Cardiovascular diseases (CVDs) stand as the predominant agents of morbidity and mortality worldwide, bearing the principal burden upon global health [1]. A significant number of cardiovascular risk factors have been recognized up to date and used for predicting outcomes and risk stratification in CVDs. Lipoprotein(a) [Lp(a)] is a liver-derived lipoprotein first identified by Kåre Berg in 1963 [2]. In contemporary discourse, Lp(a) has emerged as a compelling novel risk factor for cardiovascular conditions [3, 4, 5, 6, 7]. Substantial evidence has established that Lp(a) contributes to the pathogenesis of atherosclerosis, vascular calcification, inflammation, and thrombosis [7]. Numerous observational studies have posited a robust correlation between Lp(a) levels and the incidence and prognosis of CVDs [4, 8, 9, 10]. However, these results are predicated on observational data, which are susceptible to confounding variables and the possibility of reverse causation. Prior investigations have probed the causal impact of Lp(a) concentrations on the risk of coronary artery disease (CAD) and peripheral artery disease (PAD) [11, 12, 13]. Notwithstanding, the data for these studies were aggregated from identical cohorts, presenting a complete overlap in samples, which might inflate the perceived causal influence of Lp(a) on the aforementioned conditions. Consequently, the extent of the causal association of Lp(a) with a broad spectrum of CVDs has not been definitively established.

While a definitive pharmacological therapy to reduce Lp(a) levels is still lacking, multiple targeted therapies (e.g., antisense oligonucleotides, siRNAs) are under investigation in clinical trials [14]. Given the structure of oxidized phospholipid components and apolipoprotein(a), it is hypothesized that Lp(a) may facilitate platelet aggregation [15, 16]. Additionally, aspirin has been shown to reduce the production of Lp(a) by inhibiting the expression of apo(a) mRNA in the liver, a process that may not rely on cyclooxygenase-1 [17, 18]. Consequently, it is proposed that individuals with elevated Lp(a) levels may benefit from aspirin therapy [19]. Some studies have suggested that individuals with elevated Lp(a) levels may derive cardiovascular benefit from aspirin therapy, even in the absence of established cardiovascular disease. These findings, however, are based on observational data and genetic subgroups, and their generalizability remains to be clarified [20, 21, 22].

Mendelian randomization (MR) constitutes an innovative methodological approach that employs genetic markers to determine the existence of a causal relationship between a putative risk factor and diseases of interest. Owing to the random inheritance and lifelong stability of genetic variants, MR is less susceptible to confounding factors and reverse causality, thereby serving as a surrogate for randomized clinical trials [23, 24, 25].

The present study was designed to elucidate the role of Lp(a) levels in nine CVDs, including CAD, atrial fibrillation (AF), heart failure (HF), pulmonary embolism, deep vein thrombosis, hypertension, rheumatic and non-rheumatic valve diseases, and PAD. Furthermore, it sought to assess the influence of aspirin on Lp(a) concentrations.

In this MR study, the findings revealed a causative association between elevated Lp(a) levels and an increased risk of CAD, AF, HF, hypertension, and PAD. Notably, genetically predicted higher SA levels were associated with lower Lp(a) levels in our analysis. While this observation may support a potential role for aspirin in modulating Lp(a)-related cardiovascular risk, this hypothesis requires validation in prospective clinical studies.

Growing evidence from observational studies supported the role of Lp(a) in atherogenesis and thrombosis [42, 43, 44]. The Copenhagen City Heart Study from the general Danish population examined 9330 individuals for 10 years and found that elevated Lp(a) levels were related to an increased risk of myocardial infarction (MI) [42]. A meta-analysis of 36 prospective studies summarized a total of 126,634 individuals indicated that there was a continuous and robust association between Lp(a) concentration and the risk of CAD [43]. Another analysis from the UK Biobank database enrolled 460,506 individuals demonstrated a linear relationship between Lp(a) and the risk of CAD during a median of 11.2 years follow-up [44]. Two large genetic epidemiological studies causally revealed Lp(a) concentration in the CAD population [12, 13]. Consistent with these findings, our MR analysis supports the causal relationship between elevated Lp(a) levels and an increased risk of CAD. Epidemiologic studies indicated only an extremely high Lp(a) concentration was correlated with a slightly increased risk of venous thromboembolism [45]. However, we did not observe a causal association between Lp(a) levels and deep vein thrombosis of lower extremities or pulmonary embolism in this study, which is consistent with a previous MR analysis [46].

