Introduction

Cardiovascular diseases (CVDs) have emerged as a predominant cause of global mortality and morbidity, presenting a significant public health challenge due to the escalating disease burden [1, 2]. Extensive genetic, biological, evidence-based, and epidemiological research consistently highlight cholesterol as a key pathogenic risk factor for atherosclerosis, underscoring the importance of lipid management in mitigating and regulating the risk of cardiovascular diseases (CVDs) [3,4,5]. LDL-C stands out as a well-established mediator of atherosclerosis, representing a primary target for intervention in both primary and secondary CVD prevention strategies [6]. Despite significant reductions in LDL-C levels, patients continue to experience recurrent CVD events. Hypertriglyceridemia emerges as a potential influential factor contributing to this lingering risk. Observational and genetic epidemiological studies robustly suggest that triglycerides, apolipoproteins, and remnant cholesterol in the progression of CVDs [7]. Nevertheless, the causal relationship between lipid profiles, apolipoprotein levels, cardiac structure, function remains inadequately explored. Unraveling these causal connections is essential for assessing and mitigating cardiovascular risks in individuals exhibiting elevated lipid and apolipoprotein traits.

Cardiovascular magnetic resonance imaging (CMR) has gained widespread recognition as the premier non-invasive method for assessing cardiovascular structure and function, providing valuable insights into the risk factors and pathological status of CVDs [8, 9]. The reliability and consistency of CMR make it a preferred modality for assessing primary CVD prevention, with its derived parameters acting as biomarkers and endophenotypes for a range of clinical outcomes [10]. Moreover, comprehensive genomic association studies have pinpointed thousands of SNPs linked to lipid profiles, apolipoprotein patterns, or CMR characteristics [11,12,13,14,15,16,17,18,19,20]. These studies have significantly enhanced our genetic comprehension of the structural and functional alterations in the heart and aortic system throughout the progression of disease. Notably, early detection and intervention of these variations are essential for preventing and treating CVDs.

As an emerging approach, MR has been widely applied to establish causal links between commonly associated traits and diseases [21, 22]. By assuming that genetic variations are inherited randomly, MR analysis can reduce the influence of confounding variables and minimize bias from reverse causality. In this context, we utilized a two-sample MR design to systematically explore the potential causal relationships between lipid profiles, apolipoprotein patterns, and CMR traits.

Methods

Study Design

A schematic summary of the study design is given in Fig. 1. The MR analysis was grounded on three fundamental assumptions [21]. Assumption 1: Genetic variants are strongly associated with the exposure variables. Assumption 2: Genetic variants are independent of any potential confounding factors. Assumption 3: Genetic variants affect the outcome variables solely through the exposure pathway. This study adheres to the reporting standards outlined in the STROBE-MR checklist guidelines (https://www.Strobe-mr.org/). The genome-wide association studies (GWAS) summary data employed in this research were openly accessible for download, and each original study obtained written informed consent from participants and ethical approval from relevant committees.

Fig. 1
figure 1

A schematic summary of the study design

Full size image

Data Sources

GWAS Summary Statistics of Lipids and Apolipoproteins

Summary statistics for serum lipid traits were sourced from the Global Lipids Genetics Consortium (GLGC) [11,12,13]. The GLGC is a global collaborative effort among researchers aimed at elucidating the genetic underpinnings of quantitative lipid traits. This consortium, which comprises 930,672 individuals of European ancestry excluding those from the UK Biobank, encompasses parameters such as total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and triglycerides (TG). Additionally, genetic associations related to remnant cholesterol (RC), apolipoprotein A-I (ApoA-I), and apolipoprotein B (ApoB) were derived from GWAS studies conducted by Richardson et al. For a more comprehensive understanding of the specific GWAS data mentioned above, further details can be obtained from the original literature.

GWAS Summary Statistics of CMR Traits

In our analysis, we scrutinized 32 cardiac and aortic MRI-derived traits categorized into 20 left ventricle traits, 4 right ventricle traits, 4 ascending aorta traits, and 4 descending aorta traits [14,15,16,17,18,19,20]. The summary-level data for GWAS on these CMR characteristics primarily originated from seven studies within the UK Biobank, encompassing participant numbers from 29,506 to 43,230. For further details on the specific CMR traits and the covariates included in the GWAS analysis model, refer Table 1.

