Rare Variants in Genes of the Cholesterol Pathway Are Present in 60% of Patients with Acute Myocardial Infarction

Acute myocardial infarction (AMI) is a pandemic in which conventional risk factors are inadequate to detect who is at risk early in the asymptomatic stage. Although gene variants in genes related to cholesterol, which may increase the risk of AMI, have been identified, no studies have systematically screened the genes involved in this pathway. In this study, we included 105 patients diagnosed with AMI with an elevation of the ST segment (STEMI) and treated with primary percutaneous coronary intervention (PPCI). Using next-generation sequencing, we examined the presence of rare variants in 40 genes proposed to be involved in lipid metabolism and we found that 60% of AMI patients had a rare variant in the genes involved in the cholesterol pathway. Our data show the importance of considering the wide scope of the cholesterol pathway in order to assess the genetic risk related to AMI.


Introduction
Acute myocardial infarction (AMI) is defined as myocardial cell death due to prolonged ischemia [1]. It is the most severe type of coronary artery disease (CAD) and one of the main causes of death in developed countries [2].
Epidemiological studies have identified the following three major categories of risk factors for AMI: unchangeable factors (age, gender, and family history); variable factors (smoking, alcohol intake, lack of exercise, poor diet, high blood pressure, diabetes, dyslipidemia, and metabolism syndrome); and emerging factors (abnormal levels of C-reactive protein (CRP), fibrinogen, coronary artery calcification (CAC), homocysteine, and lipoprotein(a)) [3]. The environmental and lifestyle components of this triad of AMI have been well-documented since the 1960s, starting with the Framingham study [4]. However, it Is important to note that genetic predisposition has been stated to account for 40-50% of the variability in the development of CAD [4,5].
However, to date, the range of genes underlying the heritable component of AMI is not fully known. According to Musunuru et al. (2016) [6], 25% of AMI loci are lipid-related, including the following: LDLR; PCSK9; APOB; SORT1; ABCG5/G8; LPA; ABO; TCF21; SH2B3; APOE; APOA1/A5; LPL; TRIB1; and LIPA. These data reinforce the role of cholesterol homeostasis in the genetic landscape of AMI. According to the Kyoto Encyclopedia of Genes and Genomes (KEGG), the main cholesterol pathway includes more than 40 genes that operate on many planes, including cholesterol uptake, efflux, transport, storage, utilization, and/or excretion [7].
Most studies [8][9][10][11][12][13][14][15][16] have focused on the analysis of a limited number of genes in the cholesterol pathway, converging on the secretion of the HDL and LDL molecules, without considering the complexity of the pathway as a whole. Most of these studies have focused on single nucleotide polymorphisms (SNPs) [12,[14][15][16][17] without taking into account the study of rare variants in the whole pathway. An ambitious strategy of studying the rare variants in the whole pathway is important as a rare variant can be considered, not the cause of the disease, but an additional risk factor [18][19][20].
Thus, in this paper we aimed to study potential rare variants in the cholesterol pathway genes in patients who presented AMI in order to obtain a broader spectrum of variants that may be involved.

Characteristics of the Study Participants
The principal characteristics of the 105 patients included in the study are summarized in Table 1. The data are also shown to be disaggregated between patients with mutation and without mutation. There was not a significant difference between the groups in any of these variables.

Classification of Variants Identified in the Cholesterol Pathway
The 40 genes of the cholesterol metabolism pathway ( Figure 1) presented a mean coverage of 117.06 ± 22.14-fold ( Figure S1).
After the analysis of the sequencing data, 474 unique genetic variants in the codifying region, and ±10 intronic bases of the 40 genes related to cholesterol metabolism, were identified. Considering the variants in the 105 patients, the total number of variants was 8805. On average, each patient carried around 83 variants in the analyzed region. All data are referred to in terms of percentage, except age.