Studies of evaluating the impact of Lp(a) levels on AF are limited and the relationship between the two has not been evaluated effectively. A community-based cohort study with a median follow-up of 13.9 years indicated that elevated Lp(a) levels were not correlated with the risk of AF [47]. However, this observational cohort study did not include extremely high Lp(a) concentrations and excluded individuals aged over 65 years old; the results may be influenced by potential confounders and sample selection bias. In contrast, a recent observational study conducted in the UK biobank database suggested each 50 nmol/L elevation in Lp(a) was correlated with an increased risk of AF occurrence [Hazard ratio (HR): 1.03; 95% CI: 1.02–1.04; p 0.05] [48]. Consistent with our study, the MR analyses revealed that Lp(a) may be a causal mediator in the incidence of AF [48]. A large prospective cohort from Europe demonstrated that Lp(a) levels were significantly associated with the occurrence of peripheral artery disease [49]. Our results by MR analysis further complemented the causal association between Lp(a) levels and the risk of PAD.

The Multi-Ethnic Study of Atherosclerosis (MESA) study, which included 6809 participants, suggested that elevated Lp(a) levels were associated with a higher risk of HF [Lp(a) $\ge$50 mg/dL; HR, 1.87; p = 0.006] in the white population, which is consistence with our results [50]. Similarly, although limited data exists on the role of Lp(a) in hypertension, a cohort study found that 30% of hypertensive individuals had evaluated Lp(a) levels [51]. Despite these findings, observational studies alone cannot clarify the precise association between Lp(a) levels and the incidence of HF and hypertension. Our study is the first to demonstrate a causal relationship between elevated Lp(a) levels and the risk of both HF and hypertension.

The question of whether targeted Lp(a) reduction therapies can mitigate cardiovascular event risks has ignited considerable interest. However, the current challenge is the absence of specific therapies aimed at lowering Lp(a) levels. A randomized controlled trial (RCT) by Lacaze et al. [22] examined the effect of aspirin on subjects with genotypes related to high plasma Lp(a), revealing that aspirin diminished the incidence of major adverse cardiovascular events in individuals over 70 years old. However, this RCT could not eliminate the potential impact of other variables, such as statin therapy, which may elevate Lp(a) levels. Furthermore, direct measurements of Lp(a) levels were not taken, leaving the specific influence of aspirin on Lp(a) levels undetermined. Notably, Bhatia et al. [52] conducted an observational study using the MESA cohort and demonstrated that aspirin use was associated with a significantly lower risk of coronary heart disease events in individuals with Lp(a) $>$50 mg/dL, suggesting a potential benefit of aspirin for primary prevention in this population. However, their study was limited by the inherent drawbacks of observational studies, such as the potential bias, and the definition of aspirin use was based on patient self-reports. In contrast, our MR study leveraged genetic variants to minimize confounding and reverse causality, providing supportive causal inference regarding the inverse association between SA levels and Lp(a) concentrations, Increasing the possibility that aspirin may lead to a decrease in Lp(a). These studies underscore the urgent need for RCT to clarify the role of aspirin in managing cardiovascular risk associated with elevated Lp(a). To date, the mechanism by which aspirin may reduce Lp(a) levels remains unclear. Previous in vitro studies have shown that aspirin and sodium salicylate can inhibit the transcription and mRNA expression of apo(a) gene in human hepatocytes, thereby reducing Lp(a) synthesis, while the non-selective cyclooxygenase inhibitor indomethacin is ineffective, indicating that this process is independent of cyclooxygenase inhibition [17, 18]. These results are consistent with Mendelian randomization analyses that we found a negative correlation between SA levels and Lp(a) concentrations. Further studies are needed to clarify this mechanism and its clinical implications.

The principal merit of this investigation lies in the application of the MR methodology to establish a causal relationship between Lp(a) levels and the risks of nine CVDs within a relatively homogeneous population, distinctly without overlap in the study cohort. Our findings bolster the evidence for a causal link between Lp(a) and cardiovascular risk, aligning with the recent declaration by the European Atherosclerosis Society consensus [6].

Adhering to Mendel’s Second Law, which posits the independent assortment of alleles, each heritable trait segregates independently during gamete formation. Consequently, in a relatively homogenous population, the random distribution of genotypes amidst potential confounders allows for the inference of causality under conditions akin to those in a RCT [53]. Thus, our MR analysis is poised to circumvent biases that typically arise from confounding factors and reverse causality. Moreover, the stability of Lp(a) concentrations throughout an adult’s life, governed by genetic variation, renders it an exemplary subject for MR analysis [5]. The outcomes of this research might well reflect the lifelong implications of Lp(a) levels on cardiovascular disease risk. This MR study was specifically conducted within a European ancestry population using GWAS data for Lp(a) levels and CVDs, which serves to minimize the impact of bias and confounding variables. Additionally, MR-Egger regression was utilized to ascertain that the SNPs exert their effects on CVDs solely through Lp(a) levels and no directional pleiotropic effects were detected in this MR analysis.

