI 1.09–2.04) for CV, 1.93 (95% CI 1.31–2.84) for SD, and 1.44 (95% CI 1.06–1.95) for VIM of TG.
Conclusions
In this study, we found that higher long-term variability in any of the four lipid fractions—TC, LDL cholesterol, HDL cholesterol, and TG—was significantly associated with an increased relative risk of HF among adults with T2DM. These associations persisted even after adjusting for mean lipid levels, allocation to combined lipid therapy (simvastatin and fenofibrate), and other conventional CVD risk factors, suggesting independent prognostic implications of lipid variability. Importantly, this association was robust across various measures of variability, including CV, SD, and VIM. Our findings add to the body of evidence on the clinical importance of lipid fractions, and particularly their variability, for assessing HF risk, and highlight the importance of more uniform and less variable lipids levels over time.
Our study expands on the findings from previous studies that have also shown an association between various lipids fractions (non-HDL cholesterol, high TG, low HDL cholesterol, or high TC/HDL cholesterol ratio) and the incidence of HF (4–6). These studies only assessed lipid fractions at a single time points, and thus could not capture the variations in lipid fractions over time. Our results are also consistent with previous reports on a positive association between higher VVV in various lipid fractions and an increased risk of several ASCVD outcomes and all-cause mortality (9–17). However, these studies primarily focused on ASCVD outcomes, with HF rarely or not examined as a primary end point. In a recent retrospective study from Hong Kong, higher variability of HDL cholesterol and TC was associated with an increased risk of HF (22); however, this prior research did not specifically focus on individuals with T2DM. Our study is unique in its explicit focus on HF risk among individuals with T2DM and the examination of risk associations prospectively within a racially/ethnically diverse population.
Several mechanisms may explain the relationship between higher lipid variability and increased HF risk in adults with T2DM. One potential mechanism is lipid-induced cardiac stress and lipotoxicity (23–26). Indeed, a greater variability in lipid levels, particularly TG and LDL cholesterol, may cause intermittent lipid overloading in cardiomyocytes, leading to toxic lipid accumulation (e.g., ceramides) resulting in mitochondrial dysfunction and impaired energy production (26). Fluctuating lipid levels may disrupt the balance of lipid oxidation and energy production in the heart (23–26). Periods of high lipid availability can lead to excess fatty acid intermediates, while low lipid levels can create energy shortfalls, causing metabolic stress in cardiomyocytes and contributing to HF. Another potential mechanism is via inflammation and endothelial dysfunction (23–26). During periods of elevated LDL cholesterol or TC, circulating levels of oxidized LDL cholesterol may increase, promoting vascular inflammation and endothelial dysfunction. Endothelial dysfunction impairs nitric oxide bioavailability, leading to reduced vasodilation and increased vascular stiffness, both of which increase afterload on the heart and contribute to HF (23–26). Variations in lipid levels may also lead to cycles of oxidative stress, causing damage to cardiomyocytes. These fluctuations drive mitochondrial dysfunction, apoptosis, and adverse remodeling in the heart. Periodic lipid level changes, particularly TG, can increase sympathetic nervous system activity, promoting arrhythmias and ventricular remodeling, both of which contribute to HF (23–26).
Our findings have several important implications for clinical practice, public health, and future research. The results suggest that, beyond achieving optimal lipid levels, maintaining stable lipid profiles over time could be critical in reducing HF risk, particularly in high-risk populations such as those with T2DM. This highlights the potential benefits of more frequent lipid monitoring to detect and address excessive fluctuations. Additionally, clinicians should emphasize the importance of medication adherence and consistent lifestyle modifications to minimize lipid variability, potentially reducing HF risk. Our results on the HF risk associated with lipid variability in the context of ongoing lipid therapy suggest that this likely reflects a certain degree of underlying metabolic instability. Hence, targeting the underlying metabolic abnormalities (through a reduction of insulin resistance using lifestyle change, glucagon-like peptide 1 receptor agonists, etc.), in addition to lipid-lowering therapies may help prevent HF.
Future research should explore whether interventions that specifically reduce lipid variability can lower HF risk in this population. Additionally, it would be beneficial to investigate whether different lipid-lowering medications vary in their ability to stabilize lipid levels and whether this stabilization translates into a reduction in HF incidence. Mechanistic studies are warranted to further elucidate the pathways through which lipid variability influences HF risk.
