Bringing doctors together

or

A free, private network for verified physicians

Leading voices

Dr. Dorairaj Prabhakaran

General Cardiology

View

Dr. Trey Woods

General Cardiology · Emory Healthcare

View

Dr. Rishi Wadhera

General Cardiology · BIDMC

View

Dr. Robert Harrington

General Cardiology · WCM

View

Dr. Mike Gaziano

General Cardiology · BWH

View

Dr. Tyler Gluckman

General Cardiology · Providence

View

Dr. Cayla Pichan

General Cardiology · UT Southwestern

View

Dr. Sanjiv Kaul

General Cardiology · OHSU

View

Dr. Emily Lau

General Cardiology · MGH

View

Dr. Sarah Ashley

General Cardiology · Emory University Hospital

View
  • Discover your community

  • Stay current

  • A profile that represents you

  • Find your people

About RoonPrivacyTerms

    Overview

    Lipoprotein(a) Levels and Adverse Outcomes in Heart Failure.

    Apr 4, 2025DOI: 10.1016/j.cardfail.2025.03.016Journal ArticleJCF

    Written by: Adithya K YadalamApoorva GangavelliAlexander C RazaviYi-An KoAyman AlkhoderNisreen HarounRafia LodhiAhmed EldaidamouniMahmoud Al KasemArshed A Quyyumi

    Find these authors on Roon

    AG

    Dr. Apoorva Gangavelli

    General Cardiology · Emory University

    AR

    Dr. Alexander Razavi

    Preventive Cardiology · Emory University Hospital

    Introduction

    Heart failure (HF) is a common and complex clinical syndrome associated with significant morbidity and mortality. The lifetime risk of HF in American adults is nearly 25%, and HF prevalence is expected to increase from 6.7 million to 8.5 million by 2030. Despite significant advances in HF treatment, HF-related mortality and hospitalization rates continue to rise, with Black patients being disproportionately affected.1 Novel strategies to identify patients at higher risk of adverse HF events are needed in order to mitigate the growing burden of HF-related morbidity.HF pathogenesis and progression are driven by multiple pathobiological pathways, and biomarkers play a crucial role in HF diagnosis, risk stratification, and management. Although natriuretic peptides (B-type natriuretic peptide [BNP] and N-terminal prohormone of brain natriuretic peptide [NT-proBNP]) are the cornerstone of biomarker-based HF prognostication methods through the detection of ventricular stretch, the detection of natriuretic peptide elevation represents assessment of only one of potentially several affected physiological pathways. A multi-biomarker approach could offer a more comprehensive assessment of the entirety of the HF clinical syndrome and serve to guide personalized treatment strategies.Lipoprotein(a) [Lp(a)] is a genetically inherited lipoprotein consisting of an apolipoprotein(a) molecule of varying length attached to an apolipoprotein B-100 core. Lp(a) elevation has been increasingly recognized as a potential risk factor for HF development, and emerging evidence suggests that Lp(a) elevation might also be associated with adverse outcomes in patients with HF. Previous studies have found that elevated Lp(a) independently predicts new-onset HF, as well as progression from asymptomatic to symptomatic HF.2,3 However, it is less established if Lp(a) level elevation is associated with cardiovascular mortality or incident HF hospitalization events in patients with known HF. Herein we analyzed a high-risk HF cohort with significant representation of Black individuals, a population with a higher prevalence of elevated Lp(a) levels,4 to examine the association between elevated Lp(a) levels and cardiovascular death or HF hospitalizations, with the hypothesis that higher Lp(a) levels would independently predict cardiovascular outcomes in patients with HF.

