Non-Random Enrichment of Single-Nucleotide Polymorphisms Associated with Clopidogrel Resistance within Risk Loci Linked to the Severity of Underlying Cardiovascular Diseases: The Role of Admixture

Cardiovascular disease (CVD) is one of the leading causes of death in Puerto Rico, where clopidogrel is commonly prescribed to prevent ischemic events. Genetic contributors to both a poor clopidogrel response and the severity of CVD have been identified mainly in Europeans. However, the non-random enrichment of single-nucleotide polymorphisms (SNPs) associated with clopidogrel resistance within risk loci linked to underlying CVDs, and the role of admixture, have yet to be tested. This study aimed to assess the possible interaction between genetic biomarkers linked to CVDs and those associated with clopidogrel resistance among admixed Caribbean Hispanics. We identified 50 SNPs significantly associated with CVDs in previous genome-wide association studies (GWASs). These SNPs were combined with another ten SNPs related to clopidogrel resistance in Caribbean Hispanics. We developed Python scripts to determine whether SNPs related to CVDs are in close proximity to those associated with the clopidogrel response. The average and individual local ancestry (LAI) within each locus were inferred, and 60 random SNPs with their corresponding LAIs were generated for enrichment estimation purposes. Our results showed no CVD-linked SNPs in close proximity to those associated with the clopidogrel response among Caribbean Hispanics. Consequently, no genetic loci with a dual predictive role for the risk of CVD severity and clopidogrel resistance were found in this population. Native American ancestry was the most enriched within the risk loci linked to CVDs in this population. The non-random enrichment of disease susceptibility loci with drug-response SNPs is a new frontier in Precision Medicine that needs further attention.


Introduction
Caribbean Hispanics are underrepresented in genetic studies, limiting knowledge of genetic risk factors for both CVD severity and a poor response to clopidogrel [1].Most of the characterized SNPs associated with CVDs, such as acute coronary syndrome (ACS), coronary artery disease (CAD), and peripheral artery disease (PAD), have been identified in patients of predominantly European ancestry [2,3].Cardiovascular risk factors to inform the disease severity (e.g., type-2 diabetes, a left ventricular ejection fraction (LVEF) < 30%, an adjoined length of stent(s) ≥ 30 mm, smoking) are also predictive biomarkers of a poor response to clopidogrel [4][5][6][7].However, there is no evidence of significant associations between these markers at the genomic level.We hypothesized that genetic loci linked to these risk factors and the response to clopidogrel are inherited together or associated with each other.
Unlike Mendelian-type monogenic conditions with large effect sizes, complex traits affected by many genes with moderate-to-small effect sizes each (e.g., CVDs) are less likely to transfer across populations.Given their diverse evolutionary histories, resulting in differential allele frequency distributions and varying effect sizes across ethnicities of distinct ancestry and degree of admixture, the role of unique genomic architectures on transferability is critical and must be accounted for.Indeed, the poor transferability of findings from European-centered genomic associations to Hispanics influences how well true causal variants are captured by SNPs identified in Europeans and ultimately exacerbate health inequities.To mitigate the impact of population-specific attributes and the resulting poor transferability of genomic results on health equity, it is imperative to take ancestry into account and expand current studies to admixed populations in which the traits of interest are highly prevalent.Caribbean Hispanics are an admixed population with high genetic heterogeneity that gives rise to phenotypic diversity.By accounting for the local ancestry (LA) effect, we may also uncover novel variants associated with the clopidogrel response.Therefore, inferring genetic ancestry while assessing the relationship between ACS/CAD-and PAD-associated SNPs and those associated with clopidogrel resistance is important for reducing health disparities in this underrepresented population.Bioinformatic computational tools have been developed to facilitate the study of the global ancestry composition in individuals.This has permitted us to understand the evolutionary and migration patterns of populations [8,9].Because the SNP frequency can change from one population to another, it could serve as a genetic signature or landmark in population genetic studies.In addition, advancement in local-ancestry inference (LAI) tools has continued, helping elucidate the disease risk and optimal drug therapy in populations [10,11].LAI is described as the ancestry in a particular location of the chromosome of an individual.LAI works by matching a segment of the reference population to a segment of the chromosome of an admixed individual and assigning the respective ancestry.In the case of Caribbean Hispanics, the reference populations include European, African, and Native American.
LAI results will also depend on the applied method.Various methods for LAI have been developed, including the Local Ancestry in adMixed Populations (LAMP-LD) [12], RFMix [13], Efficient Local Ancestry Inference (ELAI) [14], and MOSAIC [15] software.LAMP-LD is a software package for the inference of the locus-specific ancestry in recently admixed populations [12].RFMix is a powerful discriminative modeling approach for rapid and robust local-ancestry inference [13].ELAI performs local-ancestry inference for admixed individuals [14], whereas MOSAIC infers segments of ancestry and characterizes admixture events involving an arbitrary number of genetically distinct groups put together [15].However, a study showed that the most accurate method is RFMix [16].Also, a study comparing LAMP-LD and RFMix showed that RFMix had a shorter run time [17].
Studies inferring local ancestry by using RFMix have been carried out on Hispanics, but not on a cohort of Caribbean Hispanic patients with ACS, CAD, and PAD receiving clopidogrel treatment [18,19].In this study, we ascertained ancestry enrichments at relevant CV loci, as well as non-random associations between clopidogrel-related genomic markers and those linked to the severity of CV diseases, in admixed Caribbean Hispanics.

