Causal association between remnant cholesterol level and risk of cardiovascular diseases: a bidirectional two sample mendelian randomization study

Abstract

Serum lipids have been associated with an increased risk of various cardiovascular diseases (CVDs) in several observational studies, but the causal inference between the remnant cholesterol (RC) levels and several CVDs risk has not been established. The purpose of this study was to investigate whether there is a causal relationship between RC levels and risk of developing CVDs by a bidirectional two-sample Mendelian randomization (TSMR) analysis. One TSMR analysis was performed using the publicly released large-scale genome-wide association study (GWAS) data. Inverse variance weighted (IVW) method was chosen as the main analysis method, and MR-Egger, weighted median, simple mode, and weighted mode were used as supplementary methods. We conducted a series of sensitivity analyses to assess the robustness of the main results, including the Cochran’s Q test, MR-Egger intercept test, leave-one-out sensitivity analysis, and funnel plot. The main IVW method revealed that genetically predicted serum level of RC is significantly associated with an increased risk of developing ischemic heart disease (OR = 1.409, 95%CI = 1.284–1.546, P value = 4.753E-13), unstable angina pectoris (OR = 1.621, 95%CI = 1.398–1.880, P value = 1.672E-10), myocardial infarction (OR = 1.526, 95%CI = 1.337–1.741, P value = 3.771E-10), cardiac arrest (OR = 1.595, 95%CI = 1.322–1.924, P value = 1.076E-06), heart failure (OR = 1.086, 95%CI = 1.009–1.169, P value = 0.028), hypertension (OR = 1.089, 95%CI = 1.043–1.136, P value = 9.458E-05), major coronary heart disease (CHD) events (OR = 1.515, 95%CI = 1.376–1.669, P value = 3.217E-17), coronary atherosclerosis (OR = 1.388, 95%CI = 1.231–1.564, P value = 7.739E-08), cardiac arrhythmias (OR = 1.067, 95%CI = 1.008–1.130, P value = 0.025), and atrial fibrillation and flutter (OR = 1.122, 95%CI = 1.039–1.211, P value = 0.003). Additionally, the causal associations between the RC levels and these CVDs remained significant after correcting for the false discovery rate (all P value < 0.05). However, this study did not find any significant association of RC with cardiomyopathy and pericarditis (both P value > 0.05). Heterogeneity existed in the IVs of RC and ischemic heart disease, unstable angina pectoris, myocardial infarction, heart failure, hypertension, major CHD events, cardiomyopathy, coronary atherosclerosis, cardiac arrhythmias and atrial fibrillation and flutter using the Cochran’s Q test (all P value < 0.05). Moreover, there was no horizontal pleiotropy in this study (all P value > 0.05). The leave-one-out sensitivity analyses showed that the causal effects between RC level and CVDs (except for heart failure, cardiomyopathy, pericarditis and cardiac arrhythmias) are not driven by a single SNP. The funnel plots showed that there is no obvious potential bias in our study. In the replication analysis, the genetically predicted RC levels were positively associated with a 43.12% higher risk of coronary artery disease. This present study supported the causal link between RC and heightened the risk of CVDs, indicating that RC-lowering treatment might be effective in preventing CVDs.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-78610-0.

Introduction

Over the past few years, atherosclerotic cardiovascular disease (ASCVD) has become a chronic epidemic worldwide, and is one of leading causes of death globally^1,2. Dyslipidaemia is considered to be the most important factor in the development and progression of ASCVD³. Millions of lives are lost annually due to sudden cardiac death (SCD), despite significant advances in cardiovascular medicine⁴. SCD is the most common and serious outcome of cardiac arrest, with an annual incidence of 0.03–0.10% in the general population⁵.