However, it is worth noting that several outcomes in our MR analysis exhibit unusually narrow confidence intervals. This pattern is likely a consequence of the large sample sizes and the small effect sizes commonly observed in GWAS. While narrow confidence intervals suggest high statistical precision, they should be interpreted in the context of the underlying data structure and study design. Specifically, large sample sizes can yield precise estimates even when the magnitude of the effect is modest, which may not always reflect stronger causal relationships. Therefore, these findings should be interpreted with caution and regarded as hypothesis-generating. Further studies—particularly those involving independent cohorts and complementary methodological approaches—are warranted to validate and extend these results.

Several limitations should be acknowledged in this study. Primarily, due to reliance on summary data, individual patient-level information was inaccessible; consequently, it was not feasible to stratify the association between Lp(a) and CVDs by sex and age within this study. Second, the research encompassed solely the European demographic; hence, additional studies are requisite to determine whether these results are applicable to other ethnicities. Third, given the inherent limitations of MR studies, although we used MR-Egger, weighted median, and Cochran Q test to detect and reduce pleiotropy and heterogeneity, residual bias for undetected pleiotropy or gene-environment interactions could not be entirely ruled out. Fourth, multiple testing may increase the risk of type I error, thus some statistically significant findings should be interpreted with caution. Fifth, our results suggested that genetically predicted SA levels were inversely correlated with Lp(a), but SA is not unique to aspirin metabolism and lacks the pharmacological effects of aspirin. Sixth, all nine genetic variants used as instruments for Lp(a) are located within the lipoprotein(a) gene locus on chromosome 6. While LD-clumping with an R² threshold of 0.001 was employed to minimize statistical correlation among the variants, it is important to acknowledge that their shared biological context may limit the ability of MR-Egger and related tests to detect horizontal pleiotropy. This potential limitation must be considered when interpreting the results, as the biological effects of these variants within the same locus may introduce confounding factors that obscure the detection of pleiotropy. Therefore, our findings provide indirect evidence for the potential Lp(a)-lowering effects of aspirin, which still require further clinical validation in the future. Finally, the observed odds ratios for hypertension, peripheral artery disease, and non-rheumatic valve diseases were relatively modest, which may be influenced by the large sample size used in this study. Therefore, these findings should be interpreted with caution. Further research is needed to reaffirm the causal links between Lp(a) levels and these three cardiovascular conditions.

In summary, our MR analysis lends credence to a causal connection between Lp(a) levels and the risk of CVDs including CAD, AF, HF, hypertension, and PAD, suggesting individuals may benefit from reducing Lp(a) levels, which can be expected to be a new target for lowering cardiovascular risk. Furthermore, aspirin may be an effective therapeutic agent to reduce CVD risks in individuals with elevated Lp(a) levels in the future, though this hypothesis warrants prospective clinical validation.

All summary-level GWAS data used in this study are publicly available from the IEU OpenGWAS database (https://gwas.mrcieu.ac.uk/). No individual-level data were used. The analysis code is available upon reasonable request from the corresponding author.

YMY and JZ contributed to the conception and design of the study and critically reviewed the manuscript for important intellectual content. JST and SH organized the database. JST, SH and WX performed the statistical analysis and drafted the initial manuscript. JYW and LLW contributed to the interpretation of results and critically reviewed the revised manuscript. SH and WX assisted with manuscript revision. All authors contributed to the manuscript. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

Ethical approval was not provided for this study on human participants because the summary data were previously published and the ethics approvals have been obtained in their institutions. Therefore, no additional ethics approvals were required for this study. All patients or their families/legal guardians gave their written informed consent before they participated in the study.

This study was supported by National Clinical Medical Research Center for Cardiovascular Diseases (NCRC2020015), High-Level Hospital Clinical Research Funding (2025-GSP-QN-27 and 2022-GSP-GG-26), and Fundamental Research Funds for the Central Universities (2024-XHQN15).

The authors declare no conflict of interest.

Artificial intelligence tools (specifically, ChatGPT by OpenAI) were used to assist with language editing. The authors reviewed, edited, and approved all content, and accept full responsibility for the integrity and accuracy of the final version. No AI tools were used in the design, data collection, analysis, or interpretation of the study.