Study Population
Caribbean Hispanic patients, self-identified as Cuban, Dominican, or Puerto Rican, on treatment with clopidogrel (75 mg/day), were included in the study.The participants were recruited at five medical facilities across the Commonwealth of Puerto Rico: The Cardiovascular Center of Puerto Rico and the Caribbean, San Francisco Hospital, Dr. Hilton Franqui Clinics, Pavia Hospital, and UPR Hospital Dr. Federico Trilla.All participants met the inclusion/exclusion criteria (Table 1) and signed an informed consent approved by the Institutional Review Board (IRB protocol number A4070417).Relevant demographic and clinical data were collected along with a blood sample in two 3.0 mL 3.2% sodium citrate tubes.Medication adherence was assessed via self-report and record reviews, using the highly validated and reliable four-item Morisky Medication Adherence Scale (MMAS-4), as described elsewhere [20].A flow diagram of the study design is depicted in Figure S1.), we decided not to include these patients in further analyses.Furthermore, the role of genomics and the impact of specific genetic markers on the clopidogrel response in stroke patients is less well understood [21].

Identification of Variants
A literature assessment was performed to identify 50 SNPs that were related to CVDs, such as ACS, CAD, and PAD.No date restrictions were employed.We used the NHGRI-EBI GWAS Catalog [22] (https://www.ebi.ac.uk/gwas/, accessed on 29 April 2021) as the search engine for SNPs associated with each of the three CVDs.First, we searched using the key term "acute coronary syndrome" and we arranged the SNPs in order of p-value significance, followed by the selection of the 20 most significant SNPs.Then, we repeated the same procedure using the key term "coronary artery disease".Some SNPs for CAD were already included for ACS; therefore, we evaluated and selected the SNP with the highest p-value for the corresponding condition.For PAD, we used the key term "peripheral artery disease" and we selected only the first 10 most significant SNPs, as the minority of the patients were on treatment with clopidogrel for PAD.The SNP positions were acquired as the IDs for further analysis steps.The positions were based on GRCh37 as the reference genome build.In addition, 10 SNPs corresponding to the top 10 risk locus signals from a GWAS on high on-treatment platelet reactivity (HTPR) performed on Caribbean Hispanic patients on clopidogrel were included (unpublished data).We evaluated whether any of these SNPs associated with either CVDs or clopidogrel resistance could also be expression quantitative trait loci (eQTLs).To this purpose, we employed the Gtex Portal (https://gtexportal.org/home/, accessed on 29 April 2021), an online tool used to evaluate possible eQTLs from our variants of interest.