Classification of Variants Identified in the Cholesterol Pathway
The 40 genes of the cholesterol metabolism pathway ( Figure 1) presented a mean coverage of 117.06 ± 22.14-fold ( Figure S1). After the analysis of the sequencing data, 474 unique genetic variants in the codifying region, and ±10 intronic bases of the 40 genes related to cholesterol metabolism, were identified. Considering the variants in the 105 patients, the total number of variants was 8805. On average, each patient carried around 83 variants in the analyzed region.
Next, the filtering strategy described in the methods section was applied ( Figure 2) in order to discriminate between the relevance of each variant. First, 259 common variants, with a frequency of higher than 0.01 in the population, were filtered out, leaving 215 rare variants. Second, the synonymous and intronic variants in more than ±5 bases on the splicing sites were excluded, reducing the variants to 129 (86 filtered out). Those 129 variants were classified using Hass et al.'s classification [21], with slight modifications which are presented in the methods section. Next, the filtering strategy described in the methods section was applied ( Figure 2) in order to discriminate between the relevance of each variant. First, 259 common variants, with a frequency of higher than 0.01 in the population, were filtered out, leaving 215 rare variants. Second, the synonymous and intronic variants in more than ±5 bases on the splicing sites were excluded, reducing the variants to 129 (86 filtered out). Those 129 variants were classified using Hass et al.'s classification [21], with slight modifications which are presented in the methods section.
From the 129 selected variants, 43 had already been described in the Human Gene Mutation Database (HGMD) [22], classified following our strategy as category I ( Table 2, ). A total of 37 of the 43 variants were related to a lipid metabolism disorder or to cardiopathy. Consequently, they were classified as category Ip (known mutation). The mutations categorized as Ip were present in 18 genes, including the top three genes with the most mutations in our patients, namely APOB (n = 11), ABCA1 (n = 4), and LDLR (n = 3) ( Table 2).  From the 129 selected variants, 43 had already been described in the Human Gene Mutation Database (HGMD) [22], classified following our strategy as category I (Table 2, ). A total of 37 of the 43 variants were related to a lipid metabolism disorder or to cardiopathy. Consequently, they were classified as category Ip (known mutation). The mutations categorized as Ip were present in 18 genes, including the top three genes with the most mutations in our patients, namely APOB (n = 11), ABCA1 (n = 4), and LDLR (n = 3) ( Table 2).  The remaining 86 variants were classified as category II. After the assessment of their potential impact on the protein by in silico prediction tools, 46 variants were classified as potentially damaging and, consequently, were classified as category IIp (potential mutation) ( Table 3). In this group, the genes APOB (n = 7), LRP1 (n = 6), and LRP2 (n = 5) were the most represented.

Distribution of the Variants of Interest in the Genes
A total of 80 mutations and potential mutations (category Ip and IIp) were identified in the studied population. These variants were present in 72% of the genes studied (29 out of 40) (Figure 3). The distribution of variants in the genes was uneven. APOB was the gene in which more of these variants were identified (n = 18). Then, genes LRP2, ABCA1, LPA, and LRP1 had between six to eight variants. Finally, the 24 remaining genes had between one to three variants of category Ip (known mutation) and IIp (potential mutation) mutations ( Figure 3). These group had 44% of the variants identified in our population. in which more of these variants were identified (n = 18). Then, genes LRP2, ABCA1, LPA, and LRP1 had between six to eight variants. Finally, the 24 remaining genes had between one to three variants of category Ip (known mutation) and IIp (potential mutation) mutations ( Figure 3). These group had 44% of the variants identified in our population. Figure 3. Bar graph displaying the number of mutations and potential mutations (Y axis) per gene (X axis). Red: more than 8 variants; yellow: between 8-6 variants; grey: less than 6 variants.

Distribution of the Variants of Interest in the Patients
The distribution of the 80 variants included in category Ip (known mutation) and IIp (potential mutation) showed that approximately 60% of the patients (n = 63) had a mutation or potential mutation in one of the genes analyzed ( Figure 4). The 37 variants of category Ip (known mutation) appeared 50 times in 43 patients, and 7 of these patients had more than one variant. The 43 variants of category IIp (potential mutation) appeared 44 times in 37 patients of which 7 had at least two variants.

Distribution of the Variants of Interest in the Patients
The distribution of the 80 variants included in category Ip (known mutation) and IIp (potential mutation) showed that approximately 60% of the patients (n = 63) had a mutation or potential mutation in one of the genes analyzed (

Discussion
In the present study, we examined the presence of rare variants in 40 genes proposed to be involved in lipid metabolism in AMI patients, and we identified mutations in the genes of this pathway in 60% of patients. Therefore, our data highlight the importance of analyzing rare variants in patients with AMI in the cholesterol pathway.
The importance of the cholesterol pathway has been stressed in GWAS studies that show the association between loci in lipid-related genes and susceptibility to AMI [78][79][80][81][82][83][84][85]. The association between SNPs in genes, such as APOB [85], LDLR [19,86,87], PCSCK9 [88], and AMI has been described, while other studies have reported a weak association between variants in ABCA1 and both the incidence of AMI and the risk of symptomatic CAD [89].
Although common sequence variants have been extensively studied in large genomewide association studies, it is also important to understand the contribution of rare variants to the susceptibility of AMI [18,90]. For this purpose, the selection of the genes implicated in the cholesterol pathway and the strategy used to identify rare variants associated with AMI were two key points of this study.

