Association Between Lipoprotein(a) and Cardiac Remodeling Across Race and Ethnicity in the Multi-Ethnic Study of Atherosclerosis

Abstract

BACKGROUND:

Lp(a) (lipoprotein[a]) is a known cardiovascular risk factor; however, its role in cardiac remodeling and functional changes over time across diverse racial and ethnic groups remains underexplored.

METHODS:

MESA is a prospective multi-ethnic cohort study of individuals without a history of cardiovascular disease on enrollment (2000–2002), conducted across 6 sites in the United States. Participants with baseline Lp(a) measurements and cardiac magnetic resonance imaging at both baseline and 10-year follow-up exam were included. Lp(a) was treated as both a log-transformed continuous variable (per SD log) and a categorical variable based on data-driven Lp(a) terciles. Multivariable regression models adjusted for sociodemographic, and cardiovascular risk factors, including coronary artery calcium and interim myocardial infarction, were used to assess associations between Lp(a) and longitudinal changes in left ventricular and atrial structure and function over a decade across different racial/ethnic groups.

RESULTS:

A total of 2366 participants were included. The average age at baseline was 60±9 with 53% women, 43% White, 24% Black, 21% Hispanic, and 12% Chinese. Each 1-SD increase in log-transformed Lp(a) was associated with an increase in left ventricular end-systolic volume index (β, 0.60 [95% CI, 0.02–1.18]), and left atrial minimum volume index (β, 0.81 [95% CI, 0.09–1.52]), and a decline in left ventricular ejection fraction (β, −0.75 [95% CI, −1.34 to −0.17]), and total left atrial emptying fraction (β, −1.17 [95% CI, −2.09 to −0.24]) in Hispanic subjects over a decade. No significant associations were seen in White, Black, or Chinese participants. The observed findings persisted after adjusting for coronary artery calcium, interim myocardial infarction, and atrioventricular decoupling, and when Lp(a) was treated as a categorical variable with race-specific terciles.

CONCLUSIONS:

Elevated Lp(a) levels were independently associated with maladaptive left ventricular and left atrial remodeling in Hispanic adults over a decade, while no statistically significant relationships were observed in White, Black, and Chinese participants. This suggests a unique susceptibility of Hispanic individuals to Lp(a)-mediated cardiovascular remodeling, independent of ischemic pathways.

WHAT IS KNOWN

• Lipoprotein(a) is a genetically determined lipoprotein and an established risk factor for atherosclerotic cardiovascular disease, with marked variation in circulating levels across racial and ethnic groups.

• Prior studies have demonstrated racial and ethnic heterogeneity in the association between lipoprotein(a) and clinical cardiovascular outcomes, including coronary heart disease and heart failure.

WHAT THE STUDY ADDS

• In a multi-ethnic cohort free of clinical cardiovascular disease at baseline, higher lipoprotein(a) levels were independently associated with adverse longitudinal left ventricular and left atrial remodeling over a decade among Hispanic adults only.

• These associations were observed at lower lipoprotein(a) thresholds than current guideline cutoffs and were independent of coronary artery calcium and interim myocardial infarction, suggesting nonischemic pathways of myocardial remodeling.

• The findings support the need for population-specific lipoprotein(a) risk assessment and provide a mechanistic rationale for evaluating emerging lipoprotein(a)-lowering therapies on myocardial structure, fibrosis, and heart failure risk, particularly in populations with heightened susceptibility.

Cardiac remodeling encompasses structural and functional changes in the heart, occurring as an adaptive response to altered hemodynamic loading conditions or myocardial injury. These changes include variations in left ventricular (LV) and left atrial (LA) volume, mass, and function, which often precede the onset of clinically overt heart failure (HF).¹ Notably, maladaptive cardiac remodeling serves as a robust predictor of adverse cardiovascular outcomes, including myocardial infarction (MI) and stroke.¹ Identifying modifiable and nonmodifiable risk factors for cardiac remodeling remains a key objective in cardiovascular research, with lipoproteins emerging as significant contributors.^2–4

