Lipoprotein(a) and acute myocardial infarction: Differential risk across age and sex subgroups

Abstract

Objective

Lipoprotein(a) is an identified risk factor for acute myocardial infarction; however, its role in specific subgroups is inconclusive. This case–control study examined whether classical risk factors modify the association between lipoprotein(a) levels and acute myocardial infarction across subgroups.

Methods

Our study involved 2946 patients with initial acute myocardial infarction and 14,571 controls. Data were processed using multiple imputation and propensity score-based matching. Binary logistic regression, stratification, and sensitivity analyses were performed to explore associations.

Results

The acute myocardial infarction group had a significantly greater proportion of men, smokers, and alcohol consumers as well as higher lipoprotein(a) levels. After risk factor adjustment in Model 2, the odds ratios and 95% confidence intervals for lipoprotein(a) (Quartile 2–Quartile 4) were 1.11 (0.97, 1.28), 1.24 (1.07, 1.42), and 1.45 (1.26, 1.66), respectively. In older patients, the adjusted odds ratio for acute myocardial infarction in Quartile 4 (vs. Quartile 1) was 2.15 (1.76–2.63)—higher than that in younger patients. Among female participants, the odds ratio for Quartile 4 was 2.61 (1.98–3.45), exceeding that in men. Compared with the younger/male group, the older/female group exhibited escalating acute myocardial infarction risks (adjusted odds ratios: Quartile 2 = 1.48 (1.00–2.18); Quartile 3 = 1.97 (1.38–2.81); and Quartile 4 = 2.88 (2.04–4.05)), culminating in a 2.88-fold elevation in Quartile 4. Repeat analyses of the complete dataset yielded the same results.

Conclusions

High lipoprotein(a) levels independently increased the risk of initial acute myocardial infarction in older women, suggesting its role as a lipid-lowering therapy target in this population.

Introduction

Acute myocardial infarction (AMI), the ultimate manifestation of coronary artery disease (CAD), remains a major reason of disability and sudden death, primarily due to coronary artery occlusion. Despite significant advancements in medical care and the adoption of advanced cardiovascular technologies in China, the prevalence and fatality ratios of AMI have exhibited an upward trend over the past two decades. ¹ Numerous studies have identified diverse risk factors for cardiovascular diseases, including biological factors, such as hypertension, dyslipidemia, creatinine abnormality, type 2 diabetes mellitus (T2DM), and tobacco addiction as well as social determinants such as marital status and educational level.^2–12

The function of lipoprotein(a) (Lp(a)) as a residual-risk factor for CAD has aroused increasing concern. Lp(a) is a liver-synthesized, low-density lipoprotein-like structure consisting of an apolipoprotein B-100 (apo B-100) linked via covalent bonds to a major glycoprotein called apolipoprotein(a) (apo (a)). ¹³ Genetic influences account for approximately 90% of plasma Lp(a) levels. These levels display considerable variations among distinct ethnic groups and maintain a relatively unchanging state across a person’s lifespan. ¹⁴ The LPA gene is a primary determinant of Lp(a) levels, while dietary and environmental factors exert only minimal effects.^15–18

Higher Lp(a) level is a widely accepted risk factor for atherosclerotic cardiovascular diseases (ASCVD), supported by Mendelian randomization assessments, epidemiologic analyses, and genome-wide association explorations. ¹⁹ The 2018 Clinical Practice Guideline for Cholesterol Management has suggested the use of Lp(a) as a risk evaluation booster in those with ASCVD risk, complementing evaluations based on classical risk factors. ²⁰ A study involving seven ethnic groups has revealed that elevated Lp(a) levels independently augment the risk of AMI, irrespective of established risk elements. In a Chinese cohort, Lp(a) level was the lowest, whereas isoform size was the largest. However, distinct characteristics, combined with the relatively low number and proportion of Chinese women in the study, may limit the universality of the results. ²¹ Emerging genetic studies have further confirmed the relationship between increased Lp(a) levels and ASCVD. Notably, treatment measures to reduce Lp(a) levels may mitigate the ASCVD risk.^22,23 These research advancements have driven the development of innovative remedies to reduce Lp(a) levels. Promisingly, several Lp(a)-targeting therapies are currently undergoing clinical trials in cardiovascular disease patients and are anticipated to enter the market in the near future. Thus, a pivotal research question is to identify patient subgroups that may derive the greatest benefit from prompt reduction in Lp(a) levels, particularly when other modifiable risk factors are effectively controlled. ²⁴

Considering that the current understanding of the role of Lp(a) is chiefly founded on genetic bases and prospective investigations, real-world clinical data remain scarce. Therefore, we conducted a retrospective clinical study to explore whether traditional risk factors modulate the Lp(a)-related risk of initial AMI in a large-sample Chinese cohort. Our study aimed to provide evidence for AMI prophylactic and management plans.