Table 1 Overview of genome-wide association studies used
Full size table

Instrumental Variable Selection

We selected all SNPs that robustly and independently predicted serum lipid and apolipoprotein traits at the GWAS (P < 5 × 10–8). A linkage disequilibrium (LD) analysis was conducted utilizing the European genotype in the genome of 1000 individuals as a reference panel, eliminating potential linkage disequilibrium (LDR2 < 0.001, kb = 10,000), palindromic structural SNPs (minor allele frequency > 0.42), and incompatible SNPs [23]. To assess the strength of each IVs, we calculated the F-statistic (F = β2/se2) for each SNP [24]. For the genetic prediction of lipid and apolipoprotein traits (Table S1), the F-statistics ranged from 30.01 to 1332.26. Regarding the genetic predisposition of CMR traits (Table S8), F-statistics ranged from 25.35 to 317.47, with F > 10 indicating the absence of weak instrumental bias. Additionally, reverse MR analyses were conducted to evaluate the causal impacts of genetically predicted CMR traits on serum lipids and apolipoprotein traits. The data harmonization and MR methodologies were consistent with those used in the forward MR analysis. Detailed information on the exposure and outcome datasets, as well as the IVs, can be found in Tables S1 and S8.

Statistical Analysis

In the forward MR analysis, SNPs directly associated with the CMR traits outcomes were initially excluded to minimize the potential for horizontal pleiotropy, with a significance threshold of P < 5 × 10–8. The primary analysis employed the IVW method with a random effects model [25]. Additionally, a variety of sensitivity analyses including MR-Egger [26], weighted median [27], weighted mode [28], simple mode, and penalized weighted median methods were conducted to assess potential biases in the MR framework. Furthermore, RadialMR and MR Pleiotropy Residual Sum and Outlier (MRPRESSO) (NbDistribution = 3000) were utilized to identify multi-effect outliers at any level for exposures displaying significant causal associations, with subsequent evaluation of causal association estimates post-outlier removal [29, 30].

Moreover, sensitivity analyses were performed using MR-Egger regression to validate the robustness of the IVW results (a statistically significant P-value for intercept < 0.05) [31]. Cochran’s Q statistic was used to evaluate the heterogeneity of individual causal effects, and a leave-one-out analysis was conducted to scrutinize the influence of potentially outlying and pleiotropic SNPs on the causal estimates [32, 33]. To address potential pleiotropic effects in the study, Multivariable MR (MVMR) analyses were conducted to adjust for other lipid and apolipoprotein traits [34]. Additionally, adjustments were made for potential confounders such as body mass index (BMI), hypertension (HTN), and type 2 diabetes (T2D) (Table 1). MVMR enables the simultaneous evaluation of causal effects among multiple factors and outcomes, utilizing IVW and Egger methods for analysis.

All statistical analyses were conducted using the “TwoSampleMR,” MRPRESSO, and Radial-MR packages within the R software. The MR estimates were reported as effect sizes (β) along with 95% confidence intervals (CI). In the univariate MR analysis, the statistical significance threshold was set at a P-value < 2.23E-04 (0.05/7 serum lipids and apolipoproteins traits × 32 CMR traits) following Bonferroni correction to address multiple testing concerns. A P-value falling between 2.23E-04 and 0.05 was considered nominally significant for a potential causal association. In the MVMR analysis, a P-value < 0.05 was deemed statistically significant.

Colocalization and Susie Analysis

We conducted a colocalization analysis to assess the shared genetics among serum lipids, apolipoproteins, and CMR traits using the R package “coloc” [35, 36]. The analysis was based on default priors: p1 = 10–4, p2 = 10–4, p12 = 10–5, where p1, p2, and p12 represent the prior probabilities that a SNP in the examined region is significantly associated with the CMR traits outcome, or both, respectively. The colocalization analysis provides five different hypotheses for posterior probabilities: PPH0 indicates no association with any trait; PPH1 suggests an association with lipid levels and apolipoproteins but not with CMR traits; PPH2 indicates an association with CMR traits but not with lipid levels and apolipoproteins; PPH3 indicates an association with both CMR traits and lipid levels and apolipoproteins, with apparent causal variation; and PPH4 indicates associations with both CMR traits and lipid levels and apolipoproteins, with shared causal variation. During the analysis, we simultaneously used the coloc. abf and coloc. susie functions to calculate posterior probabilities. We observed that the combination of low values for PPH3 and PPH4, along with high values for PPH0, PPH1, and/or PPH2, indicated limited capacity for the colocalization analysis. Therefore, we restricted the analysis to traits with PPH3 + PPH4 ≥ 0.8 to ensure the reliability and validity of the results [36].