    Methods

    The Emory Cardiovascular Biobank (EmCAB) is a prospective registry of approximately 8,000 adults who were recruited prior to undergoing coronary angiography at 3 Emory Healthcare-affiliated hospitals in Atlanta, Georgia.5 Individuals with a history of severe congenital heart disease, severe valvular heart disease, severe anemia, recent blood transfusion, active inflammatory disease, active cancer, or dementia at the time of recruitment were excluded from enrollment into EmCAB. EmCAB participants enrolled between May 2003 and November 2023 with a past medical history of HF and non-missing data for Lp(a), NT-proBNP, and left ventricular ejection fraction (LVEF) were eligible for inclusion into the current study sample. EmCAB participants with a history of heart transplantation were not eligible for inclusion into the study sample. Participants with missing covariate data utilized in multivariable regression modeling were incrementally excluded from the present study (Figure S1). EmCAB participants were interviewed at the time of enrollment to collect clinical information related to demographics, medical and surgical history, and medication history.
    Comorbidities, such as prior history of heart failure, obstructive coronary artery disease (CAD; defined as ≥50% stenosis of an epicardial coronary artery, history of prior myocardial infarction (MI), history of percutaneous coronary intervention (PCI), or history of coronary artery bypass graft surgery), diabetes mellitus, and hypertension, were documented per clinician diagnosis and/or treatment. Participant-reported medical history was validated by medical record review. Height and weight at the time of enrollment were used to calculate body mass index (BMI). Estimated glomerular filtration rate (eGFR) was calculated by the Chronic Kidney Disease Epidemiology Collaboration Equation.7 EmCAB was approved by the institutional review board of Emory University (Atlanta, GA).
    Fasting arterial blood samples were collected prior to cardiac catheterization and subsequently stored at −80 °C. Serum concentrations of Lp(a) and NT-proBNP were measured utilizing the Abbott Architect platform (Abbott Laboratories). Total cholesterol (TC) and high-density lipoprotein cholesterol (HDL-C) were assayed (Beckman Coulter).
    Heart failure phenotypes were classified by LVEF, which was documented by review of the most recent transthoracic echocardiogram (TTE) nearest to the time of EmCAB recruitment or by left ventricular angiography during cardiac catheterization. Participants with an LVEF ≤40% were classified as heart failure with reduced ejection fraction (HFrEF), LVEF 41–49% as heart failure with mildly reduced ejection fraction (HFmrEF), and LVEF ≥50% as heart failure with preserved ejection fraction (HFpEF).6 Although targeted analyses with the HFmrEF subgroup were attempted, these analyses were ultimately deferred due to a lack of statistical power related to the small size of the HFmrEF subgroup (N=89). Additionally, EmCAB participants with HF and comorbid obstructive CAD were defined as having ischemic HF. Those participants with HF and no history of obstructive CAD were defined as having nonischemic HF.
    The primary outcome was defined as the composite of the first occurrence of either cardiovascular death or HF hospitalization. The secondary outcome was defined as all-cause death. EmCAB participants are prospectively followed for adverse outcomes by blinded personnel through medical record review, Social Security Death Index, state record review, and telephone interview at 1 year, 5 years, and every subsequent 5 years following enrollment. Cause of death is adjudicated by two blinded cardiovascular medicine physicians. Medical records were obtained to corroborate self-reported events. Cardiovascular death was defined as death from MI, HF, stroke, pulmonary embolism, or as a complication during a cardiovascular-related procedure. MI was adjudicated using the third and fourth universal definition of MI.8,9