Materials and methods

Patients

This retrospective case–control study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. ²⁵ The electronic medical records of consecutive inpatients at the Department of Cardiology at Taixing People’s Hospital between May 2014 and June 2023 were reviewed. The medical records of all inpatients were anonymized via removal of all identifying details. Given the retrospective design of the investigation, the Institutional Review Board (IRB) dispensed with the need for informed consent. The IRB of the hospital sanctioned the study plan. The present research adheres strictly to the Declaration of Helsinki (1975, as revised in 2024).

To minimize the confounding factors in the study, patients who met the following criteria were not included: age <18 years, history of myocardial infarction, aortic dissection, pulmonary embolism, thyroid dysfunction, chronic nephropathy (defined by creatinine level >442 µmol/L or estimated glomerular filtration rate <60 mL/min/1.73 m²), severe acute infections, and uncontrolled hyperglycemia (blood glucose (BG) >22.2 mmol/L).

Clinical and laboratory data

General information was obtained from digital archives, including inpatient clinical demographics such as age, sex, marital status, educational level, lifestyle determinants (tobacco use pattern and alcohol consumption level), medical history (hypertensive disorder and diabetes mellitus), and cardiopulmonary signs (arterial pressure and pulse rate). Blood specimens were typically obtained on the morning after the day of admission, following an 8-h fast. Laboratory data included coagulation function indicators, such as D-dimer, creatinine, urea nitrogen, uric acid, and BG levels, and lipid profile indices. Biochemical markers were measured using the Beckman AU5400 and ACL TOP700 analyzers. Calibration and quality control procedures were performed strictly according to the established laboratory protocols to ensure the fidelity and stability of the test outputs.

Data on marital status, educational level, smoking status, and alcohol consumption were missing for 1.27%, 1.49%, 0.76% and 0.84% of the study participants, respectively. Systolic blood pressure (SBP) data were missing for 1.48%, and diastolic blood pressure (DBP) data were missing for 1.48% of the participants. Missing data were also observed for heart rate (HR, 3.55%); blood urea nitrogen (BUN), uric acid (UA), and serum creatinine (Scr) levels (2.67%); D-dimer levels (10.91%); prothrombin time (PT, 9.08%); activated partial thromboplastin time (APTT, 9.96%); thrombin time (TT, 10.42%); BG level (8.60%); total cholesterol (TC) and triglyceride (TG) levels (7.09%); low-density lipoprotein cholesterol (LDL-C) level (7.11%); high-density lipoprotein cholesterol (HDL-C, 7.14%); and apolipoprotein A1 (apoA1), apolipoprotein B (apoB), and Lp(a) levels (7.24%). No other variables had missing data.

Definitions and diagnoses

Initial AMI was defined as the primary occurrence of AMI diagnosed at our institution, in line with the universal definition of myocardial infarction, with no prior episode of myocardial infarction.

Primary AMI was defined as the first documented episode of AMI diagnosed at our institution, in accordance with the universal definition of myocardial infarction.

Patients without AMI were included in a reference group for comparison, given the unavailability of healthy controls. Thyroid dysfunction included hyperthyroidism, hypothyroidism, presence of thyroid nodules, and other thyroid-related diseases. Hypertension was defined as SBP ≥140 mmHg and/or DBP ≥90 mmHg, measured on 3 separate occasions in a clinical setting, without antihypertensive medication use. Furthermore, according to the definition, we also included patients with a history of hypertension who were on antihypertensive therapy. T2DM was diagnosed using a similar definition. Individuals who had consumed alcohol at least once a week during the preceding 12 months were characterized as alcohol consumers. Those who had smoked continuously or cumulatively for at least 6 months during their lifetime were considered smokers.

Statistical analyses

All continuous variables considerably deviating from normality were assessed using the Kolmogorov–Smirnov test for normality. Continuous variables were reported as medians along with interquartile ranges (IQRs) and subjected to analysis using the Wilcoxon rank-sum test. Nominal variables were displayed as counts (percentages) and examined using Pearson’s chi-square test.

To reduce biases due to the inadequate handling of missing data, multiple imputation was employed. This method obtained reasonable estimates for incomplete data, relying on the pattern of the available data, allowing the use of more data points. The presumption of data being missing at random (MAR) is a prerequisite for this imputation approach. Multiple imputation by chained equations (MICEs) provided a practical approach for generating imputations using a set of imputation models, with separate models for each variable. This method specified the conditional models for all variables with missing data. Predictive mean matching (pmm) was employed for quantitative continuous factors, such as laboratory data. Logistic regression (logreg) was applied to imputing binary variables, such as smoking status and alcohol consumption. Polytomous logistic regression (polyreg) was employed for nominal variables, such as marital status. Proportional odds logistic regression (polr) was utilized for ordinal variables, such as educational level. According to Rubin’s suggestion, 5 imputed datasets (m = 5) were created in this study and used for calculating the propensity score and for further analyses. ²⁶

The propensity score-based matching (PSM) approach was utilized to balance the covariates. The matching criteria included age, sex, history of hypertension, history of T2DM, marital status, and educational level. Initial AMI patients and controls were matched in a 1:2 ratio using the PSM method. Matching was performed using the “MatchIt” package in R Studio. The nearest neighbor algorithm was employed, and the matching was set to be without replacement; a caliper value of 0.2 was applied. ²⁷ Subsequently, in-depth analyses were conducted for each imputed dataset, and the outcomes were combined according to Rubin’s rule.