In 2023, a report on the global burden of cardiovascular diseases (CVDs) revealed that the number of cardiovascular deaths globally has increased significantly over time, from 12.4 million in 1990 to 19.8 million in 2022, possibly due to global population growth and aging, as well as preventable metabolic, behavioral, and environmental risk factors⁶. The abnormal metabolism of lipids, especially elevated levels of low-density lipoprotein cholesterol (LDL-C), is widely recognized as an important causative factor of ASCVD⁷. LDL-C is an important lipid indicator, and lowering LDL-C by statin therapy can significantly reduce the risk of ASCVD events and even reduce mortality⁸. However, clinical studies have shown that despite standardized treatment of LDL-C levels, a residual risk of cardiovascular events remains occur^9,10. In this context, it is essential to identify and address the residual risks of ASCVD. It is therefore important to consider the potential pathogenic role of other components of lipids in ACSVD. In recent years, remnant cholesterol (RC), as a triglyceride-rich lipoproteins (TRLs), has been found to be associated with an increased risk of CVDs^11,12, such as ischemic heart disease¹³, myocardial infarction¹⁴, and ischemic stroke¹⁵, etc. Furthermore, one recent study from China showed that the association of RC with diabetes mellitus may be mediated through insulin resistance and pro-inflammatory state¹⁶.

Currently, some scholars believed that the atherogenic capacity of RC is similar to that of LDL-C⁸. Epidemiologically, coronary heart disease is an important risk factor for cardiac arrest, and the common risk predictors of coronary heart disease and cardiac arrest may be dyslipidemia, diabetes mellitus, hypertension, smoking and obesity¹⁷. In recent years, Mendelian randomization (MR) studies have explored the causal relationship between RC and ASCVD^18,19. However, there are currently no studies reporting the role of RC in the risk of developing cardiac arrest, unstable angina pectoris, heart failure, hypertension, coronary atherosclerosis, cardiac arrhythmias, cardiomyopathy, and pericarditis, etc.

We, therefore, aimed to apply a bidirectional two-sample MR (TSMR) analysis to explore the role of genetically predicted serum RC levels in the risk of developing CVDs.

Methods

Study design

A bidirectional TSMR analysis was performed to evaluate the causal relationship between serum RC indicator and CVDs in both sexes (Fig. 1). All data were from genome-wide association study (GWAS). Our MR analysis was based on three key assumptions: (1) there is a significant association between genetic variation and exposure; (2) there is no correlation between the instrumental variable (IV) and any confounders; and (3) exposure is the only way in which genetic variation affects outcome. This study was reported based on the STROBE-MR guidance²⁰.

Fig. 1.

Overall design of the present study. SNP, single nucleotide polymorphism.

GWAS data sources

The IEU OpenGWAS project website (https://gwas.mrcieu.ac.uk/) provided the GWAS data of RC (met-d-Remnant_C). The IVs for RC were derived from this GWAS data which involved 115,078 individuals and identified a total of 12,321,875 single nucleotide polymorphisms (SNPs). The summary-level genetic data (R10 release) on outcomes were obtained from the FinnGen study conducted in December 18, 2023 (https://www.finngen.fi/en). Based on the International Classification of Diseases, patients from the FinnGen consortium were included in this study, including ischemic heart disease, unstable angina pectoris, myocardial infarction, cardiac arrest, heart failure, hypertension, major coronary heart disease (CHD) events, cardiomyopathy, pericarditis, coronary atherosclerosis, cardiac arrhythmias, and atrial fibrillation and flutter. The major CHD events in the FinnGen consortium (https://risteys.finregistry.fi/endpoints/I9_CHD) were defined as unstable angina, acute myocardial infarction, subsequent myocardial infarction and chronic ischemic heart disease, etc. The FinnGen study is a large-scale genomics initiative that has analyzed over 500,000 Finnish biobank samples and correlated genetic variation with health data to understand disease mechanisms and predispositions²¹. All data were obtained from European populations in both sexes.

In order to verify the reliability of our findings, we simultaneously applied MR analysis to explore the role of genetically inferred serum RC levels in the risk of developing coronary artery disease using the Coronary ARtery DIsease Genome wide Replication And Meta-analysis (CARDIoGRAM) GWAS data. Summary statistics of the coronary artery disease were obtained from a large-scale GWAS study by the CARDIoGRAM consortium, involving 86,995 European ethnic participants (22,233 cases and 64,762 controls)²². The MR study was performed using the publicly available GWAS data, and all original studies received ethical approval. Therefore, no additional ethical approval was required. Details of the phenotypes are described in Table 1.