Local-Ancestry Inference
Genotyped data were used to estimate haplotypes using Shapeit v2.17 software.RFMix software v1 was performed to infer local ancestry by using the code available at https:// github.com/indraniel/rfmix/blob/master/RunRFMix.py(accessed on 7 October 2021).For the reference population, we used genotyped data from 107 individuals of the Iberian population in Spain (IBS) to represent European ancestry, 61 individuals of Yoruba ancestry in Ibadan, Nigeria, (YRI) to represent African ancestry, and 103 Continental Native American individuals from the America population of the Human Genome Diversity Project (HGDP) to represent Native American (NA) ancestry.The IBS and YRI populations came from the NHGRI Sample Repository for Human Genetic Research of the Coriell Institute (https: //www.internationalgenome.org/data-portal/population, accessed on 7 October 2021).The America population was defined as Native American, and the samples came from Surui and Karitiana in Brazil, Piapoco and Colombian in Colombia, and Maya and Pima in Mexico (https://www.internationalgenome.org/data-portal/data-collection/hgdp, accessed on 7 October 2021).

Python Scripts
Because computational tools have allowed us to analyze high-throughput data, we assessed the non-random enrichment in the association between clopidogrel resistance and the severity of underlying ACS, CAD, and PAD by determining the close proximity (±0.5 cM) between SNPs.Three Python scripts were developed fittingly for the purpose of each objective: (1) to estimate the local ancestral proportions by variant, (2) to determine whether variants of interest are found in close proximity (±0.5 cM), and (3) to acquire random variants from an existing dataset.The scripts can be found at https://github.com/mariangelimonero/Non-random_enrichment (accessed on 13 June 2022).The open-source Python v3.10.2 application was used for this purpose.

Ancestry Search
Ancestry data from each individual were obtained via RFMix software (https://github.com/indraniel/rfmix, accessed on 13 June 2022) [17].The Python script was developed to determine the ancestry within each locus genotyped or imputed across the whole genome of the 510 samples and available in the dataset.As the input file, we used the information about the location files in cM and the forward-backward files containing ancestry information for each variant.The script consisted of opening the location files and forward-backward files of each chromosome and appending them to a list.This information was used to gather ancestry data for the variants and was stored as an "all_chr.anc"file extension for each of the samples.A text file with the samples to be included in the analysis was set on the script.Ancestries were matched with the given sample names by appending them to a list.The rs number and the chromosome number were obtained and appended to their respective lists.The two lists were then transformed into a data frame and then concatenated into a single data frame.Also, the list containing the information about the position and cM was transformed into a data frame and concatenated with the data frame of the rs number and chromosome, creating a single data frame.A column with the calculation of the average ancestry by variant was created.Finally, the previously created data frames were concatenated, creating a file with the columns of rs numbers, chromosomes, positions, cM values, average ancestries, and subsequent samples.

Random Search
A Python script was developed to randomly find and select another 60 variants from the dataset.A text file was provided as input, with the positions of the rs numbers of interest that we did not want to include as random variants.The ancestry file developed in the previous ancestry script was used to provide information about ancestry from the 60 random signals.The script was developed to search for n random signals.It was designed to append all rs numbers in a list, excluding the ones of interest.The function shuffle was applied to the list.The output was given with the quantity of variants desired and the columns of rs numbers, chromosomes, positions, cM values, average ancestries, and subsequent samples.

Bracket Search
A Python script was developed to find two variants of interest in proximity (i.e., located within ± 0.5 cM of each other).A list of the variants of interest and the ancestry file developed in the previous ancestry search step were used as the input file.The variants of interest and the ancestry file were added to individual lists.The intersection function was used to determine the variants of interest that were available in the ancestry list.If the intersection was true, then the matching variants were written in the output file.The output file was then converted into a data frame, and we determined whether the variants were +0.5 or −0.5 cM far apart from each other.Repeated matching pair variants were eliminated.This created a csv file with only the position and cM of the variants in proximity.The data from ancestry and matching variants were converted into a csv file with columns of rs numbers, chromosomes, positions, cM values, and subsequent samples.