Lp(a) levels were categorized into quartiles and included in logistic regression models. Quartile 1 (Q1) was used as reference set, and for Q2, Q3, and Q4 relative to Q1, the odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated. In the context of these matched case–control datasets, logistic regression models were constructed for patients with and without initial AMI. In the unadjusted model, no covariates were considered. Model 1 was adjusted for age, sex, smoking habit, hypertension, T2DM level, and LDL-C level. Model 2 was further adjusted for Scr, D-dimer, HDL-C, and apoB levels; SBP; and Model-1 covariates. Logistic regression Model 3 was constructed by adjusting for Model 2 and all remaining covariates.

In addition, we calculated the medians for age and LDL-C, HDL-C, TG, TC, apoA1, and apoB levels to divide patients into low (≤median) and high (>median) groups. Stratified analyses were subsequently conducted based on these medians and the following variables: (a) sex (male = 1, female = 0); (b) hypertension (yes = 1, no = 0); (c) T2DM (yes = 1, no = 0); (d) smoking status (smoker = 1, nonsmoker = 0); (e) alcohol consumption status (drinker = 1, nondrinker = 0); (f) educational level (YOE) (0 years = 0, 1–6 years = 1, 7–9 years = 2, 10–12 years = 3, >12 years = 4); (g) and marital status (MS) (married = 1, unmarried = 2, divorced = 3, widowed = 4). Then, based on the median values for age and sex, the patients were stratified into the following four groups for further analyses: (a) younger/male; (b) younger/female; (c) older/male; and (d) older/female. The primary analysis was performed using multiple imputation. Sensitivity analyses were conducted using complete cases without PSM.

We used R software (version 4.4.1, R Foundation for Statistical Computing in Vienna, Austria) and Statistical Package for Social Sciences (SPSS) software (version 26.0, SPSS, Inc., located in Chicago, Illinois) to perform all statistical analyses. Statistical significance was rigorously determined based on a two-tailed p-value <0.05.

Results

Clinical characteristics

In total, 20,018 consecutive hospitalized patients were initially considered for analyses. After applying the exclusion criteria, 2501 patients were excluded, resulting in a final cohort of 17,517 patients, including 2946 with initial AMI and 14,571 controls. The selection process for study participants is depicted in Figure 1. Characteristics of the initial AMI patients and controls are displayed in Table 1. The median age of the study population was 69 years (IQR: 62–76). Compared with controls, the initial AMI group demonstrated significantly higher prevalence of traditional cardiovascular risk factors, including male sex (74.98% vs. 60.27%), smoking status (46.19% vs. 22.87%), and alcohol consumption (29.56% vs. 18.79%). Notably, Lp(a) levels were markedly higher in initial AMI patients than in controls (median (IQR): 14.70 (7.30–29.00) vs. 11.60 (5.70–23.90) mg/dL, p < 0.001). Similarly, participants with initial AMI had higher hemoglobin (Hb) levels and elevated levels of D-dimer, PT, BG, TC, TG, LDL-C, and apoB. Figure 2 illustrates a distinct distribution pattern of Lp(a) levels.

Figure 1.

Study flow diagram for patient selection. eGFR: estimated glomerular filtration rate; Scr: serum creatinine; BG: blood glucose; AMI: acute myocardial infarction.

Table 1.

Clinical characteristics and laboratory parameters.