Table 1.

Summary information of the data sets.

Traits	Consortium	Year	Sample size	Case/Control	SNPs	Population	Web source
Expose
Remnant cholesterol	IEU OpenGWAS	2020	115,078	NA	12,321,875	European	https://gwas.mrcieu.ac.uk/datasets/met-d-Remnant_C/
Outcomes
Ischemic heart disease	Finngen	2023	412,181	69,008/343,173	21,306,349	European	https://r10.finngen.fi/pheno/I9_IHD
Unstable angina pectoris	Finngen	2023	379,082	14,281/364,801	21,305,730	European	https://r10.finngen.fi/pheno/I9_UAP
Myocardial infarction	Finngen	2023	369,139	26,060/343,079	21,305,485	European	https://r10.finngen.fi/pheno/I9_MI_STRICT
Cardiac arrest	Finngen	2023	213,123	2,471/210,652	21,296,925	European	https://r10.finngen.fi/pheno/I9_CARDARR
Heart failure	Finngen	2023	387,444	29,672/357,772	21,305,881	European	https://r10.finngen.fi/pheno/I9_HEARTFAIL_EXMORE
Hypertension	Finngen	2023	412,113	122,996/289,117	21,306,348	European	https://r10.finngen.fi/pheno/I9_HYPTENS
Major CHD events	Finngen	2023	412,181	46,959/365,222	21,306,349	European	https://r10.finngen.fi/pheno/I9_CHD
Cardiomyopathy	Finngen	2023	318,492	6,338/312,154	21,303,898	European	https://r10.finngen.fi/pheno/I9_CARDMYO
Pericarditis	Finngen	2023	313,231	1,077/312,154	21,303,730	European	https://r10.finngen.fi/pheno/I9_PERICARD
Coronary atherosclerosis	Finngen	2023	394,668	51,589/343,079	21,306,040	European	https://r10.finngen.fi/pheno/I9_CORATHER
Cardiac arrhythmias	Finngen	2023	313,778	74,000/239,778	21,303,819	European	https://r10.finngen.fi/pheno/CARDIAC_ARRHYTM
Atrial fibrillation and flutter	Finngen	2023	261,395	50,743/210,652	21,301,317	European	https://r10.finngen.fi/pheno/I9_AF
Coronary artery disease	CARDIoGRAMplusC4D	2011	86,995	22,233/64,762	2,420,360	European	https://www.cardiogramplusc4d.org/data-downloads/

SNP, single nucleotide polymorphism, CHD coronary heart disease.

Selection criteria for the instrumental variables

First, candidate IVs were selected from SNPs that reached genome-wide significance levels (P value < 5 × 10^− 8, linkage disequilibrium R² threshold < 0.001 and window size = 10,000 kb). In the reverse MR analysis, when the screening criterion was set to P value < 5 × 10^− 8 (cardiac arrest, pericarditis), only 1 and 0 SNP was identified, respectively. Therefore, a more relaxed P value was used (P value < 5 × 10^− 6). Second, we chose SNPs with F-statistics greater than 10 as potential IVs in order to mitigate the effect of potential bias. The F-statistic is calculated as follows: F=[(N-K-1)/K]*[R²/(1-R²)]²³. N is the sample size of the exposure database, K is the number of SNPs, and R² is the proportion of variants explained by SNPs in the exposure database. R² is calculated as follows: R² = 2*MAF*(1-MAF)*β². MAF is the minor allele frequency, and β is the allele effect value. Furthermore, the previously identified RC-associated SNPs were then extracted from the CVDs pooled data (ischemic heart disease, unstable angina pectoris, myocardial infarction, cardiac arrest, heart failure, hypertension, major CHD events, cardiomyopathy, pericarditis, coronary atherosclerosis, cardiac arrhythmias, atrial fibrillation and flutter, and coronary artery disease). SNPs that were missing were substituted with highly correlated SNPs (R² > 0.8), and those lacking appropriate replacements were removed. Then, we pooled data from the RC and CVDs datasets and excluded SNPs directly associated with CVDs (P value < 5 × 10^− 8). We conducted MR-PRESSO analysis before MR analysis to remove outliers with potential pleiotropy, ensuring the reliability of our MR estimates. As a result of these steps, the remaining SNPs were ultimately used as genetic tools.

