The genetics of common variation affecting platelet development, function and pharmaceutical targeting


Andrew D. Johnson, The Framingham Heart Study, 73 Mt. Wayte Ave., Suite #2, Framingham, MA 01702, USA.
Tel.: +1 508 663 4082; fax: +1 508 820 0340.


Summary.  Common variant effects on human platelet function and response to anti-platelet treatment have traditionally been studied using candidate gene approaches involving a limited number of variants and genes. These studies have often been undertaken in clinically defined cohorts. More recently, studies have applied genome-wide scans in larger population samples than prior candidate studies, in some cases scanning relatively healthy individuals. These studies demonstrate synergy with some prior candidate gene findings (e.g., GP6, ADRA2A) but also uncover novel loci involved in platelet function. Here, I summarise findings on common genetic variation influencing platelet development, function and therapeutics. Taken together, candidate gene and genome-wide studies begin to account for common variation in platelet function and provide information that may ultimately be useful in pharmacogenetic applications in the clinic. More than 50 loci have been identified with consistent associations with platelet phenotypes in ≥ 2 populations. Several variants are under further study in clinical trials relating to anti-platelet therapies. In order to have useful clinical applications, variants must have large effects on a modifiable outcome. Regardless of clinical applications, studies of common genetic influences, even of small effect, offer additional insights into platelet biology including the importance of intracellular signalling and novel receptors. Understanding of common platelet-related genetics remains behind parallel fields (e.g., lipids, blood pressure) due to challenges in phenotype ascertainment. Further work is necessary to discover and characterise loci for platelet function, and to assess whether these loci contribute to disease aetiologies or response to therapeutics.

Importance, measurement and heritability of platelet traits

Platelets play critical roles in thrombosis and haemostasis and are the targets of multiple pharmaceutical targeting strategies and endogenous activation pathways. Severe inherited disorders impacting platelet function have been described underlining the clinical relevance of platelet function [1,2]. Evidence further indicates that successful modulation of platelet function provides significant survival benefit, and that variation in platelet volume or function may be an independent risk factor and predictor for disease or outcomes [3–12]. Although studies of anti-platelet drug resistance have significant limitations in terms of phenotypic definition, there is general agreement that a significant portion of individuals exhibit anti-platelet resistance [13,14]. This, along with the underlying importance of platelet biology and the severity of rare platelet disorders, has provided significant motivation for the study of platelet-related genetics.

This article reviews knowledge on common genetic influences on platelet development, function and pharmaceutical targeting, with particular emphasis on recent developments employing genomic technologies. In particular, a catalogue of > 450 platelet-related genome-wide association study (GWAS) results is provided in Table S1. Common platelet traits in genetic studies include assays of cell parameters usually performed in whole blood (e.g., platelet count [PLT] and mean platelet volume [MPV]), measurements of platelet function by aggregation response in whole blood or platelet rich plasma (PRP) to a variety of agonists (e.g., collagen), mimetics or stimulants (e.g., shear stress), and less direct measurements of platelet function (e.g., P-selectin levels, fibrinogen levels). Each measurement has potential limitations and focuses most on specific portions of the platelet life cycle or activation cascades.

One gold standard for platelet function testing is light aggregometry in PRP [15], but this measure has a drawback for genetic studies in being challenging to scale to the large sample numbers required to achieve significant statistical power in outbred populations. Cell counts, mainly PLT and MPV in whole blood, are more scalable, and thus have been more widely applied in large genetic studies. While PLT and MPV provide less specific information about platelet function, with modern methods they have reasonable coefficients of variation [16], and the largest genetic study to date on platelet function provided evidence for partial overlap with PLT/MPV loci indicating they may be good proxy measurements for discovering some loci that contribute to platelet function variability [17].

A high degree of heritability for platelet quantitative traits has been widely established, first with twin studies of PRP epinephrine-induced aggregation, PLT and MPV [18,19]. Studies on the heritability (h2) of platelet measures are summarised in Table 1 with the general pattern for h2 being: MPV > PLT > platelet function tests. A large study of PRP aggregation in families in the Framingham Heart Study (FHS) established the heritability of platelet function to a range of agonists within a general population [20]. The Johns Hopkins GeneStar (GS) study later provided further evidence of heritability in both European ancestry and African ancestry families, and to both whole blood (WB) and PRP responses with a range of measures, and in pre- [21] and post- [22] aspirin conditions. Animal model studies, particularly in families of baboons [23] and mouse crosses [24,25], confirm the importance of inherited alleles to platelet measures in other species.