Our findings should be interpreted within the context of several limitations. First, our study was observational, and there is a possibility of unmeasured residual confounding. Second, the ACCORD study did not collect data on left ventricular ejection fraction or HF subtype data, thus precluding an examination of the effect of lipid variability on the risk of HF with reduced ejection fraction versus HF with preserved ejection fraction. Likewise, we did not have data on apolipoprotein levels, which have also been shown to be associated with increased HF risk (27). We also did not have data on the extent to which the intermittent adherence to drugs, and the modification of therapeutic regimen, may influence lipid variability. Finally, given that our study relied on only six time points to assess lipid variability, we may have underestimated the variability and consequently the magnitude of our effect estimates. Prior studies have established that VVV of a biological variable increases with the number of visits used to calculate it (28).
Despite the aforementioned limitations, our study has several strengths, including the use of a large and diverse prospective cohort (including White, Black, and Hispanic individuals), the assessment of lipid measurements at predetermined regular intervals for all participants, the inclusion of several indices of lipid variability, the ascertainment of outcomes following a standardized protocol, the accounting for mean lipid levels over the study period, and a relatively long duration of follow-up.
In conclusion, our study demonstrated that higher VVV in lipid fractions—TC, LDL cholesterol, HDL cholesterol, and TG—was independently associated with an increased risk of incident HF in adults with T2DM. These findings underscore the importance of not only targeting optimal lipid levels but also maintaining consistent control of lipid profiles over time to mitigate HF risk in this high-risk population.
This article contains supplementary material online at https://doi.org/10.2337/figshare.28641857.
Article Information
Acknowledgments. The authors thank the staff and participants of the ACCORD Study for their valuable contributions.
J.B.E.-T. is an editor of Diabetes Care but was not involved in any of the decisions regarding review of the manuscript or its acceptance.
The views expressed in this manuscript are those of the authors and do not necessarily represent the views of the National Heart, Lung, and Blood Institute or the National Institutes of Health.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. A.D.K. and J.B.E.-T. drafted the manuscript and performed statistical analysis. All authors contributed to critical revisions of the manuscript for important intellectual content. J.B.E.-T. conceived of and designed the study, obtained funding, provided administrative, technical, or material support, and supervised. J.E.B.-T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Handling Editors. The journal editor responsible for overseeing the review of the manuscript was Frank B. Hu.
Funding Statement
J.B.E.-T. is funded by National Institutes of Health/National Heart, Lung, and Blood Institute grant K23 HL153774.
Supporting information
References
- 1. Dei Cas A, Khan SS, Butler J, et al. Impact of diabetes on epidemiology, treatment, and outcomes of patients with heart failure. JACC Heart Fail 2015;3:136–145 [DOI] [PubMed] [Google Scholar]
- 2. Aune D, Schlesinger S, Neuenschwander M, et al. Diabetes mellitus, blood glucose and the risk of heart failure: a systematic review and meta-analysis of prospective studies. Nutr Metab Cardiovasc Dis 2018;28:1081–1091 [DOI] [PubMed] [Google Scholar]
- 3. Oktay AA, Paul TK, Koch CA, Lavie CJ.. Diabetes, cardiomyopathy, and heart failure. In Endotext. Feingold KR, Anawalt B, Blackman MR, et al., Eds. South Dartmouth, MA, MDText.com, 2000. Accessed 26 September 2023. Available from https://www.ncbi.nlm.nih.gov/books/NBK560257/ [Google Scholar]
- 4. Velagaleti RS, Massaro J, Vasan RS, Robins SJ, Kannel WB, Levy D.. Relations of lipid concentrations to heart failure incidence: the Framingham Heart Study. Circulation 2009;120:2345–2351 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Ebong IA, Goff DC, Rodriguez CJ, Chen H, Sibley CT, Bertoni AG.. Association of lipids with incident heart failure among adults with and without diabetes mellitus: Multiethnic Study of Atherosclerosis. Circ Heart Fail 2013;6:371–378 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Dhingra R, Sesso HD, Kenchaiah S, Gaziano JM.. Differential effects of lipids on the risk of heart failure and coronary heart disease: the Physicians' Health Study. Am Heart J 2008;155:869–875 [DOI] [PubMed] [Google Scholar]