MR analyses

We performed five MR analyses, including the inverse-variance-weighted (IVW), weighted median, MR-Egger regression, simple mode and weighted mode, to evaluate the potential causal connection between serum RC levels and higher risk for CVDs. Among these five methods, the IVW method was used as the main analysis method to evaluate the effect estimates. A scatter plot was utilized to determine the causal relationship between RC indicator and CVDs, and Cochran’s Q test was used to test the heterogeneity of IVs. The intercept of MR-Egger regression was performed to test whether causality estimates were affected by the multiplicity of genetic variants. A leave-one-out sensitivity analysis were used to assess the robustness of the results. Funnel plots were used to visualize the symmetry of the distribution of the test effect estimates. All MR analyses were performed using the “TwoSampleMR” package (version 0.5.7) and R software (version 4.2.3). A two-sided P value of < 0.05 was taken as statistically significant.

To correct for false-positive results that may be brought about by the various comparison methods involved in this study, we calculated the false-discovery rate (FDR) adjusted P values across the main analyses using the R software (version 4.2.3). In addition, the statistical power of causal effect estimates was calculated using an online calculation tool in our MR study (https://sb452.shinyapps.io/power/). In general, a power greater than 80% was deemed appropriate^24,25.

Results

Instrumental variables

A total of 56 SNPs strongly associated with RC were initially screened from the GWAS data. Eventually, all included SNPs with F-statistic > 10 were obtained after screening by a pre-established process. Details of these SNPs were provided in Supplementary file 1 (Table 1- Table 12).

Causal effect of RC on CVDs

As shown in Figs. 2 and 3, we focused on the results of the IVW method. The IVW method showed that genetically predicted serum level of RC is significantly associated with an increased risk of developing ischemic heart disease (OR = 1.409, 95%CI = 1.284–1.546, P value = 4.753E-13, FDR-corrected P value = 2.850E-12), unstable angina pectoris (OR = 1.621, 95%CI = 1.398–1.880, P value = 1.672E-10, FDR-corrected P value = 6.680E-10), myocardial infarction (OR = 1.526, 95%CI = 1.337–1.741, P value = 3.771E-10, FDR-corrected P value = 1.130E-09), cardiac arrest (OR = 1.595, 95%CI = 1.322–1.924, P value = 1.076E-06, FDR-corrected P value = 2.160E-06), heart failure (OR = 1.086, 95%CI = 1.009–1.169, P value = 0.028, FDR-corrected P value = 0.033), hypertension (OR = 1.089, 95%CI = 1.043–1.136, P value = 9.458E-05, FDR-corrected P value = 1.622E-04), major CHD events (OR = 1.515, 95%CI = 1.376–1.669, P value = 3.217E-17, FDR-corrected P value = 3.860E-16), coronary atherosclerosis (OR = 1.388, 95%CI = 1.231–1.564, P value = 7.739E-08, FDR-corrected P value = 1.860E-07), cardiac arrhythmias (OR = 1.067, 95%CI = 1.008–1.130, P value = 0.025, FDR-corrected P value = 0.033) and atrial fibrillation and flutter (OR = 1.122, 95%CI = 1.039–1.211, P value = 0.003, FDR-corrected P value = 0.005). However, this study did not find any significant association of RC with cardiomyopathy (OR = 0.927, 95%CI = 0.826–1.041, P value = 0.201, FDR-corrected P value = 0.201) and pericarditis (OR = 1.252, 95%CI = 0.992–1.581, P value = 0.058, FDR-corrected P value = 0.064).