Table 1.   Heritability studies for platelet-related traits
Phenotype descriptionCohort descriptionHeritability observationsCitation
  1. *Heritabilites are given for the two populations as European ancestry/African ancestry. Where multiple doses were studied the maximum heritability is given. Heritabilities under a model fully adjusted for pre-aspirin treatment and other covariates are given. n.s. indicates the authors did not accept the hypothesis that the trait was significantly heritable. PCT, plateletocrit; PDW, platelet distribution width.

Epinephrine-induced PRP aggregation32 male twin pairs (17 MZ, 15 DZ)PRP Epi = 0.65 in MZ
PRP Epi = 0.43 in DZ
Gaxiola et al. [19]
PLT, MPV (Coulter)92 male and female twin pairsPLT = 0.86, MPV = 0.88Whitfield and Martin [18]
PLT105 twin pairsn.s.Dal Colletto et al. [81]
PLT (Coulter)385 twin pairs (192 female, 193 male)PLT = 0.80Evans et al. [35]
PLT (Bayer H3 RTX autoanalyzer)712 twin pairs (234 MZ, 478 DZ)PLT = 0.57Garner et al. [82]
Epinephrine, ADP and collagen induced PRP aggregation1041 siblings and 928 spousesADP = 0.44
Epinephrine = 0.44
Collagen = 0.62
O’Donnell et al. [20]
PLT6111 Italian ancestry individuals in SardiniaPLT = 0.64Pilia et al. [83]
Whole blood aggregation (ADP, Collagen, AA, Tbx), PFA closure epinephrine (PFA-100), PRP aggregation (ADP, Collagen, AA, Epinephrine), PLT, MPV687 European ancestry, 321 African ancestry individuals in familiesPLT = 0.60/0.67*
MPV = 0.55/0.71
WB ADP = 0.51/0.39
WB Collagen = 0.28/n.s.
WB AA = 0.58/n.s.
PFA closure(epi)  = 0.24/0.61
PRP Collagen = 0.26/0.79
PRP Coll. Lag = 0.50/0.47
PRP AA = 0.38/n.s.
PRP Epi = 0.36/0.76
PRP ADP = 0.42/0.78
Bray et al. [21]
Post-aspirin responses with or without adjustment of baseline measures: most as above in Bray et al., additionally TXB2 (ELISA), Tx-M (ELISA), beta-thromboglobulin release (ELISA)1144 European ancestry, 736 African ancestry individuals in familiesPRP Collagen = 0.29/0.39*
PRP Coll. Lag = 0.23/n.s.
WB Collagen = 0.27/0.28
PRP ADP = 0.22/n.s.
WB ADP = 0.20/n.s.
PRP Epi = 0.33/n.s.
PFA closure(epi) = n.s./n.s.
TXB2 = 0.38/n.s.
Tx-M = n.s./0.35
B-thromboglobulin = 0.33/n.s.
Faraday et al. [22]
PLT, MPV, PCT, PDW1101 Italian ancestry individuals in S. TyrolPLT = 0.55, MPV = 0.79, PCT = 0.53, PDW = 0.37Marroni et al. [84]
PLT, MPV1803 Italian ancestry ind. in Val BorberaPLT = 0.60, MPV = 0.73Traglia et al. [85]
PLT (Coulter)12 517 individuals in SardiniaPLT = 0.54Biino et al. [86]

Early studies into the genetics of platelet traits

Early studies into the genetics of specific platelet-related genes largely stemmed from a focus on cells derived from alloimmune thrombocytopenias and efforts to identify the alloantibodies [26]. Isolation and sequencing of cDNAs due to the anuclear nature of platelets led to the discovery of coding region polymorphisms. Early variant discovery was restricted by this approach, the nascent state of information on the human genome sequence and infeasibility of sequencing large populations for the coding regions and surrounding sequence. Most early studies focused on cell surface receptors and their ligands, a candidate gene approach, as these were among the most highly expressed RNAs and proteins with recognised roles in platelet function. A series of coding and proposed regulatory polymorphisms were further studied for their functional effects in association with diseases including venous thrombosis (VT), stroke and myocardial infarction (MI).