Statistical Analysis
To characterize the study cohort, a descriptive analysis of the demographics and clinical parameters was performed.Categorical data were summarized as frequencies and percentages.Continuous variables were reported as mean ± standard deviation (SD) and standard error of mean (SEM), minimum and maximum values, and median.Statistically significant differences in the non-random enrichments between groups were assessed, using either chi-square or Fisher's exact probability tests for categorical variables.All statistical analyses were performed using PLINK v1.9.

Data Cohort
This is a secondary analysis of a dataset from a multicenter clinical study protocol (NCT03419325) [20].For this purpose, a subcohort of 510 Caribbean Hispanic patients on treatment with clopidogrel and with full genetic and clinical data available was used.Patients were diagnosed with CVDs such as ACS, CAD, and/or PAD.All diagnoses were obtained from the patient's electronic medical records at each participating medical facility using the corresponding ICD-9/10-CM codes, which were subsequently confirmed by the study cardiologists.Patients receiving either a 600 mg or 300 mg loading dose of clopidogrel followed by a maintenance dose of 75 mg daily or who started with a 75 mg daily maintenance dose (i.e., alone or as a component of dual antiplatelet therapy (DAPT) in mainly post-PCI patients with ACS or stable CAD with elective stenting) were enrolled in the study between January 2018 and June 2020 from five different medical facilities across the island.Enrollment criteria (Table 1) were minimal, with the goal of gathering an "all comers" population representative of real-world clinical practice.We observed high adherence scores (>85%) during the study period.Participants with a significant lack of adherence to therapy were removed from subsequent analyses.The cohort was genotyped for approximately 1.4 million SNPs with the Infinium TM Multi-Ethnic AMR/AFR Genotyping Array (MEGA) BeadChip by Illumina (San Diego, CA, USA).
Table 2 describes the baseline characteristics of the participants (n = 510).The average age of all the participants in this study was 68 years old, with 55% identified as male.Overall, 78.3% were aged 60 years old or above and 18.6% were within the 45-59-year-old range.Largely, most were middle-aged males, with a high prevalence of conventional risk factors (i.e., overweight (28.4 kg/m 2 ), hypertension (83.9%), hypercholesterolemia/dyslipidemias (71.9%), and type-2 diabetes (54.8%), among others).Furthermore, 20% of participants were on proton pump inhibitors (PPIs) (mainly pantoprazole), whereas statins and calcium channel blockers (CCBs) were prescribed in 79.2% and 26.8% of patients, respectively.Patients who were taking aspirin administered as part of DAPT represented 63.3% of the total cohort.In 75.9% of participants, clopidogrel was given for stable CAD/ACS indication.No differences in the baseline characteristics (e.g., comorbidities) existed among the patients after stratifying by carrier status or indication.

Expression Quantitative Trait Locus (eQTL) SNPs
The intronic variant rs55791371 (Tables 3 and 4 and S9) is an eQTL that resides within the gene encoding the BRG1 protein.The rs9349379 SNP is another identified eQTL variant located at an intronic position in the phosphatase and actin regulator 1 (PHACTR1) gene (Tables 4 and S3).Moreover, the rs10455872, rs118039278, and rs140570886 SNPs are also eQTL variants associated with thrombus formation (Tables 3 and 4, and S3).In addition, the rs55791371 variant (Tables 3 and 4, and S9) is an eQTL for the SLC44A2 gene that encodes a choline transporter protein.