Fig. 2.

Summary results of the MR analysis for the association between remnant cholesterol and ischemic heart disease (a), unstable angina pectoris (b), myocardial infarction (c), cardiac arrest (d), heart failure (e), and hypertension(f) using the five methods. OR: odds ratio, CI: confidence interval. SNP, single nucleotide polymorphism.

Fig. 3.

Summary results of the MR analysis for the association between remnant cholesterol and major coronary heart disease events (a), cardiomyopathy (b), pericarditis (c), coronary atherosclerosis (d), cardiac arrhythmias (e), and atrial fibrillation and flutter (f) using the five methods. OR: odds ratio, CI: confidence interval. SNP, single nucleotide polymorphism.

As shown in Table 2, there was no evidence of horizontal multiple validity in the pleiotropy test (all P value > 0.05). The summary results of forest plots of causal effects between serum RC and CVDs were set forth in Supplementary file 2 (Fig. 1–Fig. 12). As shown in Table 2, the statistical power of causal inference was greater than 80% except for two outcomes (cardiomyopathy: 34.70%; pericarditis: 50.30%).

Table 2.

Tests of pleiotropy of selected SNPs and heterogeneity between SNPs.

Outcomes	Number of SNPs	Power(%)	Heterogeneity test		Pleiotropy test
			Cochran’s Q	P value	Intercept	P value
Ischemic heart disease	40	100	106.681	3.320E-08	-0.005	0.350
Unstable angina pectoris	44	100	97.319	4.338E-06	-0.004	0.639
Myocardial infarction	41	100	102.888	1.912E-07	-0.007	0.338
Cardiac arrest	52	100	66.526	0.071	-0.007	0.455
Heart failure	49	83.50	74.710	0.008	0.002	0.652
Hypertension	47	100	84.534	4.622E-04	-0.0004	0.833
Major CHD events	40	100	93.124	2.540E-06	-0.007	0.162
Cardiomyopathy	51	34.70	68.012	0.046	0.003	0.566
Pericarditis	52	50.30	46.524	0.652	-0.0004	0.970
Coronary atherosclerosis	39	100	118.703	3.105E-10	-0.006	0.293
Cardiac arrhythmias	48	98.10	105.474	2.223E-06	-0.0006	0.831
Atrial fibrillation and flutter	48	100	100.502	9.321E-06	0.002	0.669

SNP, single nucleotide polymorphism, CHD, coronary heart disease.

The effect of IVs on exposure and outcome was further visualized by the scatter plots. As shown in Figs. 4, 5 and 6, genetically predicted elevated RC was associated with an increased risk of ischemic heart disease, unstable angina pectoris, myocardial infarction, cardiac arrest, heart failure, hypertension, major CHD events, coronary atherosclerosis, cardiac arrhythmias, and atrial fibrillation and flutter, and the direction of causal effects obtained by the five methods was almost consistent. Nonetheless, it had been determined that genetically influenced RC does not cause cardiomyopathy or pericarditis.

Fig. 4.

Scatter plots of the causal effects. SNP effects on remnant cholesterol and ischemic heart disease (a), unstable angina pectoris (b), myocardial infarction (c), and cardiac arrest (d). SNP, single nucleotide polymorphism.

Fig. 6.

Scatter plots of the causal effects. SNP effects on remnant cholesterol and pericarditis (a), coronary atherosclerosis (b), cardiac arrhythmias (c), and atrial fibrillation and flutter (d). SNP, single nucleotide polymorphism.

Fig. 5.

Scatter plots of the causal effects. SNP effects on remnant cholesterol and heart failure (a), hypertension (b), major coronary heart disease events (c) and cardiomyopathy (d). SNP, single nucleotide polymorphism.