A thorough review of platelet candidate gene studies is beyond the scope of this review. Readers are directed to other recent reviews that cover this topic in more depth [26–28]. Overall, studies from the candidate gene era did not produce many results that were consistently replicated in independent populations for roles in either platelet function and/or disease. There are a variety of potential explanations for this observation including the heterogeneity in ex vivo assays used in platelet function studies. However, the modest successes of candidate gene studies in platelets are not inconsistent with the experiences of other genetics fields of this era (e.g., blood pressure genetics), and were likely due to limited statistical power in small population samples and the relatively incomplete knowledge of common alleles in human populations available in that time period.

There are a few notable exceptions of platelet candidate genes and polymorphisms that show relatively consistent positive results in the literature including the GP6 and ADRA2A receptors. Another important exception is CYP2C19, a cytochrome P-450 enzyme responsible for metabolism of the pro-drug form of clopidogrel, polymorphisms of which have been consistently associated with clopidogrel response. Some candidate genes have re-emerged in genome-wide studies (see below) providing a validated role for these genes in genetic variability [17,29]. A similar pattern has been observed for blood pressure loci [30] and other traits [31,32], showing that the information gained from candidate gene studies remains valuable in the genome-wide era.

Genome-wide human linkage and animal QTL studies

First attempts at genome-wide genetic studies for platelet traits employed microsatellite studies to conduct linkage analysis in families, or quantitative trait locus (QTL) mapping in animals. Most linkage analyses use a moderate number of markers and search for cosegregation of markers in individuals’ within families with a disease or other trait, generating LOD scores or similar statistics. Results from linkage and QTL studies are summarised in Table 2. While these studies produced several genome-wide significant LOD scores, and even more suggestive signals, in most cases they did not lead to fine mapping, identification or replication of novel platelet function genes. An exception is a linkage study of PLT in Asian Indian pedigrees which fine-mapped putative functional mutations in THPO and GP9 [33]. Remarkably, a recent GWAS for PLT in Japanese individuals had a strong peak signal at the same THPO variant (rs6141) identified by fine mapping of the linkage scan [34].

Table 2.   Human linkage and animal QTL studies for platelet traits
Phenotype descriptionSample descriptionHuman or orthologous human regions (LOD)Citation
  1. MCV, mean corpuscular volume; Hb, hemoglobin; RDW, red cell distribution width; PDW, platelet distribution width; PCT, plateletocrit.