Non-Random Enrichment of SNPs
After a literature review, we identified 20 SNPs associated with ACS, 20 SNPs associated with CAD, and 10 SNPs associated with PAD (Table 3).Although some SNPs were repeated in two different CVDs, we kept the SNP for the condition with the strongest evidence.Additionally, we obtained the ten top signals from a GWAS on clopidogrel resistance among Caribbean Hispanic patients in our laboratory (Table 3).The total of these 60 SNPs was analyzed with the Python script developed for the bracket search.To determine whether the SNPs associated with a poor clopidogrel response in our cohort (i.e., top 10 GWAS hits) are occurring close to or within the same genetic loci of SNPs linked to CVDs, our Python script was used.Not a single SNP linked to the severity of CVDs was found either upstream (+0.5 cM) or downstream (−0.5 cM) of the corresponding loci of any SNP associated with clopidogrel resistance.Additional functional information of these 60 SNPs was assessed and is reported in Tables S1-S11 (Supplementary Materials).

Local-Ancestry Inferences (LAIs)
The developed script for the ancestry search was able to identify the locus-specific enrichment of ancestries related to the European (Iberians), African (Yoruba), and Native American contributions within each of the selected SNPs in each individual.However, we were only able to infer the ancestry for 10 out of the 50 SNPs associated with CVDs (Table 4; Figure 1).This is because the available databases from the reference parental populations lacked the necessary information on these other 40 SNPs genotyped in our cohort of Caribbean Hispanics, which is required by RFMix software to make LAIs.As a control group, we also estimated the ancestry proportions at 60 random SNPs for the purpose of ancestry enrichment comparison.Our results showed ≥50% enrichment of Native American, African, and European ancestries in 25, 20, and 2 of these 60 variants, respectively (Table 5; Figure 2).

Discussion
Understanding the genetic architecture of an admixed population is key to the prevention, diagnosis, and treatment of CVDs.Previous GWASs have identified SNPs associated with the risk of CVDs [24,45,46].Although these studies are informative enough about potential risk markers, admixed populations are neglected because most of the participants are of European descent.The top ten signals with the lowest p-values in our GWAS on the HTPR in CV patients on clopidogrel were used for testing their non-random enrichment with risk variants of CVDs.