Characteristics	Missing	Total (n = 17,517)	Control (n = 14,571) (83.18%)	Initial AMI (n = 2946) (16.82%)
N (%)	n (%)	Median (Q1, Q3)	Median (Q1, Q3)	Median (Q1, Q3)	p-values
Age, year	17,517 (0.00%)	69 (59, 77)	69 (60, 77)	69 (57, 77)	0.122
Scr, µmol/L	17,050 (2.67%)	70.90 (58.50, 88.00)	70.80 (58.40, 87.80)	71.40 (59.40, 88.60)	0.050
UA, µmol/L	17,049 (2.67%)	370.80 (305.00, 454.80)	371.75 (305.20, 458.83)	366.10 (303.60, 438.00)	<0.001
BUN, mmol/L	17,050 (2.67%)	6.26 (5.10, 7.90)	6.26 (5.10, 7.90)	6.27 (5.00, 7.87)	0.332
D-dimer, ng/mL	15,606 (10.91%)	178 (116, 323)	172 (113, 312)	210 (134, 384)	<0.001
PT, s	15,927 (9.08%)	11.50 (10.80, 12.40)	11.40 (10.70, 12.30)	11.90 (11.20, 12.80)	<0.001
APTT, s	15,772 (9.96%)	32.00 (29.50, 35.20)	32.10 (29.80, 35.00)	31.10 (25.70, 38.10)	<0.001
TT, s	15,692 (10.42%)	13.90 (13.00, 14.80)	14.00 (13.20, 14.90)	12.80 (0.00, 14.20)	<0.001
BG, mmol/L	16,012 (8.59%)	5.65 (4.92, 6.92)	5.52 (4.87, 6.68)	6.45 (5.47, 8.17)	<0.001
TC, mmol/L	16,276 (7.08%)	4.09 (3.40, 4.81)	4.02 (3.35, 4.74)	4.44 (3.80, 5.13)	<0.001
TG, mmol/L	16,276 (7.08%)	1.11 (0.78, 1.65)	1.10 (0.77, 1.62)	1.19 (0.81, 1.80)	<0.001
LDL-C, mmol/L	16,271 (7.11%)	2.29 (1.77, 2.85)	2.23 (1.73, 2.79)	2.60 (2.10, 3.13)	<0.001
Lp(a), mg/dL	16,249 (7.24%)	12.10 (6.00, 24.80)	11.60 (5.70, 23.90)	14.70 (7.30, 29.00)	<0.001
HDL-C, mmol/L	16,267 (7.14%)	1.11 (0.92, 1.34)	1.11 (0.92, 1.35)	1.08 (0.90, 1.28)	<0.001
apoB, g/L	16,248 (7.24%)	0.82 (0.65, 1.01)	0.80 (0.63, 0.99)	0.93 (0.76, 1.11)	<0.001
apoA1, g/L	16,248 (7.24%)	1.12 (0.95, 1.31)	1.13 (0.95, 1.32)	1.09 (0.93, 1.27)	<0.001
HR, bpm	16,896 (3.55%)	78 (67, 90)	77 (67, 89)	80 (70, 92)	<0.001
SBP, mmHg	17,258 (1.48%)	135 (120, 150)	136 (121, 151)	126 (111, 142)	<0.001
DBP, mmHg	17,257 (1.48%)	80 (71, 90)	80 (71, 90)	78 (69, 88)	<0.001
T2DM	17,517 (0.00%)				0.218
0		14,184 (81.08%)	11,823 (81.14%)	2361 (80.14%)
1		3333 (19.02%)	2748 (18.86%)	585 (19.86%)
HTN	17,517 (0.00%)				<0.001
0		8320 (47.50%)	6778 (46.52%)	1542 (52.34%)
1		9197 (52.50%)	7793 (53.48%)	1404 (47.66%)
Male	17,517 (0.00%)
0		6526 (37.25%)	5789 (39.73%)	737 (25.02%)	<0.001
1		10,991 (62.75%)	8782 (60.27%)	2209 (74.98%)
Smoking status	17,384 (0.76%)				<0.001
0		12,733 (73.24%)	11,174 (77.13%)	1559 (53.81%)
1		4651 (26.76%)	3313 (22.87%)	1338 (46.19%)
Alcohol consumption	17,370 (0.84%)				<0.001
0		13,795 (79.42%)	11,758 (81.21%)	2037 (70.44%)
1		3575 (20.58%)	2720 (18.79%)	855 (29.56%)
Educational level (YOE)	17,256 (1.49%)				<0.001
0		2812 (16.29%)	2331 (16.22%)	481 (16.70%)
1		5012 (29.04%)	4146 (28.84%)	866 (30.06%)
2		5493 (31.83%)	4578 (31.84%)	915 (31.76%)
3		2423 (14.04%)	1999 (13.90%)	424 (14.72%)
4		1516 (8.78%)	1321 (9.19%)	195 (6.77%)
MS	17,294 (1.27%)				0.119
1		16,909 (97.78%)	14,101 (97.87%)	2808 (97.26%)
2		285 (1.65%)	225 (1.56%)	60 (2.08%)
3		46 (0.27%)	35 (0.24%)	11 (0.38%)
4		54 (0.31%)	46 (0.32%)	8 (0.28%)

AMI: acute myocardial infarction, Scr: serum creatinine, UA: uric acid, BUN: blood urea nitrogen, PT: prothrombin time, APTT: activated partial thromboplastin time, TT: thrombin time, BG: blood glucose, TC: total cholesterol, TG: triglyceride, LDL-C: low-density lipoprotein cholesterol, Lp(a): lipoprotein(a), HDL-C: high-density lipoprotein cholesterol, apoA1 : apolipoprotein A1, apoB: apolipoprotein B, HR: heart rate, SBP: systolic blood pressure, DBP: diastolic blood pressure, T2DM: type 2 diabetes mellitus, HTN: hypertension, YOE: years of education, MS: marital status; Q1: quartile 1; Q2: quartile 2; Q3: quartile 3; Q4: quartile 4.