Results

Lipids and Apolipoproteins Traits and Cardiovascular Magnetic Resonance Traits

Specifically, we utilized the MRPRESSO and RadialMR methods to identify any multi-pleiotropic outliers across different levels and reassessed the causal estimates after the removal of these outliers, as detailed in Table S2. In the thorough analysis of all outcome data, a genetic liability for Lipids and apolipoproteins traits showed a correlation (Pvalue < 0.05) with 87 CMR traits. Of these, 32 CMR traits withstood multiple hypothesis testing (Fig. 2 and Table S2). We observed that, genetically predicted higher levels of LDL-C corresponded to increased descending aorta strain (DAo strain) (β = 0.098; 95% CI 0.060–0.135; P = 2.69E-07; Pirruccello et al. 2023), ascending aorta strain (AAo strain) (β = 0.079; 95% CI 0.041–0.117; P = 5.19E-05; Pirruccello et al. 2023), and AAo aortic distensibility (β = 0.078; 95% CI 0.037–0.118; P = 2.00E-04; Pirruccello et al. 2023).

Fig. 2
figure 2

Balloon plot of the association of lipids, apolipoproteins traits with 32 CMR features. Summary of non-significant, suggestive, and significant associations of genetically predicted lipids and apolipoproteins traits with CMR traits (inverse-variance weighted method estimates results)

Full size image

As for HDL-C, genetically predicted higher levels of HDL-C significantly positively correlated with left ventricular radial peak diastolic strain rate (LV-PDSRll) (β = 0.176; 95% CI 0.094–0.259; P = 2.89E-05;Thanaj et al. 2022) and the left ventricular longitudinal peak diastolic strain rate (LV-PDSRrr) (β = 0.059; 95% CI 0.035–0.084; P = 2.44E-06; Thanaj et al.2022). In addition, we also observed that genetically determined HDL-C levels were inversely correlated with left ventricular regional wall thickness (LVRWT). Specifically, left ventricular end diastole anterior wall thickness (LVED-A) (β = − 0.131; 95% CI − 0.173–− 0.089; P = 9.21E-10; Ning et al. 2023), left ventricular end diastole anteroseptal wall thickness (LVED-AS) (β = − 0.093; 95% CI − 0.133–− 0.052; P = 7.82E-06; Ning et al. 2023), Left ventricular end diastole inferior wall thickness (LVED-I) (β = − 0.108; 95% CI − 0.149–− 0.066; P = 3.45E-07; Ning et al. 2023), left ventricular end diastole inferolateral wall thickness (LVED-IL) (β = − 0.098; 95% CI − 0.140–− 0.057; P = 3.62E-06; Ning et al. 2023), Left ventricular end diastole inferoseptal wall thickness (LVED-IS) (β = − 0.107; 95% CI − 0.148–− 0.065; P = 5.53E-07; Ning et al. 2023), and Left ventricular end systole inferoseptal wall thickness (LVES-IS) (β = − 0.090; 95% CI − 0.132–− 0.048; P = 2.93E-05; Ning et al. 2023) were negatively correlated with HDL-C levels. Additionally, higher HDL-C levels were also correlated to decreased left ventricular mass (LVM) (β = − 1.845; 95% CI − 2.799–− 0.890; P = 1.51E-04; Khurshid et al. 2023) and left ventricular mass index (LVMI) (β = − 0.952; 95% CI − 1.387–− 0.517; P = 1.78E-05; Khurshid et al. 2023).