Discussion
Understanding the genetic architecture of an admixed population is key to the prevention, diagnosis, and treatment of CVDs.Previous GWASs have identified SNPs associated with the risk of CVDs [24,45,46].Although these studies are informative enough about potential risk markers, admixed populations are neglected because most of the participants are of European descent.The top ten signals with the lowest p-values in our GWAS on the HTPR in CV patients on clopidogrel were used for testing their non-random enrichment with risk variants of CVDs.
SNP Annotation and Tissue Expression: Seven of these signals (Table 3) are located close to a cluster of cytochrome P450 genes on chromosome 10q23.33,which is where the highly polymorphic CYP2C19 gene is mapped to (reference GRCh38.p14assembly: from 94,762,681 to 94,855,547).These seven SNPs form a cluster ~8 kb long on chromosome 10, suggesting they are segregated.The closest gene is located more than 1 Mbp away from this cluster, implying possible distant genetic interactions with other genes on chromosome 10.The bio-activation of clopidogrel is mediated by the CYP2C19 metabolic pathway and, therefore, CYP2C19 genotypes are predictive of clopidogrel resistance [47].Particularly, the CYP2C19*2 variant (rs4244285) has been associated with clopidogrel resistance in previous studies on Europeans [48,49].However, rs4244285 was not among the top 10 signals in our study with Caribbean Hispanics.The rs7916697 SNP, an ATHO7 intronic variant encoding a transcription factor for the regulation of the retinal ganglion cells and optic nerve, was among the 10 most significant variants selected instead.Notably, ATHO7 is highly expressed in the liver (i.e., a GTEx value of 10, similar to that in brain tissues (GTEx = 12); https://www.genecards.org/cgi-bin/carddisp.pl?gene=ATOH7 (accessed on 10 September 2023)).Because this GTEx value is one of the highest across all tissues, this gene may be involved in the regulation of other gene expressions in the liver.However, the possible functional importance of other genes co-localized with the SNPs detected on chromosome 10 is unclear.We also selected the intronic variants rs4646743, rs4745950, and rs1900005, located within the region encoding an lncRNA called LINC02640, while the rs1900003 and rs1900002 variants located ~2 kb upstream of the LINC02640 were also considered.In addition, the intronic rs3796692 variant that is related to the ncRNA called LOC105377582 was included.Finally, the single-nucleotide variant rs4021557, located ~7 kb downstream of the MAP2K3 gene involved in the regulation of the MAPK kinase pathway and MAPK kinase-mediated signaling cascades, was considered.
Reports of numerous SNPs associated with CVDs can be found in the available literature [38,50,51].We identified 50 of these genetic variants significantly associated with the severity of CVDs (e.g., ACS, CAD, and/or PAD).These variants (Table 3) were associated with inflammatory pathways, cholesterol transport, and regulation and signaling cascades in previous GWASs [38,50,51].Although most of these CVD-linked variants occur within intronic regions, they can either enhance or reduce gene expressions [52].The eQTLs are genetic loci that alter gene expression and trait variation quantitatively [53].These eQTL SNPs can change a quantitative trait through different mechanisms, such as epigenetic modifications, transcription, pre-mRNA processing, and post-transcriptional processing (mRNA stability and translation rate) [53].
BRG1 is part of the SWI/SNF complex, with ATPase and helicase activity, that participates in chromatin remodeling.Although the underlying pathway related to BRG1 is not yet fully elucidated, a recent human in vitro study demonstrated that BRG1 is highly expressed in hypertrophic cardiomyopathy and can alter myosin expression [54].The rs9349379 SNP in the PHACTR1 gene is an eQTL that has been found to be associated with CAD risk.PHACTR1 protein regulates the remodeling of the heart after myocardial infarction (MI), increasing its expression and the total protein levels after the event [55].MI is commonly caused by coronary artery occlusion due to a thrombus; therefore, eQTL variants associated with thrombus formation (e.g., rs10455872, rs118039278, and rs140570886 in Tables 3 and 4, and S4) are expected to be identified as signals of GWAS significance.These three SNPs have been correlated to one another for affecting the trait expression for lipoprotein A, a low-density lipoprotein (LDL) with apolipoprotein A, and have been associated with CAD and PAD [21,36,37,56,57].The relevance of rs55791371 (Tables 3  and 4, and S9) as an eQTL for the SLC44A2 gene encoding the choline transporter was demonstrated in studies in vitro.Authors have suggested that ATP will be available for platelet aggregation by regulating the choline transport into the mitochondria through SLC44A2, thus resulting in a higher risk for thrombus formation [58].
Non-Random Enrichments: In this study, we found that relevant genetic risk loci associated with the severity of CVDs in Europeans are not enriched with SNPs linked to clopidogrel resistance among Caribbean Hispanic patients.Therefore, we did not identify a possible genetic marker with a dual role of predicting the severity of the CVDs and the response to clopidogrel in this target population.A possible explanation could be the limited number of SNPs to consider from the reference parental populations, narrowing the pool of possible variants to be studied.Databases containing genetic information from worldwide populations need to expand their variant catalog diversity, especially to account for relevant SNPs within the Native American populations.
Due to the emergence of novel bioinformatic tools, the genetically inferred ancestral composition of Caribbean Hispanics can now be better understood.To determine whether the enrichment was different from that observed in Caribbean Hispanics, the resulting ancestral enrichments at each of the 60 random SNPs were compared to those within the variant loci previously found to be associated with CVDs in Europeans.We found that the ancestral composition in the vicinity of the ten CVD-linked variants tested was similar to that around the 60 random variants.We speculate that certain locus-specific ancestry enrichments (e.g., African) at these risk loci can explain, to some extent, a possible breakup of the overlapping between SNPs related to clopidogrel resistance and the severity of CVDs or lack thereof in Caribbean Hispanics.The occurrence of novel haplotype blocks as a result of altered linkage disequilibrium (LD) patterns and the substantial admixture in Caribbean Hispanics might interfere with the expected associations between both types of SNPs.
Major Contribution: Our study is unique in that we carried out a targeted approach using bioinformatic tools and newly developed scripts to identify ancestry enrichments and non-random associations in a quite heterogeneous and admixed population of Caribbean Hispanic patients.Significant enrichments of African and Native American ancestry, but not of clopidogrel-associated genomic markers, were found within these 10 CVD-specific risk loci.We firmly believe that the ancestry enrichments observed at these 10 disease-specific relevant loci linked to the severity of CVDs will help predict the prevalence of significantly different inter-ethnic phenotypes even for highly admixed populations like Caribbean Hispanics.Consequently, such information could be of help for future admixture mapping studies in this population in order to refine the discovery of novel predictive signals by stratifying individuals regarding their ancestral enrichment at the locus of interest.That is, genetic loci showing high enrichment of ancestry related to the ethnic group with the highest prevalence of the surveyed phenotype will be targeted for more in-depth scrutiny, as long as the phenotype shows differential risk by ancestry.
Recently admixed populations, like Caribbean Hispanics, have a genome that is reminiscent of a mosaic, with sections inherited from Africa and others inherited from Europe or native populations.Consequently, genetically inferred global and local ancestry is an important consideration in the drug response specifically in admixed populations at the gene level.The Hispanic population is highly admixed, but their ancestral proportions differ across different ethno-geographic locations due to historical context and other determinants.This admixture highlights challenges for the current CPIC guideline for clopidogrel selection, which does not discriminate between patients with different ethnic-geographic backgrounds.For example, should any amount of enrichment with European ancestry or certain CVD-linked risk loci require an individual to obtain CYP2C19 genotyping prior to benefiting from PGx-guided antiplatelet therapy?Therefore, our results could also help to delineate optimal strategies for the individualization of antiplatelet therapies based on the ancestral origin and non-random CVD-related enrichment of patients at risk.
Study limitations include a relatively modest sample size in the surveyed cohort when compared to the usual larger number of participants in similar studies within other continental populations (i.e., 2750 Europeans in the International Clopidogrel Pharmacogenomics Consortium (ICPC) study) [59].However, our patients represent a more diverse group than that of the ICPC study.Ancestral proportions, both local and global, are inherently biased due to the European-centered nature of the reference genomes used for their inferences [60].Although of minimal impact, collection biases due to the potential for patient noncompliance and the exclusion of stroke patients are additional limitations of this analysis.Finally, our work did not include data regarding the use of alternative antiplatelet drugs (e.g., prasugrel, ticagrelor), which may be preferred in many clinical scenarios.