Animal studies
 PLT (Sysmex XE9000 hematology analyzer)1126 F2 progeny from a CBA/CaH × QSi5 intercrossOrthologous regions not given
PLT (LODs 11.2, 6.8, 3.6, 3.5, 3.4)
Cheung et al. [24]
 PLT, MPV (Advia 120 whole blood analyzer)279 F2 progeny each from crosses of NZW/LacJ × SM/J and C57BLKS/J × SM/JMPV – 12q12-13 (24.2)
MPV – 16q13-p21 (2.7)
PLT – 9q33-34 (3.5)
PLT – 11p15 (3.3)
PLT – 2p25 (3.2)
Peters et al. [25]
 PLT, MPV (Coulter)582 baboons (410 female, 172 male) in 11 pedigreesPLT/MCV – 1p34.2-p36.23 (3.1)
PLT/Hb – 19p13.12-q13.41 (2.2)
PLT/MPV – 10q26.3 (2.0)
PLT/RDW – 3p23-p22.3 (1.6)
PLT/RDW – 10q26.3 (1.5)
MPV/RDW – 20p12.1-q11.22 (2.4)
MPV/RDW – 10q23.3 (2.0)
MPV/MCV – 9q34.2-q34.3 (1.7)
MPV/RBC – 8p21.3-p22 (1.6)
Bertin et al. [23]
 PLT, MPV, PDW (CD1700 whole blood analyzer)1033 F2 progeny of White Durox × Erhualian porcine crossesOrthologous regions not given
PLT (F-val 12.1, 10.1)
PDW (F-val 21.7, 11.1, 9.6) MPV (F-val 10.1)
Yang et al. [87]
 PLT, MPV, PDW, PCT (TEK-II mini automatic hemocyte analyser)368 purebred piglet (Landrace, Large White, Songliao Black)Orthologous regions not given
PLT (LR 17.89, 14.61)
PDW (LR 22.24, 19.12)
Gong et al. [88]
Human studies
 PLT (Coulter)745 twin pairs (327 MZ, 418 DZ)PLT – 19q13.11-q13.32 (2.59)
PLT – 10q11.23-q22.1 (2.18)
PLT – 5q32-q33.2 (1.97)
PLT – 2q24.3-q31.1 (1.42)
PLT – 5q23.1 (1.26)
PLT – 15q11.2 (1.15)
Evans et al. [35]
 PLT (Bayer automated analyzer)6 Asian Indian families (125 with PLT + markers)PLT – 3q23 (3.26)
PLT – 3q26.2 (2.52)
Garner et al. [33]
 Epinephrine, ADP and collagen induced PRP aggregationUp to 724 individuals in familiesADP – 7q21.12 (2.0)
Collagen – Xp11.4 (2.0)
Yang [57]
 Pre- and Post-aspirin responses with or without adjustment of baseline measures: most as above in Bray et al. Table 1, additionally Tx-M (ELISA)1231 European ancestry, 846 African ancestry individuals in familiesWB ADP post AA – 5q11.2 (3.6)
WB ADP post AA – 5q12.3 (3.6)
WB ADP post AA – 5q13.1 (2.2)
WB ADP post EA – 5q35.3 (2.2)
WB epi post EA – 3p14.2 (2.2)
WB epi post EA – 11q23.3 (2.2)
WB coll. pre AA – 1q23.3 (2.2)
WB coll. post EA – 4p12 (2.2)
WB coll. post EA – 4p14 (2.2)
WB coll. post EA – 4q12 (2.2)
WB coll. post EA – 6q25.3 (2.2)
Coll. lag pre AA – 19q13.41 (2.2)
Mathias et al. [36]

There are a few notable consistencies across the human and animal studies. Human chromosome region 19q13.41 was encompassed in linkage peaks for PLT in twins [35], for collagen lag time response in African-Americans [36], and for an orthologous region in the baboon genome in a bivariate analysis of PLT and haemoglobin levels [23]. This region contains GP6, the direct receptor for platelet activation by collagen, and a previous candidate gene with characterised functional coding alleles. This locus was further validated in a recent GWAS for PRP collagen lag time [17].

Regions orthologous to human chromosome 9q33-q34.3 were identified in mouse PLT studies [25] and in bivariate analysis of MPV and mean corpuscular volume in baboons [23]. Notably, this large region of the human genome contains the human blood group antigen locus ABO. While this locus has not emerged in large human linkage or genome-wide association studies (GWAS) of PLT, MPV or platelet function to date, some candidate studies have indicated an effect of ABO blood group on particular platelet function measures [37–39]. Furthermore, a GWAS study of soluble P-selectin and ICAM-1 levels found the ABO locus strongly linked with these measures [40]. The effects of ABO-type matching on platelet transfusions may be of potential important to clinical outcomes. A meta-analysis indicates that ABO-matched transfusions consistently result in higher outcome platelet counts [41], though further studies are needed to determine conclusively whether this is of definitive clinical benefit.

Functional genomics studies of platelet traits

Genome-wide genetic studies, transcriptomic (RNA) and proteomic studies have the potential to dramatically advance scientific understanding because they are relatively unbiased and current technologies probe a significant fraction of common human variation, expressed transcripts and proteins. Genome-wide genetic studies have flourished due to extensive maps of common genetic variation (e.g., the HapMap and the 1000 Genomes Project), as well as the stability of DNA and relative availability of samples in large biobanks and cross-sectional population studies. These studies, referred to as GWAS, employ hundreds of thousands or millions of SNP genotypes (or estimates) per individual to conduct large-scale genetic association analysis usually in population-based samples. Transcriptomic and proteomic studies are challenged more than DNA studies by sample availability, technical challenges in assay design and coverage, and the potential impact of confounder variables like tissue specificity and sample quality. In particular, due to the transient platelet life span, challenges in isolation of pure cell populations, and lower transcript and protein copy numbers such studies in platelets face major challenges. Nonetheless, several studies have shown value in non-genetic ‘omics’ approaches to uncovering novel genetic candidates in platelet function.

