HDL Levels as a Novel Predictor of Long-Term Adverse Outcomes in Patients with Heart Failure: A Retrospective Cohort Study

Introduction

Cardiovascular disease (CVD) is the leading cause of death globally, with Heart Failure (HF) emerging as a significant terminal event within this spectrum.¹ HF is a complex and treacherous syndrome, marked by elevated illness and mortality rates, diminished daily functioning and life satisfaction, and imposing a considerable economic burden on healthcare systems.

Geographical differences shape the epidemiology of HF. While it seems to have plateaued or may be declining in developed countries, an aging population, advancements in coronary artery disease interventions, and the use of life-extending treatments have contributed to a global increase in HF cases.² An analysis of medical insurance data from 50 million urban workers in China suggests a substantial HF patient population of 12.05 million, with an annual rise of approximately 2.97 million new patients.³ The Chinese Heart Failure Patient Registration Study (China-HF) reports a 4.1% mortality rate among hospitalized HF patients,⁴ highlighting the strain HF places on China’s public health system and the urgent need for prevention strategies.

Effective strategies include the early identification of at-risk populations and timely interventions for modifiable risk factors, which are crucial for slowing HF progression, improving patient longevity, and reducing the incidence, severity, and mortality rates associated with HF.

A well-established body of evidence links blood lipoproteins to HF onset. High-density lipoprotein particles (HDL-P) in the bloodstream are known to influence processes such as atherosclerosis, inflammation, and endothelial health, all of which are integral to HF pathogenesis. HDL cholesterol (HDL-C) has long been recognized for its cardioprotective potential, with numerous epidemiological studies showing that each 1 mg/dL increase in HDL-C levels corresponds to a 2 to 3 percentage point decrease in cardiovascular mortality rates.^5–10

However, recent research in epidemiology and genetics indicates that HDL-C levels may not uniformly predict heart-related health issues in every case.^11–13 The precise mechanisms by which HDL-C is associated with different cardiovascular events remain incompletely understood, particularly for high-risk individuals such as HF patients. Our review focuses on a cohort of 4442 individuals diagnosed with HF, with long-term monitoring to understand how HDL-C levels are related to disease progression and treatment outcomes.

Methods

Methodological Approach and Participant Selection

We conducted a comprehensive review of medical records from 4442 individuals hospitalized for HF at the First Affiliated Hospital of Xinjiang Medical University, covering a period of ten years from July 2012. Our study, a substantial retrospective cohort analysis, included patients who met the established HF diagnostic criteria¹⁴ and had complete clinical records. The study adhered to the Declaration of Helsinki, and the research protocol was approved by the hospital’s Ethics Committee (ethics number 202207–019). Given the study’s retrospective nature, informed consent from participants was waived, as per the Ethics Committee’s approval. Additional study design details can be found under the identifier NCT06092658 on the ClinicalTrials.gov registry. Participants were included if they were 18 years or older and fulfilled the HF diagnostic standards outlined in the 2018 Chinese Guidelines for the Diagnosis and Treatment of Heart Failure.¹⁴ Participants had to have a minimum HF diagnosis duration of 3 months to be eligible for the study. Informed consent was required for participant enrollment. We excluded individuals with late-stage cancers, significant organ impairment, immune system disorders, and blood-related conditions, as well as those with a familial predisposition to psychiatric disorders or a history of recent (within three months) surgical intervention or severe injury. After applying these criteria, the study enrolled 3203 individuals with available HDL-C measurements. The standard flow chart for discharge is depicted in Figure 1.

Figure 1 Flowchart depicting the patient selection process for the study on long-term mortality and main adverse cardiovascular events in heart failure (HF) patients categorized by high-density lipoprotein cholesterol (HDL-C) levels.