    Demographics and clinical factors were reported as frequencies (percentages) for categorical variables and means (SD) or medians [IQR] for all continuous variables, as appropriate. The Pearson’s Chi-squared and Kruskal-Wallis rank-sum tests were utilized to assess for between-group differences for categorical and continuous variables. The Mann-Whitney U Test was utilized to compare Lp(a) levels between participants with HFrEF and those with HFpEF.
    The risk of cardiovascular death or HF hospitalization across the entire spectrum of Lp(a) levels was first visualized with a penalized spline constructed using the smoothHR R package.10 Variance inflation factors (VIF) were calculated in multivariable linear regression modeling for the primary outcome of cardiovascular death or HF hospitalization to assess for multicollinearity between covariates utilized in multivariable regression modeling. VIF ≥5 was considered indicative of moderate collinearity and consideration for covariate removal. Lp(a) was then analyzed in all further survival analyses by low-risk (<30 mg/dL), intermediate risk (30–49 mg/dL), and high risk (≥50 mg/dL) categories, per the 2024 National Lipid Association consensus statement on Lp(a).11 Cumulative incidence curves were created to visualize the association between Lp(a) groups and the primary composite outcome of cardiovascular death or HF hospitalization. The independent association between Lp(a) levels and cardiovascular death or HF hospitalization was assessed by Fine and Gray competing-risk regression modeling, while accounting for non-cardiovascular death.12 Regression modeling was adjusted for demographics (age, sex, and race [Black race vs. non-Black race]), traditional cardiovascular risk factors (BMI, diabetes mellitus, hypertension, TC, HDL-C, eGFR, aspirin use, statin use, angiotensin-converting enzyme [ACE] inhibitor or angiotensin receptor blocker [ARB] use, and beta-blocker use), LVEF, ischemic HF etiology, and NT-proBNP levels. We additionally examined the relationship between Lp(a) and NT-proBNP level elevation for the prediction of adverse cardiovascular outcomes with cumulative incidence curves and competing-risk regression analyses. In these models, Lp(a) level elevation was defined as ≥30 mg/dL, and NT-proBNP elevation was defined as greater or equal to the sample median (870 pg/mL). The association between Lp(a) levels and all-cause death was assessed by Cox proportional hazards modeling.
    Sensitivity analyses were performed to explore for heterogeneity of effect by examining the interactions between elevated Lp(a) levels and demographics, LVEF, ischemic HF etiology, and other significant predictors of adverse cardiovascular outcomes discovered in multivariable regression modeling. The Benjamini-Hochberg procedure was applied to account for multiple hypothesis testing. For these analyses, Lp(a) was dichotomized at 30 mg/dL, age at 65 years, eGFR at 60 mL/min/1.73 m2, and LVEF at 40%. Additional post-hoc sensitivity analyses were performed to assess for time-varying effects of Lp(a) on the risk of cardiovascular death or HF hospitalization. A competing-risk model was fitted with the Lp(a) risk groups of interest, all covariates from the fully adjusted model, and an interaction term for each Lp(a) risk group of interest multiplied by the natural log of time to the primary outcome, and risk estimates for the Lp(a) 30–49 mg/dL and >50 mg/dL groups compared to the <30 mg/dL group at 1, 2, and 3 years were manually calculated. Lastly, to consider the impact of missing covariate data, multiple imputation by chained equations was performed by predictive mean matching. Fine and Gray competing-risk regression models were then fitted with the Lp(a) risk groups of interest and all covariates from the fully adjusted model in each of the 10 imputed datasets, and model estimates were pooled using Rubin’s Rules to compute subdistribution hazard ratios, 95% confidence intervals, and P-values.
    All analyses were performed with R 4.2.2 (https://www.R-project.org). P values <0.05 were considered statistically significant.