- 7. Arnett DK, Blumenthal RS, Albert MA, et al. 2019 ACC/AHA guideline on the primary prevention of cardiovascular disease: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Circulation 2019;140:e596–e646 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Di Angelantonio E, Sarwar N, Perry P, et al.; Emerging Risk Factors Collaboration . Major lipids, apolipoproteins, and risk of vascular disease. JAMA 2009;302:1993–2000 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Kim MK, Han K, Park Y-M, et al. Associations of variability in blood pressure, glucose and cholesterol concentrations, and body mass index with mortality and cardiovascular outcomes in the general population. Circulation 2018;138:2627–2637 [DOI] [PubMed] [Google Scholar]
- 10. Manemann SM, Bielinski SJ, Moser ED, et al. Variability in lipid levels and risk for cardiovascular disease: an electronic health record-based population cohort study. J Am Heart Assoc 2023;12:e027639. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Kim MK, Han K, Kim H-S, et al. Cholesterol variability and the risk of mortality, myocardial infarction, and stroke: a nationwide population-based study. Eur Heart J 2017;38:3560–3566 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Bangalore S, Breazna A, DeMicco DA, Wun C-C, Messerli FH, TNT Steering Committee and Investigators . Visit-to-visit low-density lipoprotein cholesterol variability and risk of cardiovascular outcomes: insights from the TNT trial. J Am Coll Cardiol 2015;65:1539–1548 [DOI] [PubMed] [Google Scholar]
- 13. Clark D, Nicholls SJ, St John J, et al. Visit-to-visit cholesterol variability correlates with coronary atheroma progression and clinical outcomes. Eur Heart J 2018;39:2551–2558 [DOI] [PubMed] [Google Scholar]
- 14. Cao Y-X, Li L, Zhang H-W, et al. Visit-to-visit variability of lipid and cardiovascular events in patients with familial hypercholesterolemia. Ann Transl Med 2021;9:556. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Waters DD, Bangalore S, Fayyad R, et al. Visit-to-visit variability of lipid measurements as predictors of cardiovascular events. J Clin Lipidol 2018;12:356–366 [DOI] [PubMed] [Google Scholar]
- 16. Liu X, Wu S, Song Q, Wang X.. Visit-to-visit variability of lipid measurements and the risk of myocardial infarction and all-cause mortality: a prospective cohort study. Atherosclerosis 2020;312:110–116 [DOI] [PubMed] [Google Scholar]
- 17. Bangalore S, Fayyad R, Messerli FH, et al. Relation of variability of low-density lipoprotein cholesterol and blood pressure to events in patients with previous myocardial infarction from the IDEAL trial. Am J Cardiol 2017;119:379–387 [DOI] [PubMed] [Google Scholar]
- 18. Buse JB, Bigger JT, Byington RP, et al.; ACCORD Study Group . Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial: design and methods. Am J Cardiol 2007;99:21i–33i [DOI] [PubMed] [Google Scholar]
- 19. Ginsberg HN, Elam MB, Lovato LC, et al.; ACCORD Study Group . Effects of combination lipid therapy in type 2 diabetes mellitus. N Engl J Med 2010;362:1563–1574 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Warnick GR. Enzymatic methods for quantification of lipoprotein lipids. Methods Enzymol 1986;129:101–123 [DOI] [PubMed] [Google Scholar]
- 21. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D.. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 1999;130:461–470 [DOI] [PubMed] [Google Scholar]
- 22. Chan JSK, Satti DI, Lee YHA, et al. High visit-to-visit cholesterol variability predicts heart failure and adverse cardiovascular events: a population-based cohort study. Eur J Prev Cardiol 2022;29:e323–e325 [DOI] [PubMed] [Google Scholar]
- 23. Da Dalt L, Cabodevilla AG, Goldberg IJ, Norata GD.. Cardiac lipid metabolism, mitochondrial function, and heart failure. Cardiovasc Res 2023;119:1905–1914 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Schulze PC, Drosatos K, Goldberg IJ.. Lipid use and misuse by the heart. Circ Res 2016;118:1736–1751 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Wende AR, Abel ED.. Lipotoxicity in the heart. Biochim Biophys Acta 2010;1801:311–319 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Goldberg IJ, Trent CM, Schulze PC.. Lipid metabolism and toxicity in the heart. Cell Metab 2012;15:805–812 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Ingelsson E, Arnlöv J, Sundström J, Zethelius B, Vessby B, Lind L.. Novel metabolic risk factors for heart failure. J Am Coll Cardiol 2005;46:2054–2060 [DOI] [PubMed] [Google Scholar]
- 28. Levitan EB, Kaciroti N, Oparil S, Julius S, Muntner P.. Blood pressure measurement device, number and timing of visits, and intra-individual visit-to-visit variability of blood pressure. J Clin Hypertens (Greenwich) 2012;14:744–750 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.