Lp(a) (lipoprotein[a]) is a low-density lipoprotein particle that includes apolipoprotein(a) covalently bound to apolipoprotein B-100.⁵ Elevated Lp(a) levels are a recognized risk factor for atherosclerotic cardiovascular disease, including coronary artery disease, aortic stenosis, and stroke.⁶ Lp(a) levels are largely genetically determined and exhibit marked variability across different racial and ethnic groups.⁷ Importantly, previous reports have shown racial/ethnic disparities in the association between Lp(a) and clinical outcomes, including coronary heart disease,⁸ carotid atherosclerosis,⁹ peripheral arterial disease,¹⁰ and HF.¹¹ However, whether similar differences exist with regard to the link between Lp(a) levels and subclinical cardiac remodeling remains poorly understood. This knowledge gap is particularly relevant to inform clinical strategies for personalized risk stratification and intervention in the era of emerging novel therapeutic strategies for lowering Lp(a) levels.^12–17

The current study aims to address this gap by investigating the association between Lp(a) levels and longitudinal changes in cardiac remodeling indices across 4 distinct racial/ethnic groups in a multi-ethnic cohort free of cardiovascular disease on enrollment. Using data from the MESA (Multi-Ethnic Study of Atherosclerosis), this study evaluates LA and LV remodeling parameters over a decade to elucidate racial and ethnic differences in cardiac remodeling linked with Lp(a).

Methods

Data Availability

Data analyzed during the study were provided by the MESA Coordinating Center. Requests for data should be directed to the MESA Coordinating Center indicated in the Acknowledgment.

Study Population

The MESA study was established to explore the incidence and risk factors of cardiovascular disease in individuals who initially did not exhibit any noticeable clinical cardiovascular disease, cancer treated with radiation or chemotherapy, severe major illness, or cognitive impairment, as assessed by the screening interviewer at the visit 1 baseline examination.¹⁸ Between 2000 and 2002, 6814 men and women 45 to 84 years old, self-identified and categorized into mutually exclusive racial/ethnic backgrounds per study protocol as White, Black, Hispanic, or Chinese, were enrolled from 6 different locations: Baltimore, MD; Chicago, IL; Forsyth County, NC; Los Angeles County, CA; New York, NY; and St. Paul, MN. On enrollment (2000-02), MESA participants answered clinical questionnaires and had physical examination, diverse imaging studies, and laboratory tests. Institutional review boards at each field center approved the study protocol, and all participants gave written informed consent. More information about the study’s methodology can be accessed online at https://www.mesa-nhlbi.org. In the present study, we examined participants with available Lp(a) levels at baseline, who underwent cardiac magnetic resonance imaging (MRI) at baseline and 10 years later during visit 5.

Lp(a) Determination

After fasting, blood samples were collected, and tubes containing EDTA anticoagulant were managed according to uniform methodology.¹⁹ These samples were divided and kept at -70 °C until the time of analyte assessment. Previously established methods were employed to measure fasting triglyceride, total cholesterol, and high-density lipoprotein cholesterol levels.²⁰ As the exposure variable, Lp(a) mass concentrations were assessed at baseline for 6700 MESA samples by Health Diagnostics Laboratory (Richmond, VA) using a latex-enhanced turbidimetric immunoassay (Denka Seiken, Tokyo, Japan) that compensates for the diverse sizes of apo(a; apolipoprotein(a)).²¹ Total imprecision amounted to <5%.

Cardiac MRI Measurements

The cardiac magnetic resonance protocol for baseline and follow-up visits has been previously reported in detail.³ Briefly, LV structural and functional parameters, including LV mass, end-diastolic volume, end-systolic volume, mass-to-volume ratio, stroke volume, and ejection fraction, were assessed using fast gradient-echo MRI at baseline examination and steady-state free precession sequences for LV cine imaging at the follow-up visit. Calibration between the 2 cardiac magnetic resonance examinations was performed in participants who had both image sequences acquired at the MESA follow-up examination, as reported previously.³ Multimodality tissue tracking software (MTT version 6.0; Toshiba Medical Systems) was used to quantify LA volume and strain at 2- and 4-chamber cine cardiac MRI. This method has been validated previously with good to excellent intra- and inter-reader reproducibility (intraclass correlation, 0.88 to 0.98; P<0.001) and good interstudy reproducibility (intraclass correlation, 0.44–0.82; [P value range, 0.02 to <0.001]).^22,23