    Results

    Baseline characteristics of the study sample are shown in Table 1. The study sample included 1,088 EmCAB participants with a history of HF. The mean age was 67 years, 64% were men, and 24% were of self-identified Black race. Of the 76% (N=828) of study participants who did not identify as Black, 97% where White (N=786), 2% were Asian (N=14), 1% were Hispanic (N=11), <1% were Native American (N=1), and 2% identified as “Other” (N=16). Within our study sample, 60.3% (N=657) had low-risk Lp(a) levels (<30 mg/dL), 14.1% (N=153) had intermediate-risk Lp(a) levels (30–49 mg/dL), and 25.6% (N=278) had high-risk Lp(a) levels (≥50 mg/dL). Regarding HF phenotypes, 658 (60.5%) were diagnosed with HFrEF, 89 (8.2%) with HFmrEF, and 341 (31.3%) with HFpEF. Median Lp(a) levels did not significantly differ between HFrEF and HFpEF subgroups (20.4 [IQR 43.8] mg/dL vs. 18.4 [IQR 41.1] mg/dL, P=0.442). When compared to EmCAB participants without a history of HF (N=3,085), those with a history of HF had significantly higher Lp(a) levels (19.1 [IQR 43.3] mg/dL vs. 15.6 [IQR 37.5] mg/dL, P<0.001).
    Over a median follow-up of 4.3 [IQR 4.3] years and maximum follow-up of 11.1 years, 474 (44%) cardiovascular death or HF hospitalization events and 489 (45%) all-cause death events occurred. A nearly linear increase in the risk of cardiovascular death or HF hospitalization was observed with increasing Lp(a) levels (Figure 1). A significant difference between the cumulative incidence of adverse cardiovascular outcomes between Lp(a) levels by low-, intermediate-, and high-risk groupings was observed (Figure 2). Assessment for multicollinearity was assessed with VIFs. All covariates included in multivariable regression modeling had VIF values <1.5, indicative of minimal collinearity (Table S1).
    When compared to participants with Lp(a) <30 mg/dL in unadjusted competing-risk regression modeling accounting for non-cardiovascular death, those with Lp(a) ≥50 mg/dL (sHR 1.41, 95% CI 1.15–1.73, P=0.001) and 30–49 mg/dL (sHR 1.30, 95% CI 1.00–1.68, P=0.046) had a significantly higher risk of cardiovascular death or HF hospitalization (Table 2, Model 1). After adjustment for demographics (Table 2, Model 2) and further adjustment for traditional cardiovascular risk factors, obstructive CAD history, LVEF, ischemic HF etiology, and cardiovascular medication use (Table 2, Model 3), no significant attenuation of risk estimates was observed. In the fully adjusted model including NT-proBNP levels, both Lp(a) ≥50 mg/dL (sHR 1.38, 95% CI 1.11–1.72, P=0.004) and 30–49 mg/dL (sHR 1.35, 95% CI 1.04–1.76, P=0.025) groups remained significantly associated with a higher risk of cardiovascular death or HF hospitalization when compared to those with Lp(a) <30 mg/dL) (Table 2, Model 4).
    Cumulative incidence modeling of cardiovascular death or HF hospitalization accounting for competing non-cardiovascular death demonstrated the highest risk of adverse cardiovascular outcomes in participants with both Lp(a) and NT-proBNP elevation (Figure 3). When compared to participants with low Lp(a) and NT-proBNP levels in multivariable-adjusted competing-risk regression modeling, participants with elevation of both biomarkers had a substantially greater risk of adverse cardiovascular outcomes (sHR 2.71, 95% CI 2.05–3.60, P<0.001). Poorer risk prediction was observed in those with solely elevated NT-proBNP levels (sHR 2.14, 95% CI 1.63–2.82, P<0.001) or Lp(a) levels (sHR 1.46, 95% CI 1.07–1.99, P=0.016), when compared to those with low levels of both biomarkers.
    Lp(a) ≥50 mg/dL was also associated with the secondary endpoint of all-cause death (HR 1.30, 95% CI 1.12–1.51, P<0.001) after full adjustment when compared to Lp(a) <30 mg/dL, whereas Lp(a) 30–49 mg/dL was not (HR 1.09, 95% CI 0.91–1.30, P=0.378) (Table S2).
    Sensitivity analysis revealed no significant interactions between Lp(a) levels and the included covariates (age, race, ischemic HF etiology, LVEF, eGFR, and ACE/ARB use) for the prediction of cardiovascular death or incident heart failure hospitalization after adjustment for multiple hypothesis testing. Nominal significance was observed for the interaction between Lp(a) levels and ischemic HF etiology (P=0.06) but did not meet false discovery rate-adjusted thresholds (Figure 4).
    In post-hoc analyses examining for time-varying effects of Lp(a) with cardiovascular death or HF hospitalization risk, the interaction terms with time for both intermediate- and high-risk Lp(a) elevation were found to be significant and with negative coefficients, indicative of attenuation of Lp(a)-associated risk over time. Both Lp(a) 30–49 mg/dL (sHR 5.03, 95% CI 3.58–7.09, P<0.001) and >50 mg/dL (sHR 4.60, 95% CI 3.28–6.46), P<0.001) were strongly predictive of adverse cardiovascular risk at 1 year when compared to Lp(a) <30 mg/dL but trended towards significance at 3 years (Lp(a) 30–49 mg/dL: sHR 1.30, 95% CI 0.99–1.70, P=0.057; Lp(a) >50 mg/dL: sHR 1.25, 95% CI 0.98–1.58, P=0.068) (Table S3).
    In a sensitivity analysis accounting for missing covariate data, variables with missing data (HDL-C 9.7%, total cholesterol 9.3%, BMI 6.7%, diabetes mellitus 1.1%, and hypertension 1.0%) were imputed. Pooled estimates revealed that Lp(a) 30–49 mg/dL vs. Lp(a) <30 mg/dL trended towards significance for the prediction of cardiovascular death or incident HF hospitalization (sHR 1.24, 95% CI 0.97–1.59, P=0.086), whereas Lp(a) ≥50 mg/dL vs. Lp(a) <30 mg/dL remained significantly predictive (sHR 1.34, 95% CI 1.09–1.64, P=0.005).