Figure 2.

A histogram for the distribution of lipoprotein(a) concentrations. The x-axis represents the Lp(a) concentrations in mg/dL, ranging from 0 to 150, and the y-axis represents the percentage of the population.

Multiple imputation and propensity score-matching

Five imputed datasets (m = 5) were created in this study, and the PSM approach was employed to achieve covariate balance. The matching variables included age, sex, history of hypertension, history of T2DM, MS, and educational level. The characteristics of the study population after multiple imputation and PSM are presented in Table 2. After matching, the characteristics of the initial AMI and control groups were balanced across all matched factors.

Table 2.

Clinical characteristics and laboratory parameters between initial AMI patients and controls after multiple imputation and PSM.

Characteristics	Controls (n = 5886) (66.65%)	Initial AMI (n = 2945) (33.35%)	p-values
N (%)	Median (Q1, Q3)	Median (Q1, Q3)
Age, year	69 (58, 77)	69 (57, 77)	0.3633
Scr, µmol/L	72.36 (60.84, 89.92)	71.4 (59.5, 88.61)	0.0288
UA, µmol/L	382.25 (313.27, 467.74)	365.62 (303.42, 437.03)	<0.001
BUN, mmol/L	6.49 (5.30, 8.16)	6.26 (5.00, 7.87)	<0.001
D-dimer, ng/mL	178.3 (118.40, 330.05)	216.2 (136.00, 410.60)	<0.001
PT, s	11.5 (10.90, 12.50)	11.9 (11.12, 12.76)	<0.001
APTT, s	32.2 (29.9, 35.08)	31.52 (26.38, 39.08)	<0.001
TT, s	13.9 (13.1, 14.8)	12.94 (0.00, 14.32)	<0.001
BG, mmol/L	5.59 (4.97, 6.68)	6.49 (5.48, 8.28)	<0.001
TC, mmol/L	3.97 (3.29, 4.72)	4.42 (3.77, 5.11)	<0.001
TG, mmol/L	1.11 (0.80, 1.63)	1.17 (0.81, 1.78)	0.002
LDL-C, mmol/L	2.20 (1.69, 2.76)	2.59 (2.08, 3.12)	<0.001
Lp(a), mg/dL	11.68 (5.80, 23.65)	14.84 (7.44, 29.37)	<0.001
HDL-C, mmol/L	1.11 (0.92, 1.34)	1.07 (0.90, 1.27)	<0.001
apoB, g/L	0.78 (0.62, 0.99)	0.92 (0.75, 1.11)	<0.001
apoA1, g/L	1.15 (0.98, 1.34)	1.08 (0.93, 1.26)	<0.001
HR, bpm	78 (67, 90)	80 (70, 92)	<0.001
SBP, mmHg	135 (120, 150)	126 (111, 142)	<0.001
DBP, mmHg	80 (71, 90)	78 (69, 88)	<0.001
T2DM			0.1754
0	4794 (81.5%)	2360 (80.10%)
1	1091 (18.5%)	585 (19.90%)
HTN			0.3822
0	3018 (51.30%)	1541 (52.30%)
1	2867 (48.70%)	1404 (47.70%)
Male			0.3949
0	1423 (24.20%)	737 (25.00%)
1	4463 (75.80%)	2208 (75.00%)
Smoking status			0.005
0	4140 (70.30%)	1584 (53.80%)
1	1746 (29.70%)	1360 (46.20%)
Alcohol consumption			0.005
0	4498 (76.40%)	2073 (70.40%)
1	1387 (23.60%)	872 (29.60%)
Educational level (YOE)			0.5928
0	934 (15.90%)	490 (16.60%)
1	1854 (31.50%)	884 (30.00%)
2	1861 (31.60%)	939 (31.90%)
3	840 (14.30%)	432 (14.70%)
4	396 (6.70%)	199 (6.80%)
MS			0.0572
1	5780 (98.20%)	2866 (97.30%)
2	83 (1.40%)	61 (2.10%)
3	14 (0.20%)	10 (0.30%)
4	8 (0.10%)	8 (0.30%)

AMI: acute myocardial infarction; PSM: propensity score-matching; Scr: serum creatinine; UA: uric acid; BUN: blood urea nitrogen; PT: prothrombin time; APTT: activated partial thromboplastin time; TT: thrombin time; BG: blood glucose; TC: total cholesterol; TG: triglyceride; LDL-C: low-density lipoprotein cholesterol; Lp(a): lipoprotein(a); HDL-C: high-density lipoprotein cholesterol; apoA1 :  apolipoprotein A1; apoB: apolipoprotein B; HR: heart rate; SBP: systolic blood pressure; DBP: diastolic blood pressure; T2DM: type 2 diabetes mellitus; HTN: hypertension; YOE: years of education; MS: marital status; Q1: quartile 1; Q2: quartile 2; Q3: quartile 3; Q4: quartile 4.