Statistical Analysis

Data were presented as mean (SD) or median (interquartile range [IQR]) for continuous variables based on normality and counts and percentages for categorical variables. Group comparisons were made using ANOVA, paired t test, Fisher exact test, or χ² test as appropriate. Multiple comparisons for continuous and categorical parameters used the Tukey-Kramer significant difference test and Bonferroni correction, respectively. Lp(a) was treated as both a log-transformed continuous variable (per SD) and a categorized variable based on terciles. Since there was a significant interaction between race/ethnicity and Lp(a; P<0.001), analyses were reported stratified by race/ethnicity.

Multivariable linear regression was used to assess the association of changes in LA and LV parameters with Lp(a) and changes in risk factors over 10 years of follow-up with complete-case analysis. Demographic and risk factor covariates chosen a priori according to clinical relevance or their biological impact on the exposure and outcome variables included age, sex, race/ethnicity, education, body mass index, smoking status, systolic blood pressure, diastolic blood pressure, use of anti-hypertensive medications, total cholesterol, high-density lipoprotein cholesterol, use of lipid-lowering medications, fasting plasma glucose, and estimated glomerular filtration rate. All models were also adjusted for baseline values of the respective LA and LV measures being studied. For longitudinal analysis, the change in risk factors for continuous variables was assessed using the difference (year-10 follow-up exam—baseline). For medication use (hypertension and lipid), participants were categorized as no (not taking meds at baseline and year-10 follow-up exam), stop (taking meds at baseline but not at year-10 follow-up exam), new (not taking at baseline but taking at year-10), and long-standing (taking meds at both exams). Similarly, 4 categories were considered for smoking—never (nonsmokers at baseline and year-10 follow-up exam), stop (smokers at baseline but not at year-10 follow-up exam), new (nonsmokers at baseline but smokers at year-10 follow-up exam), and long-standing (smokers at both exams). A series of sensitivity analyses was performed. First, 1 supplemental model (model 3) further adjusted for coronary artery calcium scores and excluded participants with interim MI. Additionally, for LV outcomes, a supplemental model (model 4) further adjusted for baseline left atrioventricular coupling index²⁴—defined as the ratio of LA to LV volume at end-diastole—and LA total emptying fraction, while for the LA outcomes, model 4 adjusted for baseline left atrioventricular coupling index and LV ejection fraction. Finally, analyses were repeated with race/ethnicity-specific Lp(a) terciles as the predictor variable to account for the difference in distributions of Lp(a) among racial/ethnic groups.

The models’ fit to the data was assessed by inspecting plots of residuals against all covariates and against predicted values to look for nonrandom patterns that may indicate nonlinearity or heteroscedasticity. Moreover, the Shapiro-Wilk test assessed the normal distribution of models’ residuals, and variance inflation factors were used to assess the influence of multicollinearity in the adjusted models. A 2-sided Bonferroni-adjusted P value of <0.05 was considered statistically significant. All statistical analyses were performed using R version 4.4.2 (The R Foundation).