Conversely, we observed that TG levels were negatively correlated with LV-PDSRll and LV-PDSRrr, and positively associated with LVRWT. Specifically, genetically predicted higher levels of TC significantly negatively correlated with LV-PDSRll (β = − 0.249; 95% CI − 0.344–− 0.154; P = 2.71E-07) and LV-PDSRrr (β = − 0.083; 95% CI − 0.111–− 0.055; P = 6.38E-09). Genetically determined higher levels of TC corresponded to increased LVED-A (β = 0.136; 95% CI 0.085–0.187; P = 1.48E-07), LVED-AS (β = 0.107; 95% CI 0.059–0.155; P = 1.15E-05), LVED-I (β = 0.115; 95% CI 0.065–0.165; P = 6.57E-06), LVED-IS (β = 0.180; 95% CI 0.131–0.229; P = 7.22E-13), left ventricular end systole anterior wall thickness (LVES-A) (β = 0.134; 95% CI 0.086–0.182; P = 5.15E -08; Ning et al. 2023), LVES-AL (β = 0.140; 95% CI 0.089–0.190; P = 5.46E-08; Ning et al. 2023), LVES-AS (β = 0.136; 95% CI 0.088–0.185; P = 3.35E-08; Ning et al. 2023), LVES-AL (β = 0.117; 95% CI 0.066–0.169; P = 7.46E-06; Ning et al. 2023), and LVES-IS (β = 0.169; 95% CI 0.118–0.220; P = 6.67E-11; Ning et al. 2023). Additionally, higher TC levels were also correlated to decreased AAo aortic distensibility (β = − 0.081; 95% CI − 0.123–− 0.038; P = 2.17E-04; Pirruccello et al. 2023), left ventricular stroke volume (LVSV) (β = − 0.085; 95% CI − 0.129–− 0.042; P = 1.25E-04; Zhao et al. 2023), and right ventricular stroke volume (RVSV) (β = − 0.129; 95% CI − 0.186–− 0.073; P = 8.29E-06; Zhao et al. 2023).

As for ApoA-I and ApoB, genetically predicted higher levels of ApoA-I corresponded to decreased LVED-IS (β = − 0.087; 95% CI − 0.124–− 0.049; P = 4.78E-06; Ning et al. 2023) and LVM (β = − 1.664; 95% CI − 2.492–− 0.836; P = 8.24E-05; Pirruccello et al. 2023). Besides, genetically determined higher levels of ApoB corresponded to increased AAo strain (β = 0.076; 95% CI 0.042–0.110; P = 1.16E-05; Pirruccello et al. 2023), DAo strain (β = 0.065; 95% CI 0.035–0.096; P = 2.77E-05; Pirruccello et al. 2023), and left ventricular end systole anterolateral wall thickness (LVES-AL) (β = 0.067; 95% CI 0.032–0.103; P = 2.09E-04; Ning et al. 2023).

In the MVMR-IVW and MVMR-Egger framework, MVMR1 (adjusted for lipids and apolipoproteins) (Fig. 3 and Table S3), and MVMR2, we adjusted for three potential confounding factors (BMI, T2D, and HTN) (Fig. 3 and Table S4). We observed that the causal associations remained robust after adjusting for these factors. Notably, the most the P-values for the MVMR-Egger intercept were greater than 0.05, suggesting a low likelihood of pleiotropy, as shown in Table S3 and Table S4.

Fig. 3
figure 3

Significant associations of genetically predicted lipids, apolipoproteins levels with CMR traits. MVMR1 (lipids and apolipoproteins adjusted); MVMR2 (BMI, HTN, and T2D adjusted)

Full size image

Colocalization and Susie analysis for lipids and apolipoproteins and CMR traits.

We conducted a colocalization analysis to further strengthen the evidence of the MR results. By analyzing the results of Coloc.abf, we found that the region of KANK2 is shared by ApoB with AAo strains (PPH3 + PPH4 = 99.61%), and the region of ABCA7 is shared by ApoB with DAo strains (PPH3 + PPH4 = 99.64%) (as shown in Fig. 4A and 4B and Table S6). This result indicates that ApoB shares the same pathogenic genetic loci with the traits of these AAo strains and DAo strains. To further validate this finding, we also employed the Coloc.susie method for supplementary analysis. Unlike Coloc.abf, Coloc.susie assumes that there may be multiple causal variants at each locus, allowing for more accurate inferences. Interestingly, both the Coloc.abf and Coloc.susie analysis results indicate that PPH3 predominates at these shared pathogenic loci, suggesting that there are likely two independent causal SNPs at these loci (Table S7). Although these independent causal SNPs colocalize in the same genomic region, they may be regulated by different causal variants within these loci. However, it is worth noting that the presence of independent causal SNPs does not negate the significant colocalization phenomenon we observed, as the overall evidence still supports the existence of shared genetic structures between these traits.