Discussion
In the present study, we examined the presence of rare variants in 40 genes proposed to be involved in lipid metabolism in AMI patients, and we identified mutations in the genes of this pathway in 60% of patients. Therefore, our data highlight the importance of analyzing rare variants in patients with AMI in the cholesterol pathway.
The importance of the cholesterol pathway has been stressed in GWAS studies that show the association between loci in lipid-related genes and susceptibility to AMI [78][79][80][81][82][83][84][85]. The association between SNPs in genes, such as APOB [85], LDLR [19,86,87], PCSCK9 [88], and AMI has been described, while other studies have reported a weak association between variants in ABCA1 and both the incidence of AMI and the risk of symptomatic CAD [89].
Although common sequence variants have been extensively studied in large genomewide association studies, it is also important to understand the contribution of rare variants to the susceptibility of AMI [18,90]. For this purpose, the selection of the genes implicated in the cholesterol pathway and the strategy used to identify rare variants associated with AMI were two key points of this study.
One of the genes studied the most in order to identify low-frequency lipid-associated variants with AMI is LDLR, which codifies the low-density lipoprotein receptor. Approxi-mately 5% of patients with CAD and AMI under the age of 60 years carry heterozygous LDLR mutations [19,86,87]. These mutations are present and equally distributed throughout the gene in the form of exonic substitutions, small exonic rearrangements, large rearrangements, promoter variants, intronic variants, variants in the 3 untranslated sequence, point mutations, splice site mutations, and large deletions [87]. In our study, two missense variants, p.G269D and p.G592E, were identified in one patient each while a third variant, p.T726I, was identified in two patients. These data imply a frequency of variants in the LDLR gene of 0.038, similar to those described in the literature. These three variants were previously described in patients with AMI [91]. However, the impact on the protein is not fully understood because in one study, the authors were not able to observe a disruptive effect on LDL uptake [91]. Meanwhile, in other research [92], it was demonstrated that patients who carried the mutation presented levels of 50% for LDLR expression, LDL-LDLR binding, and LDLR uptake.
However, in our study, the major gene in which variants were found was APOB, which codified the primary apolipoprotein, including 100 chylomicrons, as well as VLDL, Lp(a), IDL, and LDL particles [93]. In fact, 18 variants were identified in the APOB gene in 19 patients, which implies a frequency of mutations in this gene in our cohort of 0.181. It is important to note that one patient carried three variants of the APOB gene (p.P145S, p.T741N, p.L1212M) and that the variants p.T3826M and p.R1128H were present in two and three patients, respectively. Of the 18 variants found in the APOB gene, 11 were previously described as being mainly associated with hypercholesterolemia and hypertriglyceridemia (category Ip (known mutation)), while 7 were only listed in databases, such as Gnomad, with a frequency of less than 0.01 (Category IIp (potential mutation)). Surprisingly, 9 out of 11 of the variants in category Ip (known mutation) were not previously associated with AMI [37,39,43,[94][95][96][97][98]. However, two of them, namely p.P2821L and p.S2429T, have been described in patients with cardiovascular artery diseases [39,45,46]. Moreover, our study is in concordance with previous studies that suggest that rare deleterious mutations in APOB, such as R3500Q/W, confer a higher risk of ischemic cardiovascular disease in mutation carriers [99]. In fact, the importance of APOB levels has been stressed because the Copenhagen City Heart Study showed that apolipoprotein B levels are associated with ischemic diseases [99].
The most important finding of this study was that 60% of AMI patients had a rare variant in genes involved in the cholesterol pathway. This fact confirms the importance of not focusing only on the gene with a higher presence of rare variants, such as LDLR in the literature or APOB in our case, but rather on the 40 selected genes of the cholesterol pathway. In our study, we have found more than three mutations in LRP2, ABCA1 [89], LPA, LRP1, CD36 genes. Additionally, we found between one and three rare variants in genes such as LDLR and PCSK9 which, in other studies, had a high impact on their association with AMI [78,79,88,100]. In fact, the low frequency of variants in the LDLR and PCSK9 genes in the group is surprising and supports the idea of the importance of analyzing rare variants in the entire pathway of cholesterol genes.
It is important to highlight that 11 genes of the 40 included, in which we did not identify rare variants, have been described in the literature to have an association with AMI [101][102][103][104][105][106][107][108][109][110][111]. Therefore, these genes should be further analyzed in following studies in order to increase the 60% of patients with rare variants detected in this study.
The main limitation of this study is that it relies on the fact that it was a descriptive study, not an association study. The sample size is small for an association study of rare variants that present a frequency lower than 0.01. We have not been able to associate the variants with cholesterol levels because in many cases there was no previous data on cholesterol levels prior to infarction due to AMI being the first event. Moreover, because a functional validation of each variant was not performed, this method may have led to a misclassification in some cases.
AMI is a pandemic in which the risk of patients can be assessed; however, conventional risk factors are inadequate to detect who is at risk early in the asymptomatic stage [4].
Genetic risks, which can be determined at birth, could help to identify patients at higher risk of presenting AMI [4]. Our data show that rare variants in genes related to cholesterol are present in AMI patients. In future clinical settings where genomic sequencing might be available for all patients, the evaluation of genetic risks would be improved by incorporating variants of the genes of the cholesterol pathway.