Results

Participant Characteristics

The baseline characteristics of the study population are summarized in Table 1. Significant differences in age, sex distribution, anthropometric measures, blood pressure, lipid profiles, and other cardiovascular risk factors were observed across racial/ethnic groups. Chinese participants were younger compared with White adults (P=0.007). Hispanic individuals had the lowest proportion of females (47%) (P=0.026). Compared with White participants, Chinese subjects exhibited lower body mass index values (24±3 kg/m²; P<0.001), while Black (30±5 kg/m²; P<0.001) and Hispanic (29±4 kg/m²; P<0.001) adults had higher body mass index values. Black participants demonstrated the highest mean systolic and diastolic blood pressures (130±21 mm Hg, 75±10 mm Hg, respectively), followed by Hispanic (124±20 mm Hg and 72±9 mm Hg, respectively), White (120±19 mm Hg and 70±10 mm Hg), and Chinese individuals (120±20 mm Hg and 72±10 mm Hg; P<0.001). Consistent with this trend, Black adults also had the highest prevalence of antihypertensive medication use (45%; P<0.001). In terms of lipid profiles, Hispanic participants had the highest total cholesterol (199±36 mg/dL) and triglyceride (135 [IQR, 98–191] mg/dL) levels, and lowest high-density lipoprotein cholesterol levels (47±13 mg/dL, P<0.001), while LDL cholesterol levels were comparable across groups (P=0.058). Black adults demonstrated significantly higher Lp(a) levels (36 [IQR, 20–63] mg/dL), compared with White (12 [IQR, 5–30] mg/dL), Chinese (13 [IQR, 8–22] mg/dL), and Hispanic individuals (14 [IQR, 6–32] mg/dL; P<0.001; Figure 1). The time between the 2 cardiac MRI exams was similar between White (3457±181 days) and Hispanic individuals (3480±166 days; P=0.083), but shorter in Black (3414±181; P<0.001) and longer in Chinese (3556±206; P<0.001) individuals compared with White subjects (Table 1).

Table 1.

Characteristics of Participants by Race/Ethnic Group at MESA Baseline Exam

Figure 1.

Lp(a) (lipoprotein[a]) distribution kernel density plot. Kernel density plots of Lp(a) concentrations stratified by race/ethnicity: White (n=1030; 43%), Chinese (n=273; 11%), Black (n=570; 24%), and Hispanic (n=493; 21%). Dashed vertical line indicates an Lp(a) threshold of 30 mg/dL; dotted vertical line indicates an Lp(a) threshold of 50 mg/dL.

Cardiac structure and function parameters also exhibited racial/ethnic differences (Table 1). Hispanic participants had the highest LV end-diastolic volume index (LVEDVi; 72±12 mL/m²), LV end-systolic volume index (LVESVi; 27±6 mL/m²), and LV end-diastolic mass index (67±10 g/ m²) compared with the other groups (P<0.001). Black individuals showed the highest LV mass-to-volume ratio (0.97±0.18; P<0.001). Left atrial minimum volume index and maximum indexed volumes were highest in Hispanic participants (13±6 mL/m² and 31±9 mL/m², respectively; P<0.001), while total LA emptying fraction (LAEF) was most reduced in Black individuals (50±10%; P<0.001; Table 1).

Lp(a) and LV Remodeling

Over a decade, trends in LV remodeling indices showed differences across Lp(a) terciles and racial/ethnic groups (Table S1). Among White adults, all terciles demonstrated significant reductions in LVEDVi and LVESVi over time (P<0.001). LV ejection fraction (LVEF) declined in the first tercile (baseline: 62.4%, follow-up: 61.2%; P=0.015) but remained stable in the second and third terciles. The mass-to-volume ratio significantly increased across all terciles, reflecting concentric remodeling (P<0.001). In Chinese participants, significant reductions in LVEDVi and LVESVi were observed across all terciles (P<0.001). Unlike White individuals, LVEF was stable during follow-up in Chinese participants. LV mass-to-volume ratio similarly increased across all terciles, indicating consistent concentric remodeling (P<0.001). Black participants demonstrated a more complex pattern, with significant decreases in LVEDVi and LVESVi in the second and third terciles (P<0.001) but no consistent change in the first tercile. LVEF remained stable in the first and second terciles, whereas the third tercile experienced a decline (baseline: 63.2%, follow-up: 61.2%; P<0.001). LV mass-to-volume ratio increased significantly across all terciles, aligning with a pattern of concentric remodeling (P<0.001). Among Hispanic participants, similar reductions in LVEDVi and LVESVi were observed across all terciles (P<0.001). However, LVEF declined significantly only in the third tercile (baseline: 63.1%, follow-up: 61.0%; P=0.039), a trend that was not observed in the first and second terciles. LV mass-to-volume ratio increased across all terciles, with the greatest change in the third tercile, highlighting a stronger concentric remodeling pattern in this group (P<0.001).