Causal effect of CVDs on RC

As listed in Supplementary file 3, in reverse MR analysis, the IVW method revealed that there exists potential reverse causality between several CVDs, including ischemic heart disease, myocardial infarction, cardiac arrest, major CHD events, pericarditis, coronary atherosclerosis, and cardiac arrhythmias, and RC (all P value < 0.05). Besides, the reverse causality between RC and unstable angina pectoris, heart failure, hypertension, cardiomyopathy, and atrial fibrillation and flutter was not observed. The results of heterogeneity and pleiotropy were shown in Supplementary file 3.

Sensitivity analyses

Subsequently, we conducted a series of sensitivity analyses to assess the robustness of the results, including the Cochran’s Q test, MR-Egger intercept test (Table 2), leave-one-out sensitivity analysis (Supplementary file 4: Fig.1–Fig.12), and funnel plot (Supplementary file 5: Fig.1–Fig.12). Heterogeneity was noted between the IVs and the outcomes of ischemic heart disease, unstable angina pectoris, myocardial infarction, heart failure, hypertension, major CHD events, cardiomyopathy, coronary atherosclerosis, cardiac arrhythmias, and atrial fibrillation and flutter using the Cochran’s Q test (all P value < 0.05), except for the outcomes of cardiac arrest and pericarditis.

According to the leave-one-out sensitivity analyses, the overall effect of RC susceptibility and ischemic heart disease, unstable angina pectoris, myocardial infarction, cardiac arrest, hypertension, major CHD events, coronary atherosclerosis, and atrial fibrillation and flutter was not driven by any individual SNP. Nonetheless, the MR estimates between RC and heart failure, cardiomyopathy, pericarditis and cardiac arrhythmias were driven by several SNPs (Supplementary file 4: Fig.1–Fig.12).

As presented in Supplementary file 5: Fig.1–Fig.12, visually apparent symmetry was displayed by the most of the funnel plots except for Figures S3 (myocardial infarction) and S7 (major CHD events). The funnel plots of thesetwo CVDs were slightly asymmetrical (Figures S3 and S7), but MR Egger showed little evidence of directional pleiotropy with a small intercept (-0.007, P value = 0.338; -0.007, P value = 0.162; Table 2). The result of funnel plots indicated the stability of our study.

Replication MR analysis

As shown in Supplementary file 6 (Fig. 1-Fig. 3), in the primary IVW MR analysis, genetically predicted elevated RC levels was statistically significantly associated with the risk of developing coronary artery disease (OR = 1.431, 95%CI = 1.159–1.768, P value = 8.695E-4). The statistical power for the outcome of coronary artery disease was equal to 100%.

Furthermore, the leave-one-out sensitivity analysis showed that the observed association did not change significantly even when each SNP was removed, suggesting the stability of the results (Supplementary file 6: Fig. 4). Funnel plot was symmetrical by the IVW method, supporting the robustness of the findings (Supplementary file 6: Fig. 5). The Cochran’s Q test in the IVW method (Q = 23.026, P value = 0.113) suggested that there is an absence of heterogeneity in the IVs. Moreover, there was no horizontal pleiotropy (P value = 0.285).

Discussion

We performed a bidirectional TSMR analysis to estimate the association between blood RC and risk of developing CVDs, and this study supported the causal association between RC and the risk of ischemic heart disease, unstable angina pectoris, myocardial infarction, cardiac arrest, heart failure, hypertension, major CHD events, coronary atherosclerosis, cardiac arrhythmias, and atrial fibrillation and flutter. Nonetheless, genetic liability to RC was not associated with cardiomyopathy and pericarditis risk. In the present study, the evidence from IVW analysis showed that for every 1 standard deviation increase in RC, the risk of ischemic heart disease, unstable angina pectoris, myocardial infarction, cardiac arrest, heart failure, hypertension, major CHD events, coronary atherosclerosis, cardiac arrhythmias, and atrial fibrillation and flutter development increased by 40.9%, 62.1%, 52.6%, 59.5%, 8.60%, 8.9%, 51.5%, 38.8%, 6.7% and 12.2%, respectively. Moreover, sensitivity analyses confirmed the stability of the findings. In the replication analysis, the genetically predicted RC levels were positively associated with a 43.12% higher risk of coronary artery disease. Furthermore, there was a potential reverse causality between several CVDs, including ischemic heart disease, myocardial infarction, cardiac arrest, major CHD events, pericarditis, coronary atherosclerosis, and cardiac arrhythmias, and RC.