Gathering Clinical Information and Establishing Outcome Measures

We compiled a comprehensive dataset encompassing demographic details, cardiovascular risk factors, and laboratory values from all enrolled patients, including hematologic and chemical blood tests. Noted risk factors included sex, age, chronic conditions such as hypertension and diabetes, and details of pharmaceutical interventions. Coronary heart disease (CHD) was classified according to the American Heart Association criteria, based on a confirmed diagnosis through prior angiographic evidence. Hypertension was identified in participants with a recorded medical history and ongoing anti-hypertensive therapy or consistent elevated blood pressure recordings (≥140/90 mmHg) on multiple visits.¹⁵ Diabetes was defined by a documented diagnosis and therapy with blood sugar-lowering medications, confirmed by elevated blood glucose thresholds: a fasting measurement >7.1 mmol/L and a postprandial level ≥11.1 mmol/L two hours after meals.¹⁶

For HDL-C level measurement, participants were required to fast for 6 to 9 hours prior to the procedure. Venous blood was then drawn and analyzed using enzymatic assays with magnesium phosphodiesterase and a method combining rapid centrifugation with targeted precipitation.

For HDL-C Level Measurement

To ensure accurate measurement of high-density lipoprotein cholesterol (HDL-C) levels, participants were required to fast for 6 to 9 hours prior to the blood draw. Venous blood samples were collected in a standardized procedure and processed using the phosphodiesterase magnesium enzyme method, which facilitates the separation of HDL particles through a combination of rapid centrifugation and targeted precipitation techniques.The HDL-C levels were measured using the I2000 automated chemiluminescence immunoassay analyzer (Abbott Laboratories, USA), a device known for its precision and reliability in clinical chemistry testing. The detection kit, procured from Fujian New Continent Biotechnology Co., Ltd., was utilized according to the manufacturer’s instructions to ensure the accuracy of the results.It is important to note that the specific testing procedure should adhere strictly to the guidelines provided in the reagent kit’s instruction manual. The reagent kit, which includes quality control serum or standards, was used to perform quality control measures throughout the testing process, ensuring the reliability and consistency of our measurements.

Follow-Up

This retrospective cohort study tracked individuals diagnosed with HF, initiating contact at the study’s outset and maintaining it at 6-month intervals up to 36 months, with additional assessments as required. The study aimed to monitor patients for at least one year, focusing on the incidence of significant cardiovascular and cerebrovascular events and the likelihood of hospital readmission. Data collected during these intervals were systematically submitted for analysis, essential for ongoing patient surveillance post-discharge and for identifying and monitoring adverse medical events. Comprehensive follow-up was ensured using outpatient consultations, telecommunication, and alternative contact methods until a critical event occurred or the study concluded. The study duration ranged from a median of 22.75 months, with an interquartile range of 12.37 to 47.11 months, and was designed to assess all-cause mortality (ACM).

Statistical Analysis

Data analyses were performed using SPSS 26.0. Continuous variables are presented as mean ± standard deviation (SD), while categorical variables are expressed as numbers (percentages). The normality of continuous variable distributions was assessed with the Kolmogorov–Smirnov test. Parametric attributes were analyzed using one-way ANOVA, and non-parametric attributes were evaluated with the Kruskal–Wallis test. Categorical data were analyzed using the chi-square (χ²) test. Statistical significance was set at P < 0.05. For continuous variables, the specific statistical test applied—either one-way ANOVA or the non-parametric equivalent—is indicated by the χ²/F/H value column. The rank sum test was used for non-parametric data. The term “n” refers to the number of participants in each quartile group.

The Cox proportional hazards model was applied in both univariate and multivariate settings to identify determinants associated with ACM. In the univariate analysis, potential predictors were assessed, and variables found to be statistically significant (P < 0.05) were included in the multivariate model for further analysis. Hazard ratios (HR) and 95% confidence intervals (CI) were calculated to quantify the associations. The Kaplan-Meier method was used to estimate survival probabilities for ACM, and the Log rank test was applied to compare survival curves between groups. Significance was set at P < 0.05.

Results

Baseline Characteristics Across Four HDL-C Categories

Enrolling a total of 3203 HF patients, our study comprised 2071 males (64.7%) and 1132 females (35.5%), with an average age of 64.18 ± 13.69 years. The follow-up duration for participants ranged from 1 to 10 years. Patients were categorized into quartiles based on plasma HDL-C levels: Q1 ≤ 0.75 mmol/L, Q2 from 0.76 to 0.93 mmol/L, Q3 from 0.94 to 1.14 mmol/L, and Q4 > 1.14 mmol/L.Comprehensive data encompassing clinical, echocardiographic, and laboratory findings for these participants are detailed in Table 1.