    Discussion

    In this prospective cohort study of over 1,000 patients with HF referred for coronary angiography, we show that elevated Lp(a) levels independently predicted the risk of cardiovascular death or HF hospitalization, even after adjustment for demographics, traditional cardiovascular risk factors, and natriuretic peptide levels. Those with both high-risk Lp(a) levels (≥50 mg/dL) and intermediate-risk Lp(a) levels (30–49 mg/dL) had a nearly 35% increased risk of adverse cardiovascular outcomes when compared to those with low-risk Lp(a) levels (<30 mg/dL). Furthermore, we show that a multi-biomarker model including both Lp(a) and NT-proBNP levels provided more robust risk prediction for adverse cardiovascular outcomes when compared to elevation in either biomarker alone. Lastly, the effect of Lp(a) on adverse cardiovascular outcome risk may diminish over time.
    Most previous studies investigating the association between elevated Lp(a) levels and HF have primarily focused on the relationship between Lp(a) and new-onset HF and did not include individuals previously diagnosed with HF.13,14 The few studies that have examined the association between Lp(a) elevation and adverse HF events in patients with known HF have included only patients with HFrEF and not those with HFpEF.15–17 Moreover, the follow-up time was restricted to 1 year in 2 prior studies, reducing the ability of these studies to detect incident HF events and introducing the potential for survivorship bias.16,18 Three of these studies were performed in China15,16,18 and one in Japan,17 limiting generalizability to broader populations. Although informative in deciphering the role of Lp(a) in patients with HF, our findings build on these previous observations in several important ways.
    In a diverse cohort of American adults with established HF, we show that higher Lp(a) levels predicted adverse cardiovascular outcomes over a period of several years, independent of important clinical variables and NT-proBNP levels. We also show that patients with HF who have at least intermediate-risk Lp(a) elevation in addition to NT-proBNP elevation have a substantially higher risk of adverse cardiovascular outcomes than those with single or no biomarker elevation. Our sensitivity analyses revealed a consistently higher risk of cardiovascular death or HF hospitalization attributable to Lp(a) elevation across multiple clinically relevant subgroups, along with nominally higher adverse cardiovascular outcome risk conferred by Lp(a) elevation in those with ischemic HF etiology when compared to those with nonischemic HF etiology. We furthermore observed a time-varying effect of Lp(a) on cardiovascular risk, such that the risk of cardiovascular death or HF hospitalization attributable to elevated Lp(a) levels decreased over time. This may be occurring for multiple reasons, the most salient of which includes diminished representation of high-risk individuals over time due to earlier adverse event occurrence in these participants, Lp(a)-related atherosclerotic/thrombotic events possibly being more impactful earlier in the disease course of those with an ischemic HF etiology, and other factors beyond Lp(a) potentially contributing more strongly to HF progression over time (e.g., worsening myocardial function, persistent neurohormonal activation, renal impairment, etc.).
    Regarding mechanism, there are several pathways through which Lp(a) could contribute to adverse HF outcomes in patients with HF beyond the well-known contribution of coronary atherosclerosis to HF pathogenesis through regional myocardial dysfunction. First, Lp(a) is a homologous molecule to plasminogen and competes for binding sites on endothelial cells.19 This process inhibits fibrinolysis and promotes intravascular thrombosis which can also lead to myocardial damage. Second, Lp(a) is known to be associated with myocardial fibrosis and cardiac remodeling.20 Increased interstitial myocardial fibrosis can impair myocardial function and lead to the progression of HF. Third, elevated Lp(a) levels are associated with increased arterial stiffness, a contributor to the progression of HF.21,22 Increased arterial stiffness leads to increased left ventricular afterload and ultimately to diastolic dysfunction, a key marker of HFpEF.23
    Our study has several strengths. Our cohort contained both HFrEF and HFpEF phenotypes, allowing us to examine the impact of elevated Lp(a) levels across the spectrum of LVEF. Furthermore, the significant representation of Black patients within our study sample may provide for our findings to be generalizable to the broader American public. Limitations include the observational nature of our analyses that cannot completely exclude residual confounding despite multivariable adjustment and our sample including only participants recruited prior to coronary angiography, increasing the likelihood of recruiting participants with ischemic rather than non-ischemic HF. Thus, interaction analyses of Lp(a) with ischemic HF and those with HFrEF are largely hypothesis-generating and should be interpreted with caution. Our findings were also limited by the number of eligible EmCAB participants with missing covariate data (17%), the exclusion of whom may have resulted in a potentially stronger association being observed between intermediate-risk Lp(a) elevation and adverse cardiovascular outcomes, as noted in a sensitivity analysis. However, given that these results relied on imputation of missing data under a specific missing data mechanism, we consider these results to serve as a supplemental analysis. Lastly, although a composite outcome incorporating HF hospitalizations was utilized in an attempt to capture a more holistic representation of HF morbidity, there is potential for hospitalizations to not have been detected by medical record review or patient interviews.

    Conclusions

    In over 1,000 patients with HF referred for coronary angiography, Lp(a) elevation ≥30 mg/dL was associated with a ≥35% higher risk of cardiovascular death or HF hospitalization after adjustment for clinical variables and natriuretic peptide levels. This risk appeared to diminish over time and may be greater in patients with ischemic HF. Whether lowering of Lp(a) levels with emerging therapies in patients with HF improves outcomes remains to be studied.24–26
    AG

    Dr. Apoorva Gangavelli

    Cardiology · General Non-Invasive Cardiology
    Emory University
    apoorva