Logistic regression analyses

Based on four Lp(a) quartiles, we performed binary logistic regression analysis to investigate the association between Lp(a) levels and initial AMI (Table 3). In the unadjusted model, for Lp(a) Q2–Q4, the ORs and 95% CIs were 1.24 (1.08, 1.42), 1.46 (1.28, 1.68), and 1.85 (1.62, 2.11), respectively. Subsequent to the adjustment for age; sex; smoking status; T2DM; hypertension; LDL-C, Scr, D-dimer, HDL-C, and apoB levels; SBP; and alcohol consumption status in Model 2, the ORs and 95% CIs for Lp(a), Q2–Q4, were 1.11 (0.97, 1.28), 1.24 (1.07, 1.42), and 1.45 (1.26, 1.66), respectively. Similar results were observed for Model 1 and Model 3. Notably, the ORs for Q3–Q4 showed statistically significant differences in the unadjusted model as well as in Models 1–3. Even after accounting for potential confounding factors, the increasing trend remained consistent.

Table 3.

Multivariate logistic regression analyses of the risks of initial AMI.

Characteristics	OR (95% CI)p-value
	Unadjusted model	Model 1	Model 2	Model 3
Lp(a) Q1	Reference	Reference	Reference	Reference
Lp(a) Q2	1.24 (1.08, 1.42)0.036	1.15 (0.99, 1.32)0.055	1.11 (0.97, 1.28)0.138	1.11 (0.96, 1.29)0.156
Lp(a) Q3	1.46 (1.28, 1.68)0.005	1.28 (1.11, 1.46)<0.001	1.24 (1.07, 1.42)0.003	1.26 (1.08, 1.46)0.003
Lp(a) Q4	1.85 (1.62, 2.11)<0.001	1.53 (1.33, 1.75)<0.001	1.45 (1.26, 1.66)<0.001	1.36 (1.16, 1.59)<0.001

The unadjusted model was not adjusted for any covariates. Model 1 was adjusted for age, sex, smoking habit, hypertension, T2DM level, and LDL-C level. Model 2 was adjusted for Scr, D-dimer, HDL-C, and apoB levels as well as SBP, alcohol consumption, and Model-1 covariates. Model 3 was adjusted for all remaining covariates plus Model-2 covariates.

AMI: acute myocardial infarction; CI: confidence interval; Lp(a): lipoprotein(a); T2DM: type 2 diabetes mellitus; LDL-C: low-density lipoprotein cholesterol; Scr: serum creatinine; HDL-C: high-density lipoprotein cholesterol; apoB: apolipoprotein B; SBP: systolic blood pressure; OR: odds ratio; Q1: quartile 1; Q2: quartile 2; Q3: quartile 3; Q4: quartile 4.

Stratified analyses

In addition, stratified analyses of the associations between Lp(a) and other AMI-related risk factors were conducted. The stratified analyses by age and sex are presented in Figure 3. Based on logistic regression Model 2, higher Lp(a) levels were associated with an increased risk of initial AMI in the study population. In Model 2, the adjusted OR for initial AMI was 1.63 (95% CI, 1.35–1.96) for Q4 (reference: Q1) in younger patients; however, in older patients, the adjusted OR for Q4 was 2.15 (95% CI, 1.76–2.63). Unexpectedly, the adjusted OR for Q4 in women was 2.61 (95% CI, 1.98–3.45), which was higher than the adjusted OR of 1.65 (95% CI, 1.42–1.91) in men. Furthermore, the adjusted ORs for Q2–Q4 in older/female participants were 1.48 (95% CI, 1.00–2.18), 1.97 (95% CI, 1.38–2.81), and 2.88 (95% CI, 2.04–4.05), respectively (Figure 4).

Figure 3.

Odds ratios (95% confidence intervals) for AMI across lipoprotein(a) quartiles, stratified by age and sex, adjusted for Model 2.

Figure 4.

Odds ratios (95% confidence intervals) for AMI across lipoprotein(a) quartiles stratified by younger/male, younger/female, older/male, and older/female and adjusted for Model 2.

Sensitivity analyses

Sensitivity analyses using complete-case data (after excluding all missing values) produced results similar to our primary analysis (data not shown).