In multivariable-adjusted analyses, significant interactions were observed between race/ethnicity and Lp(a) in relation to LV remodeling indices (P<0.001; Table S2), indicating that the association of Lp(a) with LV remodeling differed by racial/ethnic group. Compared with White individuals (reference group), Hispanic participants exhibited a strong association between higher Lp(a) levels and adverse LV remodeling indices (Table S1), while no significant associations were observed in Black or Chinese participants (Tables S2 and S3, respectively). When analyses were stratified by race/ethnicity, higher Lp(a) levels were significantly associated with adverse LV remodeling indices only in Hispanic adults (Table 2), while no significant associations between Lp(a) and longitudinal changes in LV remodeling indices were observed in White, Black, or Chinese participants (Table 2; Table S3).

Table 2.

Association of Lp(a) With Changes in LV Remodeling Indices Over a Decade Stratified by Race and Ethnicity Groups

In multivariable-adjusted analysis controlling for sociodemographic and cardiovascular risk factors, Lp(a) was associated with an increase in LVESVi (per SD log Lp(a) β, 0.60 [95% CI, 0.02–1.18]) and decrease in LVEF (per SD log Lp(a) β, −0.75 [95% CI, −1.34 to −0.17]) in Hispanic adults. Similarly, when Lp(a) was analyzed as a categorical variable, the third Lp(a) tercile was linked with an increase in LVESVi (β, 1.76 [95% CI, 0.30–3.22]) and a decrease in LVEF (β, −1.86 [95% CI, −3.32 to −0.40]) compared with the first tercile (Table 2; Figure 2). These associations remained significant even after further adjustment for coronary artery calcium scores, excluding individuals with interim MI, and also adjusting for left atrioventricular coupling index and LAEF (Table 2). Similar results were observed when Lp(a) was treated as a categorical variable with race-specific cutoffs for terciles (Table S4).

Figure 2.

Association of Lp(a) (lipoprotein[a]) terciles with left ventricular (LV) remodeling by race/ethnicity. Forest plot showing β coefficients (95% CI) adjusted for sociodemographic and traditional cardiovascular risk factors (model 2) for ΔLVESVi and ΔLVEF by second (Lp[a], 10–30 mg/dL) and third (Lp[a] ≥30 mg/dL) Lp(a) terciles, with first Lp(a) tercile as the reference (Lp[a] <10 mg/dL). LVEF indicates left ventricular ejection fraction; and LVESVi, left ventricular end-systolic volume index.

Lp(a) and LA Remodeling

Changes in the LA remodeling indices also varied across Lp(a) terciles and racial/ethnic groups over a decade. Among White, Black, and Hispanic individuals, significant increases in LA volumes and decreases in LA function and strain were observed across all terciles (P<0.001; Table S1). Among Chinese participants, LA volumes increased significantly in all terciles (P<0.001). Total LAEF declined in the first and second terciles but remained relatively stable in the third tercile. Interestingly, the LA strain showed slight improvements in the second and third terciles, suggesting preserved LA compliance in those with higher Lp(a) levels (P<0.05).

A significant interaction was observed between race/ethnicity and Lp(a) in relation to LA remodeling indices (P<0.001), indicating that the association of Lp(a) with LA remodeling differed by racial/ethnic group (Tables S5). compared with White participants (reference group), Hispanic individuals exhibited a significant association between higher Lp(a) levels and adverse LA remodeling indices, while no significant associations were observed in Black or Chinese participants (Table S5). When stratified by race/ethnicity, higher Lp(a) levels were significantly associated with adverse LA remodeling indices only in Hispanic participants (Table 3), while no significant associations between Lp(a) and longitudinal changes in LA remodeling indices were observed in White, Black, or Chinese participants (Table 3; Table S6).

Table 3.