Table 1 Clinical and Laboratory Characteristics

Table 2 presents the participants’ baseline characteristics, laboratory parameters, and diagnostic indicators in a clear and logical manner. One-way ANOVA was used to identify significant inter-quartile differences among variables. There were statistically significant differences in age, HDL-C, Low-Density Lipoprotein Cholesterol (LDL-C), total cholesterol, triglycerides, hemoglobin, apolipoprotein AI (Apo AI), apolipoprotein B (Apo B), C-reactive protein (CRP), total bilirubin, albumin/globulin ratio (A/G ratio), and left ventricular ejection fraction (LVEF) (all P < 0.05). Significant differences were also found in the medical histories of gender, coronary heart disease, hypertension, diabetes, and chronic obstructive pulmonary disease (COPD) by chi-square tests (all P < 0.001). The rank sum test revealed statistically significant differences for creatinine and N-terminal pro-brain natriuretic peptide (NT-proBNP) (P < 0.05).

Table 2 Baseline Characteristics and Laboratory Parameters by HDL-Cholesterol Quartiles

The gender distribution across HDL-C quartiles showed an inverted U-shaped pattern for males and an ascending trend for females. Notably, individuals in higher HDL-C quartiles more frequently exhibited elevated levels in age, LDL-C, total cholesterol (TC), Apo AI, Apo B, A/G ratio, and LVEF.

Our cohort had a notably higher prevalence of male HF patients. However, no statistically significant differences were observed across the quartiles for uric acid, white blood cell and platelet counts, estimated glomerular filtration rate (eGFR), total protein, or the prevalence of certain health conditions such as congenital heart disease, valvular disease, myocardial disease, and arrhythmia (P > 0.05).

Drug therapy analysis revealed notable differences in medication usage among heart failure patients categorized by HDL cholesterol levels. β-blockers, Sodium-Glucose Transporter 2 (SGLT-2) inhibitors, Angiotensin-Converting Enzyme (ACE) inhibitors or Angiotensin Receptor Neprilysin Inhibitors (ARNI), and Mineralocorticoid Receptor Antagonist (MRA) each demonstrated significant prevalence, with all P-values < 0.001, indicating their importance in treatment protocols. In contrast, the prevalence of other drugs did not reach statistical relevance and is detailed in Table 3.

Table 3 Distribution of Medications in Our Population

Comparison of Different Clinical Outcomes

In an extended follow-up spanning over a decade, there were 1354 all-cause fatalities, accounting for 42.3% of the cases, and readmission occurred in 820 instances, representing 28.7% of the cases. Table 4 presents a comparative analysis of the unfavorable outcomes across patient groups, with statistically significant differences observed in both all-cause mortality and readmission rates for each of the four groups (P < 0.05).

Table 4 Clinical Outcomes Distribution Across HDL Cholesterol Levels

The ACM rate decreased from 55.4% in Group 1 to 36.5% in Group 2, and then rose to 39.6% in Group 4, with a significant difference between the two (P=0.000). Conversely, the readmission rate slightly increased from 24.3% in Group 1 to 31.0% in Group 2, with the last group at 30.6%, which also indicated significance (P=0.030). Analysis using spline functions revealed the continuous risk ratio of HDL-C in relation to mortality and readmission. The all-cause mortality rate exhibited a U-shaped relationship with plasma HDL levels, whereas the readmission rate showed an inverted U-shaped correlation with plasma HDL concentration, as depicted in Figure 2A and B.

Figure 2 Association Between HDL-C Levels and Clinical Outcomes (A) This graph illustrates the U-shaped relationship between all-cause mortality rates and plasma HDL-C levels; (B) This graph demonstrates the inverted U-shaped correlation between readmission rates and plasma HDL-C concentration.