Platelet endothelial aggregation receptor-1 (PEAR1) was a relatively unknown human gene and receptor until it was characterised in a proteomic screen for proteins that undergo tyrosine phosphorylation upon platelet-platelet cell contact [42]. Subsequent candidate gene studies consistently found PEAR1 to be associated with platelet aggregation response to multiple, diverse agonists [43,44], and in both pre- and post-aspirin treated conditions [43]. In the largest study of PEAR1 to date, the combined meta-analysis of PRP aggregation GWAS results from GS and FHS, we found that peak variants in intron 1 of PEAR1 are strongly associated with both ADP- and epinephrine-induced aggregation [17]. Furthermore, we found that intron 1 variants were associated with PEAR1 protein levels in platelet lysates [45]. PEAR1, brought to light by proteomics, may be an important novel platelet signalling regulator, a potential drug target and its variants may play a role in platelet-related response variability in the general population, all of which require further detailed study.

Efforts at transcriptome profiling in megakaryocytes and platelets have been undertaken in relatively small sample sizes and often in clinically affected samples, thus our knowledge of transcriptome profiles in the general population is still limited ([n = 60] [46]; [n = 126] [47]; [n = 29] [48]; [n = 4 megakaryocyte samples] [49]; [n = 37] [50]). Despite limited sample sizes, these studies can provide important clues supporting platelet-related genetic signals, or suggest additional candidates for further study. Based on transcript profiling of 37 samples over normal ranges of platelet response, Goodall et al. [50] prioritised several genes for further study, finding that COMMD7 and LRRFIP1 show evidence of genetic association with MI. In follow-up functional studies employing gene silencing and proteomics approaches LRRFIP1 demonstrated evidence for effects on thrombus formation and interaction with the platelet cytoskeleton [50].

Notably, of 63 transcripts identified as correlated with platelet responses to ADP and/or a collagen mimetic [50], several are transcribed from loci, or have similar function to loci recently implicated in GWAS for platelet aggregation in PRP (RGS18; ST3GAL3 vs. ST3GAL4; ATP6V0D2 vs. ATP6V1F from [17]) and in GWAS for MPV (DNM3 [51]). Targeted functional reports (e.g., RTPCR) also provide valuable information. In published MPV and PRP GWAS studies the integration of megakaryocyte transcriptome data [49] as well as targeted functional studies on particular genes shows some loci contain transcripts or proteins with known significant expression and/or function in platelets and/or megakaryocytes [17,51]; see evidence codes ‘(R)NA’, ‘(P)rotein/roteomics’, ‘(S)ignaling’ in Tables 3 and 4. These examples demonstrate that functional genomics screening experiments can provide valuable information to guide new genetic studies or aid in the interpretation of existing study results.

Table 3.   Genome-wide association (GWAS) results for platelet traits with consistency in two or more human population samples and association with two or more distinct traits in either the same or different studies
Gene*Chr regionPhenotype associationAssociation P-valPop.No. in PMNo. in PM (PLT)Further evidence
  1. *Some regions contain > 1 gene near the genetic signal. Either the closest gene or strongest candidate gene is given. Two PubMed search was performed for each gene name on March 16, 2011. One search was only with gene name. The second search was with gene name AND (platelet OR megakaryocyte OR thrombosis OR hemostasis). The GWAS studies considered here are not counted among the results. Population abbreviations are: EA, European ancestry; AA, African ancestry; AS, Asian ancestry; H, Hispanic. Further evidence abbreviations: AM, animal model; DM, drug metabolism function; FD, functional domain; HS, human study of platelet function/disease; MK, role in megakaryocyte development or function; P, protein or proteomics evidence for a role in platelets/MK; R, RNA gene expression evidence for role in platelets/MK; S, signaling role known in platelets/MK.