CVDs are one of the most common causes of health hazards in the population, and cardiovascular events and complications can be reduced by intervening on risk factors^26–28. It is well known that LDL⁃C is recognized as the primary lipid-lowering target for reducing the risk of ASCVD²⁹. However, clinical practice and research have found that even after aggressive lipid-lowering therapy to achieve the desired targeted LDL-C levels, and good control of blood pressure and blood glucose, several patients still have a residual risk of cardiovascular events^30–32. A review³³ suggested that triglycerides and TGRLs, lipoprotein(a), and inflammation are the residual risk factors, and residual inflammatory risk may play an important role in the development of CVDs in addition to residual cholesterol risk. Furthermore, this review³³ indicated that there is no obvious evidence currently suggesting the possibility of LDL-C interference with exposure (RC).

Studies have shown that triglyceride lipoprotein-rich RC, which is considered to be one of the major residual lipid risks in patients with LDL-C compliance, is atherogenic and associated with an increased risk of CVDs^34–36. There is no uniform definition of RC, which usually refers to cholesterol in TRL. The RC is composed of very low density lipoprotein (VLDL) and intermediate density lipoprotein (IDL) in the fasting state, and RC contains celiac remnants, VLDL and IDL during the non-fasting period³⁷. RC can be measured directly or calculated using the following formula³⁸: RC = total plasma cholesterol-LDL cholesterol-HDL cholesterol. There is no consensus on the reference range for RC, and the European Atherosclerosis Society/European Federation of Clinical Chemistry and Laboratory Medicine recommends³⁹ that a fasting RC of ≥ 0.8 mmol/L (30 mg/dL) or non-fasting RC of ≥ 0.9 mmol/L (35 mg/dL) are considered abnormal.

Several previous studies have revealed the pathophysiological mechanisms of RC in the development of ASCVD, which may be involved in the initiation and progression of atherosclerosis and related events through several pathways^8,9,40. First, a high level of RC enhances penetration into the arterial walls, where RC than LDL is more easily captured and absorbed by macrophages⁴⁰. Moreover, TRLs have a larger size than LDL particles and can carry greater amounts of cholesterol, which in turn leads to foam cell formation more rapidly, thus increasing the atherogenic capacity⁴⁰. Second, high levels of non-fasting RC increase the expression of inflammatory interleukins and cytokines, which lead to the inflammatory response of monocytes and induce chronic low-grade inflammation⁹. As a result of low-grade inflammation, inflammatory cytokines stimulate monocytes to migrate to the arterial intima, and then monocytes ingest lipoprotein particles and transform into foam cell macrophages⁸. Overall, not only is RC associated with the formation of atherosclerosis plaques, but it is also associated with local inflammation. Ultimately, both processes can lead to plaque rupture, resulting in major adverse cardiovascular events.

Besides, a growing body of research suggests that people with high levels of RC are more prone to metabolic disorders, such as overweight/obesity, diabetes mellitus and metabolic syndrome, which contribute to CVDs^18,41. In the post-statin era, RC-lowering medications may be an important adjunctive or additional option to statin therapy in addressing the residual risk of cardiovascular events. Firstly, RC levels can be reduced by making the following lifestyle changes, including weight control, smoking cessation, exercise, reduction of alcohol intake, saturated fat and total calorie intake^42,43. Secondly, several of novel cholesterol-lowering drugs are currently being developed, such as pemafibrate, apolipoprotein C3 inhibitors and angiopoietin-like 3 protein inhibitors, etc^44,45. In the future, intervention approaches aimed at reducing RC may change primary and secondary prevention strategies for patients with elevated RC. In summary, the link between blood RC and CVDs provides exciting opportunities for lifestyle interventions and pharmacological treatments to control and reduce serum RC levels, thus providing new targets for the prevention and management of CVDs. In the present study, higher circulating RC levels are associated with a higher risk of developing ischemic heart disease, unstable angina pectoris, myocardial infarction, cardiac arrest, heart failure, hypertension, major CHD events, coronary atherosclerosis, cardiac arrhythmias, atrial fibrillation and flutter, and coronary artery disease. New ideas and approaches for intervening in the risk factors for high-risk populations are available from this study, thereby reducing cardiovascular risk and the healthcare costs and economic burdens associated with CVDs.