Discussion

The major findings of this study are as follows. First, the initial AMI group had a significantly greater proportion of men, smokers, and alcohol consumers than controls, along with markedly higher Lp(a) levels. Second, after adjustment in Model 2 (for age; sex; smoking status; hypertension; T2DM; LDL-C, Scr, D-dimer, HDL-C, and apoB levels; SBP; and alcohol consumption status), the ORs (95% Cls) for Lp(a) Q2–Q4 were 1.11 (0.97, 1.28), 1.24 (1.07, 1.42), and 1.45 (1.26, 1.66), respectively. Third, in Model 2, the adjusted ORs (95% CIs) for initial AMI demonstrated a graded increase across Lp(a) quartiles, with Q4 versus Q1 showing stronger associations in older patients (2.15 (1.76–2.63)) than in younger patients (1.63 (1.35–1.96)). Notably, sex-stratified analyses revealed that this association was particularly pronounced in women (2.61 (1.98–3.45)) compared with that in men (1.65 (1.42–1.91)). Furthermore, in the older/female subgroup, adjusted ORs (95% CIs) for initial AMI showed a striking dose–response relationship across Lp(a) quartiles (Q2–Q4 vs. Q1): 1.48 (1.00–2.18), 1.97 (1.38–2.81), and 2.88 (2.04–4.05), respectively. This culminated in a 2.88-fold elevation in Q4 compared with that in younger men, highlighting a significant disparity. These effect estimates were substantially higher than those observed in other demographic subgroups and the overall population.

Therefore, this investigation offers unique proof that more focus should be directed to older women with increased Lp(a) levels. Current guidelines recognize dyslipidemia as a modifiable driver of ASCVD. However, strong evidence now shows that elevated Lp(a) levels represent a persistent, LDL-C-independent residual-risk factor, which requires targeted interventions in patients treated using statins. A recent study based on the UK Biobank population has shown that the atherogenicity of Lp(a), measured on a per-particle basis, is approximately six times higher than that of LDL-C. The current genetic study provides higher quantitative estimates of Lp(a)’s atherogenicity than previous reports. ²⁸ These findings clearly demonstrate that Lp(a) is an extremely atherogenic lipid–protein complex, suggesting that it represents a viable target for intervention. Such an intervention is likely to reduce clinical risk across a broader range of individuals. A pooled analysis of five large U.S. prospective cardiovascular cohorts has demonstrated an independent association of increased Lp(a) levels with higher myocardial infarction risk, with similar hazard ratios observed in men and women. Limitations included inadequate Asian representation (2.5% of the participants) and reliance on archived frozen samples for Lp(a) assays, potentially introducing storage-related biases. ¹⁹ The Singapore Coronary Artery Disease Genetics Study, involving 2025 participants (61.4% participants of Singaporean Chinese descent), has revealed a significant association between Lp(a) levels and AMI risk, with this association being replicated in our study cohort. ²⁹

Despite mixed evidence from initial reports on the relationship between Lp(a) and CAD risk, a growing number of studies have established Lp(a) as an independent predictor of AMI occurrence in the general population.^30–33

With the ongoing increase in AMI incidence in China, exploring the link between Lp(a) levels and AMI among different populations—particularly those at high ASCVD risk—could yield critical insights for developing targeted preventive strategies.

However, current investigations on the relationship between Lp(a) level and AMI across age and sex groups remain limited and have yielded conflicting conclusions. A case–control study ³⁴ has revealed that Lp(a) is a proven, distinct risk factor for ACS in those aged <45 years, and heightened Lp(a) amounts escalate the ACS risk by approximately three-fold; additionally, the relationship persists but weakens in the 45–60 years age group and ceases to exist in individuals aged >60 years. There were no notable sex-based disparities in the relationship between Lp(a) level and ACS risk. This null finding may be attributed to the smaller overall sample size and limited number of female participants included in the analysis. As demonstrated in the Copenhagen General Population Study, ³⁵ plasma Lp(a) levels rise with advancing age. Specifically, among women, an additional rise was noted around the age of 50 years. In multivariable-adjusted models stratified by sex and age, compared with Lp(a) levels <10 mg/dL (18 nmol/L), those >40 mg/dL (83 nmol/L) were correlated with an elevated risk of myocardial infarction. A Mendelian randomization study has shown that plasma Lp(a) levels predicted by genetic factors have a positive correlation with both existing and new-onset CAD in both men and women, with similar effect sizes across sexes. ³⁶ Conversely, Zeis et al. have reported that drastically increased Lp(a) levels could be a more significant risk augmenter for mortality and ASCVD in women than in men. ³⁷

Given the limited data and inconsistent results regarding the role of Lp(a) levels in ASCVD, we conducted a retrospective study involving an extensive inpatient population to clarify its association with AMI.

Our study revealed significantly elevated Lp(a) levels in patients with initial AMI than in matched controls. In Model 2, the adjusted OR for initial AMI in the highest Lp(a) quartile (Q4 vs. Q1) demonstrated age-dependent patterns: older patients exhibited a higher risk than their younger counterparts (Q4: OR = 2.15 vs. 1.63). Notably, this association was more pronounced in women (Q4 OR = 2.61) than in men (Q4 OR = 1.65). Furthermore, the dose–response relationship across quartiles reached statistical significance, specifically in older/female participants (Q2–Q4: ORs = 1.48–2.88, p < 0.05), with effect estimates exceeding those observed in other age-sex subgroups (Q4: p < 0.05 for interaction). These findings establish an evidence base for implementing standardized Lp(a) screening protocols in older women.