Conclusions
In conclusion, even though we did not find a non-random enrichment of CVDassociated variants within sites nearby those associated with clopidogrel resistance, there are still possible novel ethnic-specific variants to be discovered and added to reference panels for future inquiries.Our data bring insightful information about the locus-specific ancestry in Caribbean Hispanics, with Native American ancestry being the most enriched within risk loci for variants linked to CVDs.The non-random enrichment of disease susceptibility loci with SNPs related to drug responsiveness needs further attention to improve the experimental designs of studies seeking the adoption of the Precision Medicine paradigm.
The use of valid bioinformatic tools to identify novel or overlooked eQTLs linked to critical biological pathways in this admixed population may further unravel relevant biomarkers of CVD severity, as well as of clopidogrel resistance.We strongly believe that the integration of information about ancestry and genetic risk variants can expedite the identification of underlying disease-related mechanisms in Caribbean Hispanics.Because most of the published studies identifying SNPs linked to CVDs (i.e., ACS, CAD, and PAD) were conducted on mainly Europeans, the assessment of a potential non-random enrichment of SNPs associated with clopidogrel resistance within risk loci linked to the CVD severity in a cohort of Caribbean Hispanics, and the comparison of LAIs at 10 CVD risk loci and 60 additional randomly selected SNPs, will also contribute to further addressing the existing health disparity among Caribbean Hispanics.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board (IRB protocol number A4070417) from the University of Puerto Rico, Medical Sciences Campus.All subjects participated voluntarily and signed an informed consent.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.Written informed consent has been obtained from the patient(s) to publish this paper.