Study Population
In this study, 105 patients diagnosed with AMI with an elevation of the ST segment (STEMI) and treated with primary percutaneous coronary intervention (PPCI) in A Coruña University Hospital (Spain) between July 2017 to January 2021, were included.
Written informed consent was obtained from every patient included in the study. The protocol of this study was in accordance with the principles of the Declaration of Helsinki, and the "Comité de Ética de la Investigación de Galicia" (ref: 2016/299) also approved it. Blood samples were collected at the time of the hemodynamic procedure and stored at −80 • C until analysis. All samples were included in the biobank of the National Biobank Network of "Instituto de Salud Carlos III" (C.0002483, 2013/109).
A chi-squared test was performed for the following variables: sex; dyslipidemia; hypertension; diabetes; and tobacco, as these variables were considered dichotomous. For age, an ANOVA test was performed.

Selection of Genes Related to Cholesterol Metabolism
To establish the main genes involved in the cholesterol metabolism pathway, the KEGG map04979 and WikiPathways WP4522 databases were analyzed. Therefore, 40 genes related to the cholesterol metabolism pathway were selected. All genes selected are involved in cholesterol uptake, efflux, transport, storage, utilization, and/or excretion.  (Table S1).

Next-Generation Sequencing
The strategy for exome sequence data analysis included several computational tools. The raw data files in the binary base call (BCL) format, generated by the NextSeq Sequencing System, were demultiplexed and converted to a standard FASTQ file by bcl2fastq Conversion Software v2.17. The coverage ( Figure S1) and Q-score were analyzed in order to establish the quality of the sequence. Based on the guidelines of the American College of Medical Genetics and Genomics, if the region analyzed presented a sequencing mean depth <30, the region was considered unsuitable for analysis. Furthermore, the threshold in the Q-score established was 30 (base call accuracy of 99.9%).
Computational biology sequence alignment to the human genome version GRCh37/ hg19 was performed with Burrows-Wheeler Aligner software 7.17, and the BAM file output was obtained. Variants were detected using the Genome Analysis Toolkit (GATK), and the output files were VCF files. All exonic regions and ±10 intronic regions of the 40 genes related to cholesterol metabolism were visualized via the Integrative Genome Viewer (IGV).

Variant Annotation and Classification
All variants identified in this study were searched in the Genome Aggregation Database (gnomAD) [112] and the Single Nucleotide Polymorphism Database (dbSNP) [113] in order to establish frequency in the general population. Moreover, to identify previous genotype-phenotype associations, all variants were checked in HGMD [22].
After the annotation of all variants, variant filtering was performed using an adapted variant filtering and prioritization strategy described by Akinrinade et al. [114] (Figure 2). In detail, the first filter used was the frequency of the variants was described in the Gnomad and/or dbSNP to be higher than 0.01. This frequency was selected following the classical classification of common variants. The second filtering excluded the synonymous and intronic variants located farther than five bases from the acceptor/donor splice sites.
The remaining variants were classified into well-defined categories described by Hass et al. [21] with slight modifications. We defined two categories with a subcategory each for the determination of the likelihood of being disease-relevant mutations (Figure 2): (a) Category I consisted of variants previously described in the literature (HGMD) to be associated with pathologies. Variants of this group associated with lipid metabolism disorders or cardiopathies were classified as Ip (known mutation). (b) Category II included all variants that were not previously pathology-associated in the literature. The subcategory IIp (potential mutation) included variants with more than three out of five in silico predictions tools to be pathogenic and all the stop-gained variants.
Thus, category Ip (known mutation) mutations were considered known mutations and category IIp (potential mutation) mutations were considered potential mutations.

Conclusions
A high prevalence of rare variants in the genes of the cholesterol pathway can be found in AMI patients. Our data show the need to consider not only one gene or a small set of genes, but the wide scope of cholesterol pathway genes in order to make a more realistic assessment of the risks related to AMI. Funding: This work was supported by a grant from "Instituto de Salud Carlos III" (PI18/01737) and a non-conditional grant from Abott Vascular.

Institutional Review Board Statement:
The protocol of this study was in accordance with the principles of the Declaration of Helsinki, and the "Comité de Ética de la Investigación de Galicia" (ref: 2016/299) had approved it. All the samples were included in the biobank of the National Biobank Network of "Instituto de Salud Carlos III" (C.0002483, 2013/109).
Informed Consent Statement: Written informed consent was obtained from every patient included in the study.

Conflicts of Interest:
Authors declare no conflict of interest.