At the Q3 quartile of plasma HDL-C levels, all-cause mortality rates reach their nadir, while readmission rates peak. Utilizing the univariate Kaplan-Meier method and applying the Pairwise Log rank test, we delineated survival curves to represent these findings, as illustrated in Figure 3. The data indicate that Group Q1 exhibits the poorest cumulative survival and the highest risk of death. In contrast, Groups Q3 and Q4 demonstrate the most favorable survival rates and the lowest risk of death, with their survival curves being nearly identical. These observations reinforce the non-linear correlation between HDL-C levels and the risk of mortality.

Figure 3 The relationship between HDL-C levels and cardiovascular risk is depicted through Kaplan-Meier survival curves, highlighting the non-linear association with all-cause mortality and readmission rates across different HDL-C quartiles.(A). Survival probabilities are assessed using the mean values of covariates. (B) Comparative survival analysis among Groups 1 to 4. (C) Risk evaluation using the mean values of covariates. (D) A specific risk comparison for Groups 1 to 4, emphasizing the U-shaped and inverted U-shaped relationships with HDL-C levels.The Kaplan-Meier method was employed to estimate the survival function, and the Pairwise Log rank test was used to statistically compare the survival curves between groups.The survival curves illustrate the variation in the probability of survival from Group Q1, with the lowest HDL-C levels, to Group Q4, with the highest HDL-C levels.The non-linear relationship between HDL-C levels and the likelihood of mortality is evident across all groups.

Analysis Outcomes from the COX Regression

In the univariate Cox regression analysis with HDL-C level as the independent variable and ACM as the endpoint, the reference was set to the HDL-C level group of 0.76–0.93 mmol/L, namely Group Q2. The hazard ratios (HR) for ACM for the remaining groups compared to Q2 were as follows: Group Q1 had an HR of 2.306 with a 95% CI of 1.922–2.767, Group Q3 had an HR of 1.078 with a 95% CI of 0.899–1.292, and Group Q4 had an HR of 1.420 with a 95% CI of 1.185–1.702, all with P < 0.05. Furthermore, after adjusting for confounding factors such as age and medication use in the multivariate Cox regression analysis, the results confirmed the persistence of this trend (all P < 0.05), as detailed in Tables 5 and 6.

Table 5 Univariate Cox Regression Analysis of All-Cause Mortality in Heart Failure Patients by HDL-C Level Quartiles

Table 6 Multivariate Cox Regression Analysis of All-Cause Mortality in Heart Failure Patients Adjusting for Age and Medication Use, and Other Factors

Discussion

HF denotes a clinical condition where the heart’s pumping capacity is compromised by diverse causes, leading to insufficient cardiac output to satisfy the body’s tissues’ metabolic requirements.¹⁷ HF is a burgeoning health risk, marked by a rise in both occurrence and fatality rates. Cardiac decline is linked to processes such as inflammation, cell death, and tissue decay within the heart and blood vessel linings, impairing the heart’s ability to contract and relax properly.¹⁸ Stress from oxidative sources, the endoplasmic reticulum, and nitrosative influences triggers an inflammatory response that hastens the decline of heart cells and attracts immune cells, promoting changes in the heart’s structure during HF.^19,20 Moreover, the overactivity of certain hormones like angiotensin II and leptin is implicated in the disruption of the heart’s pumping and filling cycles, playing a role in HF’s progression.^21–23 Previously, HDL was considered a substitute biomarker for a new inflammatory state. However, the exact role of HDL in HF remains controversial. Population based cohort studies have shown a negative correlation between elevated HDL levels and major adverse cardiovascular events (MACE). Previous epidemiological studies have shown that for every 1 mg/dL increase in HDL levels, the risk of cardiovascular death decreases by approximately 2% to 3%.^5–10 This indicates that HDL has an impact on the structure and function of the heart.²⁴ In multivariate analysis, HDL is an independent predictor of ACM and MACE. Recent epidemiological and genetic investigations have shown that higher HDL levels may not necessarily be better. Recent studies in populations without cardiovascular disease have shown that very high HDL levels may be associated with an increased risk of death.^25,26 However, the precise interplay between HDL levels and the spectrum of cardiovascular events, particularly in the context of HF, remains an enigma. Evidence suggests that HDL can mitigate oxidative stress, endoplasmic reticulum stress, and nitrosative stress, thereby curbing the secretion and release of inflammatory cytokines such as interleukin-1 β (IL-1 β) and tumor necrosis factor α (TNF- α), which are instrumental in the pathogenesis of HF.^27,28 Conversely, other lines of research indicate that HDL may paradoxically induce oxidative stress and exacerbate pro-inflammatory cytokine secretion, contributing to endothelial and myocardial cell damage, and the progression of HF.^29,30 These contradictory results indicate heterogeneity in the role of high-density lipoprotein in HF regulation, but the exact impact remains to be elucidated.