Data Availability Statement:
The data used as the reference populations are Surui and Karitiana in Brazil, Piapoco and Colombian in Colombia, and Maya and Pima in Mexico, found at https: //www.internationalgenome.org/data-portal/data-collection/hgdp (accessed on 7 October 2021).IBS and YRI populations are from the NHGRI Sample Repository for Human Genetic Research of the Coriell Institute found at https://www.internationalgenome.org/data-portal/population (accessed on 7 October 2021).The datasets from the Caribbean Hispanics analyzed during the current study are available from the corresponding author upon request.

Figure 1 .
Figure 1.Ancestry proportions around the 10 SNPs associated with CVDs (ACS, CAD, and PAD) using the Python script for ancestry search.Genotyping data of 510 Caribbean Hispanics were obtained via Infinium Multi-Ethnic AMR/AFR Genotyping Array (MEGA) BeadChip by Illumina ® .RFMix software was used to estimate locus-specific ancestries by using reference populations from

Figure 1 .
Figure 1.Ancestry proportions around the 10 SNPs associated with CVDs (ACS, CAD, and PAD) using the Python script for ancestry search.Genotyping data of 510 Caribbean Hispanics were obtained via Infinium Multi-Ethnic AMR/AFR Genotyping Array (MEGA) BeadChip by Illumina ® .RFMix software was used to estimate locus-specific ancestries by using reference populations from Iberians, Yoruba, and Native Americans.A Python script was developed to infer ancestry proportions from the list of SNPs of interest.Genes 2023, 14, x FOR PEER REVIEW 11 of 18

Figure 2 .
Figure 2. Ancestral proportions of 60 SNPs generated from ancestry search Python script.A total of 510 Caribbean Hispanic patients were genotyped via the Illumina MEGA chip microarray.Genotyped data were phased using Shapeit software.Local-ancestry inferences were obtained via RFMix software.A Python script was designed to identify ancestral proportions for available variants.Ancestries are represented as Iberian population (IBS), Yoruba population (YRI), and Native population from America (NA).

Figure 2 .
Figure 2. Ancestral proportions of 60 SNPs generated from ancestry search Python script.A total of 510 Caribbean Hispanic patients were genotyped via the Illumina MEGA chip microarray.Genotyped data were phased using Shapeit software.Local-ancestry inferences were obtained via RFMix software.A Python script was designed to identify ancestral proportions for available variants.Ancestries are represented as Iberian population (IBS), Yoruba population (YRI), and Native population from America (NA).

Table 1 .
Inclusion and exclusion criteria.
# Because less than 5% of the participants in the study cohort received clopidogrel for indications other than ACS, CAD, or PAD (i.e., a stroke or transient ischemic attack(TIA)

Table 2 .
Characteristics of participants in the study cohort.
Note: Due to rounding errors, percentages may not equal 100%.¶ MACE reports available only in a subset of 305 patients.* Bleeding events are a combination of major and minor events.# All patients with ACS and some with CAD underwent PCI (i.e., percutaneous coronary intervention) or coronary artery stenting.MIs: myocardial infarctions (i.e., STEMI and NSTEMI).CCBs: calcium channel blockers.PPIs: proton pump inhibitors.BMI: body mass index.PAD: peripheral artery disease.

Table 3 .
Identified SNPs associated with ACS, CAD, and PAD in previous literature and SNPs related to clopidogrel resistance in Caribbean Hispanics.

Table 4 .
Ancestry search for SNPs associated with cardiovascular conditions in previous literature reports.

Table 5 .
Ancestry enrichment for 60 random SNPs generated by computational tools.
Chr: chromosome; POS: position in GRCh37; cM: centimorgan.Ancestry data are presented as the average from IBS (Iberian population in Spain), YRI (Yoruba in Ibadan, Nigeria), and NA (American population) of 510 Caribbean Hispanic patients.Position is based on the human assembly GrCh37/h19 genome reference build by the Genome Reference Consortium.