Association of Lp(a) With Changes in LA Remodeling Indices Over a Decade Stratified by Race and Ethnicity Groups

In multivariable-adjusted analyses controlling for sociodemographic and cardiovascular risk factors, Lp(a) was linked with an increase in left atrial minimum volume index (per SD log Lp(a) β, 0.81 [95% CI, 0.09–1.52]) and a decrease in LAEF (per SD log Lp(a) β, −1.17 [95% CI, −2.09 to −0.24]; Table 3). When Lp(a) was analyzed as a categorical variable, the third Lp(a) tercile was associated with an increase in left atrial minimum volume index (β, 2.55 [95% CI, 0.76–4.35]) and decrease in total LAEF (β, −2.84 [95% CI, −5.16 to −0.52]) compared with the first tercile (Table 3; Figure 3). These associations remained significant even after further adjustment for coronary artery calcium scores, excluding individuals with interim MI, and also adjusting for left atrioventricular coupling index and LVEF (Table 3). Results were similar when Lp(a) was treated as a categorical variable with race-specific cutoffs for terciles (Table S7). Notably, when race-specific terciles were used, the third Lp(a) tercile in White subjects (Lp[a] ≥20 mg/dL) was associated with increased LAEF (β, 2.27 [95% CI, 0.7–3.85]), and increased LA strain (β, 2.36 [95% CI, 0.59–4.13]) compared with the first Lp(a) tercile (Lp[a] <8 mg/dL; Table S7).

Figure 3.

Association of Lp(a) (lipoprotein[a]) terciles with left atrial (LA) remodeling by race/ethnicity. Forest plot showing β coefficients (95% CI) adjusted for sociodemographic and traditional cardiovascular risk factors (model 2) for ΔLAVi_Min and ΔTotal LAEF by second (Lp[a], 10–30 mg/dL) and third (Lp[a] ≥30 mg/dL) Lp(a) terciles, with first Lp(a) tercile as the reference (Lp[a] <10 mg/dL). LAEF indicates left atrial emptying fraction; and LAVi_Min, left atrial minimum volume index.

Discussion

To the best of our knowledge, this is the first study to examine the relationship between Lp(a) levels and longitudinal alterations in cardiac remodeling across a multi-ethnic cohort over a span of a decade. Our findings provide compelling evidence that elevated Lp(a) is associated with adverse longitudinal changes in cardiac structure and function among Hispanic or Latino adults. We found that higher Lp(a) concentrations were associated with progressive increases in LV end-systolic volume and LA minimum volume, as well as declines in LV and LA function, independent of coronary artery calcification and interim MI, suggesting a role for Lp(a) in nonischemic pathways of myocardial remodeling.

Accumulating evidence has linked elevated Lp(a) levels to the incidence and progression of HF across the disease spectrum^11,25–27; however, the mechanistic pathways underlying this association remain incompletely understood. In a study of individuals with established atherosclerotic cardiovascular disease undergoing coronary angiography, elevated levels of Lp(a) and oxidized phospholipids bound to apolipoprotein(a) or apoB-100 (apolipoprotein B-100) were associated with a persistently higher risk of progression from HF stages A or B to symptomatic HF, HF hospitalization, or cardiovascular death.²⁷ Prior work from MESA demonstrated an independent association between higher Lp(a) concentrations and increased interstitial myocardial fibrosis, as reflected by elevated extracellular volume fraction and native T1 times on cardiac MRI.²⁸ These structural alterations are known to increase myocardial stiffness and impair both LV and LA function, providing a plausible biological substrate for adverse cardiac remodeling. The present study extends these findings by further elucidating the longitudinal pathway linking elevated Lp(a) to adverse cardiac remodeling and dysfunction in the absence of MI, potentially mediated through fibrosis, inflammation, and vascular–myocardial interactions.