Our study revealed that elevated Lp(a) levels played a role in the onset of AMI, especially in older women. However, the underlying mechanism remains obscure. Possible mechanisms include the following aspects.

First, Lp(a) elevates cardiovascular risk through three synergistic pathways: (a) vascular inflammation through oxidized phospholipid-mediated inflammatory signaling, adhesion molecule upregulation, and cytokine release; (b) endothelial dysfunction through oxidative stress, impaired nitric oxide synthesis, and vascular barrier disruption; and (c) prothrombotic effects exerted via the suppression of fibrinolysis, enhancement of coagulation/platelet activity, and fibrin binding via small apo(a) isoforms. Through these synergistic mechanisms, Lp(a) promotes the development of CAD and AMI.^38,39

Second, women exhibit sex-specific cardiovascular risk profiles that exacerbate ASCVD progression. Sex-specific biological mechanisms modify the impact of conventional risk factors; diabetes and metabolic syndrome confer more pronounced cardiovascular risk in women than in men, while chronic kidney impairment and autoimmune pathologies in women manifest with earlier onset and accelerated progression, promoting atherogenesis. Beyond biological mechanisms, endocrine-related transitional phases introduce unique vulnerabilities; preeclampsia, gestational dysglycemia, and premature ovarian insufficiency demonstrate independent associations with persistent endothelial dysfunction. Polycystic ovary syndrome and postmenopausal metabolic shifts worsen lipid profiles and inflammation. Sociocultural factors also play a role; atypical symptoms delay diagnoses, while higher psychosocial stress and healthcare disparities hinder timely intervention, exacerbating ASCVD progression. ⁴⁰

Third, epidemiological evidence indicates that elevated CVD risk in older women is driven by two interrelated biological pathways. First, postmenopausal transition induces systemic metabolic dysregulation, characterized by visceral adiposity, insulin resistance, and atherogenic dyslipidemia (raised LDL-C and triglyceride levels and reduced HDL-C levels). ⁴¹ These alterations collectively progress to metabolic syndrome, a proven CVD risk amplifier. Second, estrogen deficiency impairs its regulatory function in Lp(a) metabolism, chiefly through the attenuated suppression of apo (a) gene transcription, leading to elevated circulating Lp(a) levels.^42,43 As an independent genetic determinant of atherosclerosis, heightened Lp(a) cooperates with metabolic dysfunction to potentiate vascular pathogenesis. The proposed two-hit mechanism linking metabolic dysfunction to estrogen-related Lp(a) elevation not only clarifies why older women experience more rapid CVD progression than age-matched men but also informs the design of novel combination therapies to simultaneously target metabolic and lipoprotein disorders.

Collectively, emerging evidence indicates that cardiovascular risk profiles undergo dynamic alterations during the menopausal transition, driven by dual pathogenic pathways, the compounding effects of biological aging and endocrino–metabolic remodeling intrinsic to ovarian senescence. Consequently, it is critical to recognize that the pathophysiological trajectories of cardiovascular risk determinants manifest accelerated progression during this transitional phase. This accelerated progression reflects not only universal biological processes of age-related senescence but also distinct pathophysiological alterations specific to the menopausal transition, characterized by a unique endocrine–metabolic shift that potentiates cardiovascular vulnerability.

However, our study has limitations. As Lp(a) functions as a substance involved in the acute-phase response, its levels may change during acute coronary events, potentially affecting reliability. The exclusive inclusion of Chinese participants limits generalizability to other ethnic groups. Measuring Lp(a) level in mg/dL (vs. nmol/L) may reduce cross-study comparability due to apo(a) size heterogeneity. Additionally, potential residual confounding from unmeasured variables such as lipid-lowering therapies may influence the validity of our risk estimates. Finally, combining ST-segment elevation myocardial infarction and non-ST-segment elevation myocardial infarction patients may obscure subtype-specific differences. Future research should examine these groups separately to gain a comprehensive understanding of Lp(a)’s contribution in different types of AMI and refine risk assessment.

Conclusions

Our findings indicate that high levels of Lp(a) independently increase the risk of initial AMI in older women, suggesting its role as a lipid-lowering therapy target in this population.

Data availability statement

The data supporting the findings of this study are available upon request from the corresponding authors, Z.L. and Z.C. However, the data are not publicly accessible as they contain information that could compromise the privacy of the research participants.

Declaration of conflicting interests

The authors declare no conflicts of interest.

Informed consent statement

Patient consent was waived for this retrospective study. As data were collected from existing medical records without direct participant interaction, posed minimal risk, and were fully deidentified, the IRB approved the waiver. Contacting original participants was deemed impractical because a long time had passed since data collection.

IRB statement

The study was conducted in accordance with the Declaration of Helsinki (1975, as revised in 2024), and approved by the Ethics Committee of Taixing People’s Hospital (protocol code: LS2025026 and date of approval: 28 April 2025).