Fig. 4
figure 4figure 4

The colocalization results of ApoB with AAo strain and DAo strain are visualized. The lead SNP is indicated by the purple diamond. The plots on the right show the causal variant of two traits from the same locus. The-log P values of two traits from a single locus are plotted on the left. A Colocalization analysis of ApoB and AAo strain; B Colocalization analysis of ApoB and DAo strain

Full size image

Cardiovascular Magnetic Resonance Traits and Lipids and Apolipoproteins Traits

Reverse MR analysis evaluated the causal effects of 32 CMR features on circulating lipid and apolipoprotein profiles. Notably, only two CMR traits withstood multiple hypothesis testing. Specifically, genetically predicted higher levels of right ventricular stroke volume (RVSV) were associated positively with HDL-C level (β = 0.122; 95% CI 0.072–0.172; P = 1.38E-06). Additionally, higher left ventricular end diastole inferior wall thickness (LVED-I) was positively correlated with apolipoprotein A-I levels (β = 0.109; 95% CI 0.05 5–0.163; P = 6.63E-05). Besides, our MR results did not show a significant causal relationship between other CMR traits and lipids and apolipoproteins (Table S9).

MR Sensitivity Analysis Results

The MR estimates revealed no evidence of horizontal pleiotropy (MR-Egger intercept, p-value > 0.05) or heterogeneity (Cochrane's Q, p-value > 0.05) (Table S5). Furthermore, we performed a leave-one-out sensitivity analysis. Notably, all outcomes exhibited consistency and stability, as illustrated in Figures S1–S31. These analyses are illustrated through scatter plots, funnel plots, and forest plots, which can be found in Figures S1–S31. The robustness and coherence of these sensitivity assessments bolster the credibility of our research findings and reinforce our conclusions.

Discussion

In this study, we leveraged the latest large-scale GWAS summary-level data to evaluate the causal relationships between genetically predicted lipids and apolipoproteins traits and 32 CMR traits. Interestingly, we observed that genetically predicted elevated levels of HDL-C and TC traits were linked to a diverse array of CMR traits. To our knowledge, this is the most comprehensive MR study delving into the causal relationship between lipids and apolipoproteins traits and the structure and function of the cardiac and aortic systems.

The previous MR analysis has demonstrated that LDL-C and TG levels are linked to adverse alterations in cardiac structure and function, particularly concerning left ventricular mass (LVM). In contrast, HDL-C does not seem to induce significant changes in LV structure and function. These results suggest that LDL-C and TG may have a causal impact on cardiac morphology beyond their established role in atherosclerosis [6]. The extensive evidence base highlights a robust and progressive link between LDL-C levels and cardiovascular mortality, emphasizing the pivotal role of cholesterol assessment and lipid-lowering therapies in the primary and secondary prevention of CVDs [37, 38]. These morphological changes often coincide with exposure to other significant risk factors like HTN or elevated BMI, often presenting subclinically before any overt clinical manifestations emerge. However, our univariate MR analysis focusing on LDL-C versus LV phenotype did not align with the previous MR findings (Fig. 2 and Table S2). Specifically, genetic predisposition to LDL-C levels did not show an association with LVM. Compared to previous MR studies, our analysis utilized summary statistics from larger cohorts and included a more extensive set of LDL-C-associated genetic variants, thereby increasing the statistical power and robustness of our findings. Notably, our MR results indicate a significant correlation between genetically predicted higher LDL-C levels and CMR traits, including increased DAo strain, AAo strain, and AAo aortic distensibility. Elevated LDL-C levels may contribute to atherosclerosis development, leading to alterations in aortic morphology and structure [39]. Consequently, reducing LDL-C levels can mitigate the progression of aortic strain, thereby subsequently reducing the risk of cardiovascular events. This highlights the significance of managing LDL-C levels as a pivotal strategy in the prevention of CVDs.

Epidemiological studies show that HDL-C levels are inversely correlated with the risk of cardiovascular events [40]. However, recent research has introduced novel insights that challenge the conventional wisdom. For instance, studies have shown that augmenting HDL-C levels through the inhibition of cholesteryl ester transfer protein (CETP) activity does not significantly reduce the risk of developing CVDs [41]. In some cases, excessively high levels of HDL-C may even be associated with an increased risk of cardiovascular disease [42]. Our MR study discovered that genetically predicted elevated levels of HDL-C exhibited a significant and positive correlation with LV-PDSRll and LV-PDSRrr. This suggests that an increase in HDL-C levels may potentially enhance both systolic and diastolic cardiac function. Notably, our findings revealed an inverse relationship between genetically determined HDL-C levels and LVRWT. Specifically, parameters such as LVED-A, LVED-AS, LVED-I, LVED-IL, LVED-IS, and LVES-IS exhibited negative correlations with HDL-C levels, suggesting that higher HDL-C levels might be linked to a reduction in myocardial hypertrophy. Moreover, elevated levels of HDL-C are associated with reduced LVM and LVMI, further emphasizing the potential cardioprotective role of HDL-C, as illustrated in Fig. 2 and detailed in Table S2. The anti-inflammatory and antioxidant properties of HDL-C, as well as its facilitation of reverse cholesterol transport, are proposed as the underlying mechanisms of its protective role against CVDs [40, 43]. Therefore, future research should delve deeper into understanding how HDL-C levels contribute to cardiovascular health and elucidate the mechanisms through which HDL-C operates to develop more effective strategies for prevention and treatment.