In our study, we found that HDL levels are a new predictive indicator of poor prognosis in HF patients and exhibit an inverted U-shaped trend. Compared with the patient group with HDL levels between 0.75 and 1.14 mg/dL, the incidence of mortality was significantly higher in both the low concentration and high concentration groups. A study from Japan reported that the risk of atherosclerotic cardiovascular death was significantly higher in individuals with very high HDL levels.³¹ In another study targeting the elderly, it was found that the risk ratio of all-cause mortality with HDL-C>90 mg/dL was driven by both cardiovascular and non cardiovascular mortality rates.³² Liu et al analyzed a group of people with coronary heart disease and found that individuals with low (<30mg/dL) or extremely high (>80mg/dL) HDL-C levels had a higher risk of all-cause and cardiovascular death³³; Voight et al found that both low and high levels of HDL-C are associated with an increased risk of ischemic and hemorrhagic stroke.¹³

It is worth noting that our study revealed a U-shaped correlation between HDL levels and the prognosis of heart failure, specifically in terms of mortality risk. This finding is consistent with the pattern observed in the risk of readmission. The U-shaped trend suggests that both extremely low and high HDL levels are associated with poorer outcomes for HF patients. However, the underlying mechanisms for this association remain unclear and warrant further exploration.

In this study, we evaluated the relationship between HDL and HF in a population of HF patients caused by various etiologies, including coronary heart disease, hypertension, cardiomyopathy, etc. Therefore, our observation addresses an important knowledge gap in this field.

In fact, in the current algorithms used to calculate cardiovascular risk in the general population, HDL levels are considered a protective factor, and at very high levels, this protective effect does not seem to hold. In fact, it may increase the risk. Before accepting this assumption, there are a few points to consider.

It is worth noting that the exact cutoff values used to define high HDL concentrations may vary in different studies, possibly due to differences in the study population or differences in specific measurements used to measure HDL levels.

During the research process, we recruited 4442 HF patients and conducted long-term follow-up to investigate the correlation between HDL and clinical prognosis. We divided the study population into four subgroups based on HDL levels (Q1: ≤ 0.75; Q2: 0.75–0.93; Q3: 0.93–1.14; Q4:>1.14), and found that the all-cause mortality rates were lower in the Q2 and Q3 HDL groups.

It is worth noting that after adjusting for potential confounding factors, we found significant differences in the incidence of cardiovascular events among the four groups of HF subjects divided by HDL levels. This discovery may suggest the existence of a U-shaped association. However, as previously studied by other investigators, the range of HDL values at the lowest point of the U-shaped curve is quite broad, covering the majority of the population.²⁵ The change in HDL-C levels at this lowest point does not seem to lead to any related risk changes. Considering the differences in the measurement methods used and the studied population, the high-risk inflection point of HDL-C seems to occur within the range of HDL levels>1.14 mmol/L, indicating that the research results tend to lean towards a U-curve rather than a V-curve.