Our study has several advantages as follows. Firstly, the causal relationships between serum RC level and cardiac arrest, unstable angina pectoris, heart failure, hypertension, coronary atherosclerosis, cardiac arrhythmias, cardiomyopathy, and pericarditis are demonstrated for the first time using the GWAS data and MR method. Secondly, several statistical methods are used to test and correct for potential pleiotropy, such as MR-Egger and MR-PRESSO analysis. Finally, as compared to traditional observational studies, this study overcomes the effects of residual confounding effects, thereby increasing the reliability of the findings.

Our study also has several shortcomings. Firstly, due to the data are from the European participants only, therefore, the generalization of our results may be limited. In the future, the findings need to be validated in other regional populations, such as Asia, America, Africa and Oceania, etc. One previous MR study⁴⁶ assessed the potential causal relationship between traditional cardiovascular risk factors and disease risk in East Asian and European populations. The results showed that although the risk factor profiles for ischemic heart disease and atrial fibrillation were similar between East Asians and Europeans, it was suggested that elevated blood pressure may be a more relevant risk factor for the development of ischemic stroke and heart failure in East Asians. Secondly, the Winner’s curse bias may have affected the causal effect estimates in the MR study⁴⁷, due to possible potential associations between SNPs and confounders. Thirdly, the statistical power of several CVDs (pericarditis and cardiomyopathy) was limited by the limited number of sample sizes. Finally, the effect of horizontal pleiotropy may not be completely ruled out, implying that genetic variants do not, or do not only affect outcomes through exposure. However, we performed several sensitivity analyses, indicating the stability of the findings.

Conclusions

In conclusion, this study revealed the causal relationship between genetically determined RC levels and the risk of developing CVDs, thus providing new ideas for the prevention of CVDs. This study provided several causal evidences for the etiology of CVDs and offered targets for prevention and intervention to curb this adverse CVDs epidemic and their associated disease burden.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

ASCVD

Atherosclerotic cardiovascular disease

CHD

Coronary heart disease

CI

Confidence Interval

CVDs

Cardiovascular diseases

FDR

False discovery rate

GWAS

Genome wide association studies

IDL

Intermediate density lipoprotein

IV

Instrumental variable

IVW

Inverse variance weighting

LDL

Low-density lipoprotein cholesterol

MR

Mendelian randomization

OR

Odds ratio

RC

Remnant cholesterol

SCD

Sudden cardiac death

SNP

Single nucleotide polymorphism

TRL

Triglyceride rich lipoprotein

TSMR

Two sample Mendelian randomization

VLDL

Very low density lipoprotein

Author contributions

ZL, XB and JXW provided the idea of the research and designed the study. ZL and WHL made contributions to acquisition of data and analysis and interpretation of data. All authors were involved in writing and revising this manuscript. The authors read and approved the final manuscript.

Funding

This work was supported by the Basic Public Welfare Research Program of Zhejiang Province (LGD20H150001).

Data availability

The data are available in FinnGen study (https://www.finngen.fi/en), the CARDIoGRAM consortium (http://www.cardiogramplusc4d.org/data-downloads/), and the IEU OpenGWAS project website (https://gwas.mrcieu.ac.uk/).

Declarations

Ethics approval and consent to participate

The GWAS data is a database of publicly available datasets, and each study included in it was approved by the local institutional review board and ethics committee.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.