Conversely, we observed that the effect of TG levels compared to HDL-C showed an inverse trend to heart function. Specifically, we observed that TG levels were negatively associated with LV-PDSRll and LV-PDSRrr, and positively associated with LVRWT, may indicate that high TC levels may be associated with weakened cardiac function as well as myocardial hypertrophy (Fig. 2 and Table S2). High TG levels can lead to the development of atherosclerosis, which in turn affects the blood supply to the heart, increases the burden on the heart, and leads to complications such as myocardial ischemia and myocardial infarction [44, 45]. In addition, TG levels may also cause changes in the structure of the heart, such as hypertrophy and dilation of the heart muscle, which can affect heart function [6]. Notably, after accounting for potential confounders including BMI, T2D, HTN, and lipid profiles in MVMR analyses, the enduring causal relationship between HDL-C and TG levels and left ventricular remodeling continues to highlight their pivotal role in influencing LVRWT (Fig. 3 and Table S3-S4). This underscores the critical importance of managing HDL-C and TG levels as a key strategy in preventing CVDs. In addition, Xu et al. demonstrated elevated RC levels in young adulthood were related to adverse LV remodeling and dysfunction in midlife [46]. However, our MR study did not show a significant causal association between high levels of RC and cardiac structure and function (Fig. 2 and Table S2). These findings highlight the intricate interplay between cholesterol metabolism and cardiac structure. They not only deepen our understanding of the mechanistic basis of cardiac remodeling, but also highlight the importance of targeted interventions targeting lipid levels to mitigate CVDs risk.

Regarding ApoA-I and ApoB, higher genetically predicted levels of ApoA-I were associated with reduced LVED-IS and LVM. Conversely, elevated genetically determined levels of ApoB were linked to increased AAo strain, DAo strain, and left LVES-AL. Previous research has indicated a correlation between ApoA-I levels and the severity of aortic stenosis, suggesting ApoA-I potential as a biomarker for aortic valve disease [47, 48]. Alegret et al. demonstrated a positive correlation between ApoB levels and the extent of aortic stenosis and ascending aorta dilatation, implying a significant role for ApoB in aortic disease development [49]. Essentially, lipid and apolipoprotein levels are crucial not only for regulating cardiac structure and function but also for their significant association with aortic structure and function. These insights significantly advance our comprehension of the pathogenesis of cardiovascular diseases and suggest new directions for the development of preventive and therapeutic strategies.

The strength of this study is that the MR design minimizes confounding and reverse causality in observational studies and comprehensively assesses the causal relationship between the circulation of lipids and apolipoproteins traits and 32 CMR traits. Nevertheless, it is crucial to recognize certain limitations inherent in our MR analysis when interpreting the findings. Firstly, while MR provides a preliminary analysis of the causal relationship between exposure and outcome, it may not offer detailed insights into the specific biological mechanisms underlying this causality [50]. Secondly, our MR study could not fully eliminate the potential influence of unknown confounding factors, some of which may be undocumented in the current literature. Furthermore, the lack of individual-level data in MR studies restricts the ability to assess potential non-linear relationships. These limitations should be considered when interpreting the results of our analysis. Lastly, the results of our MR study were based on GWAS summary statistics from the European population, raising questions about their generalizability to other ethnic groups, and warranting further investigation.

In conclusion, this study extends the implications beyond conventional therapeutic approaches, necessitating a re-evaluation of HDL-C and TG as targets for preventive interventions. By elucidating these causal relationships, the research opens avenues for future studies focused on devising innovative therapeutic strategies or preventive measures for HDL-C and TG, which may substantially reduce the impact of CVDs on public health.