A previous survey conducted in two groups of people did not draw a conclusion suggesting a U-shaped correlation curve between HDL-C levels and the incidence of ischemic heart disease, myocardial infarction, or ischemic stroke.²⁵ However, it was observed that lower levels of HDL corresponded to higher incidence of ischemic heart disease, myocardial infarction and ischemic stroke. However, HDL levels in males and females showed a plateau at around 3.22 and 4.23 mmol/L, respectively. When the HDL concentration exceeds these values, the risk will not be further reduced.²⁵ The differences between these data and our data may be due to differences in global cardiovascular risk among the study population: the general population and HF subjects, or cardiovascular events that also include end-stage HF requiring hospitalization, stable outpatient follow-up patients, coronary artery revascularization at first appearance, angina, hypertension, and atrial fibrillation, which are early evidence of impaired heart function. In either case, a U-shaped correlation was observed between HDL levels and the incidence of cardiovascular events, and the conclusion drawn was that high levels of HDL may lead to an increase in cardiovascular disease-related mortality.

Our observational studies cannot elucidate whether the association between high HDL levels and increased cardiovascular risk is causal, as Mendelian randomization studies do not support the causal role of HDL in cardiovascular events.^13,34 Therefore, we can only speculate on the possible pathogenesis. The observed association may be a phenomenon driven by pathological and physiological abnormalities, as well as immune, genetic, and epigenetic abnormalities, which increase cardiovascular risk in ways that we do not fully understand and confirm the complexity of the pathophysiology of HDL.^35–39 Individuals with extremely high levels of HDL-C may also alter the structure and function of HDL particles.^40,41 It can be assumed that in individuals with extremely high levels of HDL, the function of HDL may be impaired, making it unable to function properly and becoming harmful to the body.

Strengths and Limitations

One limitation of our research is that the number of individuals at the highest end of the HDL concentration spectrum is relatively small. However, the number of HF subjects included in our high HDL group did not prevent the possibility of a significant increase in cardiovascular risk after adjusting for confounding factors. The second limitation is the lack of collection of information on alcohol intake, which is precisely related to higher levels of HDL-C.^42,43 Due to our study conducted at the First Affiliated Hospital of Xinjiang Medical University, where the majority of participants were locals, our results may not be generalizable to other populations. Finally, we did not determine the cholesterol efflux capacity of all patients,⁴⁴ nor did we determine HDL apolipoprotein levels, which have recently been associated with mortality in cardiovascular disease patients.⁴⁵ We were unable to evaluate whether the use of lipid-lowering drugs affected our research results.

The main advantage of our research is that there is a large population recruited retrospectively, and extremely detailed information about individual clinical conditions is obtained through medical record review, follow-up, and other methods.

Conclusions

The present study has delineated a complex, non-linear relationship between HDL-C levels and clinical outcomes in patients with HF. Our findings indicate that both very low and very high HDL-C levels are associated with increased risks of all-cause mortality, challenging the traditional view of HDL-C as a universally beneficial biomarker.

Over a decade-long follow-up, we observed a U-shaped association between HDL-C levels and mortality, with the lowest risk observed in patients within the intermediate HDL-C range. This suggests that an optimal HDL-C level for HF patients may exist, and that both extreme deficiency and excess may be detrimental. The clinical implications of these findings underscore the need for a reevaluation of HDL-C management strategies in HF, potentially leading to more nuanced treatment approaches.

Moreover, our results highlight the importance of considering HDL-C levels as part of a comprehensive risk assessment in HF patients. The observed associations, while statistically significant, also call for further research to explore the underlying mechanisms linking HDL-C levels to adverse outcomes in HF. Future studies should aim to clarify the role of HDL-C in HF pathophysiology and identify the specific patient populations that may benefit from interventions targeting HDL-C levels.

In conclusion, our study provides valuable insights into the prognostic significance of HDL-C levels in HF and contributes to the growing body of evidence that suggests a reassessment of HDL-C’s role in cardiovascular health. These findings may ultimately inform more personalized and effective treatment strategies for patients with HF.

What is Known About This Topic?

● Recently, several cohort studies reported that either extremely low or high HDL-C levels were associated with increased risk of mortality. However, the relation between HDL-C and HF remains unclear.

What Does This Paper Add?

● In the present study, a nonlinear association was identified between HDL-C levels and HF patients, with the lower risk at intermediate levels.