Multiple studies have demonstrated that LVESVi is a stronger predictor of mortality and HF–related hospitalizations than LV ejection fraction or LV end-diastolic volume, including among individuals with preserved ejection fraction.^29–33 The prognostic value of LVESVi likely reflects its direct dependence on myocardial contractility and afterload, in contrast to LV end-diastolic volume, which is largely governed by preload through the Frank–Starling mechanism.³⁴ Similarly, abnormal atrioventricular coupling has been implicated not only in HF development but also in atrial fibrillation and adverse cardiovascular events in both clinical and population-based studies.^24,35 Our prior MESA analyses demonstrated a cross-sectional association between higher Lp(a) levels and LA enlargement and dysfunction, supporting a pathway in which interstitial myocardial fibrosis leads to increased myocardial stiffness, impaired atrioventricular coupling, and early LA remodeling.²⁸ Such changes are increasingly recognized as key precursors in the progression toward HF with preserved ejection fraction.³⁶ The current study corroborates and extends these observations and provides important mechanistic insights into how Lp(a) may contribute to the development of HF by demonstrating longitudinal associations between elevated Lp(a) levels and maladaptive cardiac remodeling, including increases in LVESVi and LA minimum volume and declines in LV and LA function as subclinical phenotypes in the HF continuum.^37,38

Racial and ethnic differences in circulating Lp(a) concentrations are well established, with marked variation in both absolute levels and distributional patterns across groups.^8,39,40 Black individuals generally exhibit the highest Lp(a) concentrations and a more normally distributed profile, whereas non-Hispanic White and Hispanic or Latino populations demonstrate positively skewed distributions with lower median levels.⁷ These differences are largely attributable to genetic variation in LPA, including isoform size heterogeneity and single-nucleotide polymorphisms within and surrounding the gene locus.⁴¹ However, race/ethnicity represents complex constructs that encompass both biological factors, such as genetic ancestry, and socioeconomic and acculturation-related factors, and the extent to which each contributes to observed disparities in Lp(a)-related cardiovascular outcomes remains incompletely understood. This complexity is particularly salient in Hispanic or Latino populations, which comprise a highly heterogeneous group with substantial variation in genetic ancestry that has been shown to influence Lp(a) distributions. ⁴²

Prior findings from MESA further underscore racial and ethnic heterogeneity in the cardiovascular consequences of elevated Lp(a).^8–11 Guan et al reported that while Black participants had the highest absolute Lp(a) levels, the strength of the association between elevated Lp(a) and incident coronary heart disease varied by race/ethnicity. Notably, when applying a clinically relevant Lp(a) threshold of 50 mg/dL, the association with coronary heart disease was strongest among Hispanic individuals, exceeding that observed among Black and White participants.⁸ In addition, another MESA investigation examining the systemic vascular effects of Lp(a) identified significant associations between elevated Lp(a) levels and peripheral arterial disease, specifically among Hispanic participants.¹⁰ Such vascular effects may plausibly contribute to the cardiac remodeling patterns observed in our study, as increased arterial stiffness and afterload can promote progressive LA enlargement. These observations support the hypothesis that Lp(a)-mediated vascular pathology in Hispanic individuals may extend beyond atherosclerosis to influence myocardial remodeling through mechanisms related to vascular stiffness, myocardial fibrosis, and altered vascular–ventricular coupling. Emerging data from the Hispanic Community Health Study/Study of Latinos further demonstrate more than a 2-fold higher risk of MI at lower Lp(a) thresholds (75 nmol/L ≈30 mg/dL), reinforcing the concept that clinically meaningful Lp(a)-associated risk may occur at substantially lower levels in Hispanic populations.⁴³

The above findings suggest that higher Lp(a) levels may confer disproportionate cardiovascular risk among Hispanic populations despite relatively low median Lp(a) concentrations overall. Importantly, our findings extend this growing body of evidence by demonstrating associations between Lp(a) and subclinical cardiac structural and functional changes at lower Lp(a) levels (≥30 mg/dL in the main analysis and ≥23 mg/dL in the race-specific analysis) than current guideline-directed cutoffs, thereby strengthening the rationale for consideration of race-specific Lp(a) thresholds in risk assessment. In contrast to Hispanic participants, we did not observe significant associations between Lp(a) levels and cardiac remodeling among Black subjects, despite this group having the highest absolute Lp(a) concentrations. This may reflect complex gene–environment interactions that differ across racial and ethnic groups, downstream biological susceptibility, competing risk pathways, or limited power to detect modest subclinical effects within race-stratified analyses rather than an absence of Lp(a)-related cardiovascular effect per se. Collectively, these findings underscore the challenges of interpreting Lp(a)-associated risk across diverse populations and highlight the need for more nuanced approaches that integrate genetic ancestry, social context, and population-specific distributions when evaluating Lp(a)-related cardiovascular risk.

Several limitations should be considered when interpreting our findings. First, subgroup and race-stratified analyses were constrained by sample size, which may have limited statistical power to detect associations in some racial or ethnic groups. Second, although we adjusted for a comprehensive set of sociodemographic and traditional cardiovascular risk factors, residual confounding from unmeasured or imperfectly measured variables, including social, environmental, or ancestry-related factors, cannot be excluded. Third, the cardiac remodeling indices were available at only 2 time points, and therefore, we opted to use multivariable linear regression models to assess longitudinal change rather than linear mixed-effects models, which are better suited for settings with more frequent repeated measures. Although this approach has been used in prior MESA investigations of cardiac remodeling,^2,44 it does not fully account for within-person correlation over time and assumes a linear change between the 2 imaging examinations, which may oversimplify the true remodeling trajectory. Fourth, informative dropout over the extended follow-up period and the possibility of differential loss to follow-up by race or ethnicity may have introduced selection bias, potentially influencing the observed associations. Finally, as the MESA cohort consists of community-dwelling adults without clinical cardiovascular disease at baseline, our findings may not be generalizable to populations with established cardiovascular disease or to groups outside the demographic composition of MESA.

In conclusion, Hispanic or Latino participants demonstrated a disproportionate susceptibility to maladaptive cardiac remodeling associated with Lp(a). Our findings underscore the limitations of treating race or ethnicity as homogeneous or purely biological constructs in cardiovascular research and emphasize the need for population-specific approaches to Lp(a) risk assessment and interpretation, and suggest that Lp(a)-associated cardiovascular risk may extend beyond atherosclerotic disease to include subclinical myocardial injury and remodeling. As therapies targeting Lp(a) reduction continue to advance, future studies should evaluate their impact on cardiac structure, myocardial fibrosis, and HF risk, particularly in populations that may be uniquely vulnerable, such as Hispanic individuals. Integrating genetic ancestry, social determinants of health, gene–environment interactions, and longitudinal imaging phenotypes will be critical to refining risk stratification and informing precision prevention strategies across diverse racial and ethnic groups.

ARTICLE INFORMATION

Acknowledgments

The authors thank the other investigators, the staff, and the participants of the MESA study (Multi-Ethnic Study of Atherosclerosis) for their valuable contributions. A full list of participating MESA investigators and institutions can be found at http://www.mesa-nhlbi.org. The views expressed in this article are those of the authors and do not necessarily represent the views of the National Heart, Lung, and Blood Institute; the National Institutes of Health; or the US Department of Health and Human Services. This article has been reviewed and approved by the MESA Publications and Presentations Committee. All authors declare that the submitted work is original and has not been published and is not under consideration for publication elsewhere. The views expressed in this article are those of the authors and do not necessarily represent the views of the National Heart, Lung, and Blood Institute; the National Institutes of Health; or the US Department of Health and Human Services.

Disclosures

None.

Supplemental Material

Tables S1–S7

Supplementary Material

hcq-19-e013261-s001.pdf ^{(518.7KB, pdf)}

hcq-19-e013261-s002.pdf ^{(209.4KB, pdf)}

Funding Statement

This research was supported by contracts 75N92025D00022, 75N92020D00001, HHSN268201500003I, N01-HC-95159, 75N92025D00026, 75N92020D00005, N01-HC-95160, 75N92020D00002, N01-HC-95161, 75N92025D00024, 75N92020D00003, N01-HC-95162, 75N92025D00027, 75N92020D00006, N01-HC-95163, 75N92025D00025, 75N92020D00004, N01-HC-95164, 75N92025D00028, 75N92020D00007, N01-HC-95165, N01-HC-95166, N01-HC-95167, N01-HC-95168 and N01-HC-95169 from the National Heart, Lung, and Blood Institute, and by grants UL1-TR-000040, UL1-TR-001079, and UL1-TR-001420 from the National Center for Advancing Translational Sciences (NCATS).