CYP2C1910q23.33Clopidogrel response and outcomes [29]1.5e-13EA, AA, AS, H> 500192DM, FD, HS
PEAR11q23.1PRP Epi, ADP [17], Aspirin PRP response [43], WB coll [44]4.9e-19, 3.8e-16, 3.8e-4, 2.6e-4EA, AA22P [42,45]
PIK3CG7q22.3PRP Epi [17], MPV [51,56,56]3.1e-9, 1.6e-33EA21013AM [59,60],R [56],S [59,60]
JMJD1C10q21.2PRP Epi [17], MPV [51], PRP ADP [54]1.6e-8, 3.3e-21, P < 0.05EA, AA100MK [65]
MRVI111p15.4PRP ADP, Epi [17]2.0e-8, 7.6e-7EA, AA162AM [58],R [89],S [58]
RGS181q31.2PRP Epi, ADP [17]6.8e-7, 4.0e-5EA, AA196R [90–92],P [93], S [93]
TAOK117q11.2MPV [51,55], PRP ADP [17]1.4e-22, 6.9e-5EA30 
P2RY123q25.1PRP ADP, Epi [17]8.2e-6, 9.5e-4EA453395AM, FD, HS, R, P, S
GP1Bα17p13.2PLT[34]2.1e-12AS2516AM, FD, HS, R, P, S
PRP Epi [17]0.03EA
PIP5K1B9q21.11PRP ADP, Epi [17]2.2e-7, 8.0e-6EA50 
HMGB1L120q13.31PRP ADP, Epi [17]1.1e-6, 5.2e-6EA00 
SETBP118q12.3PRP Epi, ADP [17]1.3e-6, 5.4e-6EA80 
FBXL75p15.1PRP Epi, ADP [17]2.9e-6, 2.3e-5EA10 
ATP6VOD28q21.2-q21.3PRP Epi, ADP [17]7.9e-6, 8.7e-6EA00 
KIAA080218p11.22PRP Epi, ADP [17]9.6e-6, 7.2e-5EA00 
FLJ3974315q26.3PRP Epi, ADP [17]1.5e-5, 3.3e-5EA00 
WBSCR177q11.22PRP ADP, Epi [17]1.7e-5, 6.3e-5EA30 
CUBN10p13PRP ADP, Epi [17]1.9e-5, 8.1e-5EA170 
MIPOL114q13.3-q21.1PRP Epi, ADP [17]2.2e-5, 4.9e-5EA50 
NUP2103p25.1PRP Epi, ADP [17]4.2e-5, 8.8e-5EA440 
SVIL10p11.23PRP Epi, ADP [17]4.4e-5, 8.2e-5EA, AA170 
THSD415q23PRP ADP, Epi [17]4.4e-5, 8.2e-5EA30FD
STMN48p21.2PRP ADP, Epi [17]5.6e-5, 9.0e-5EA230 
KLHL316p12.1PRP Epi, ADP [17]6.0e-5, 6.3e-5EA20 
GMDS6p25.3PRP ADP, Epi [17]6.1e-5, 8.1e-5EA00 
ADAMTS25q35.3PRP Epi, ADP [17]6.8e-5, 7.1e-5EA240FD
Table 4.   Genome-wide association (GWAS) results for platelet traits with consistency in two or more human population samples with strong association with only one trait
Gene*Chr regionPhenotype associationAssociation P-valPop.No. in PMNo. in PM (PLT)Further evidence
  1. *Some regions contain > 1 gene near the genetic signal. Either the closest gene or strongest candidate gene is given. Two PubMed search was performed for each gene name on March 16, 2011. One search was only with gene name. The second search was with gene name AND (platelet OR megakaryocyte OR thrombosis OR hemostasis). The GWAS studies considered here are not counted among the results. Population abbreviations are: EA, European ancestry; AA, African ancestry; AS, Asian ancestry. Further evidence abbreviations: AM, animal model; FD, functional domain; HS, human study of platelet function/disease; MK, role in megakaryocyte development or function; P, protein or proteomics evidence for a role in platelets/MK; R, RNA gene expression evidence for role in platelets/MK; S, signaling role known in platelets/MK.

ADRA2A10q25.2PRP Epi [17]3.2e-12EA, AA> 500> 500FD, HS [94], R [95], P [96]
NEURL10q24.33PRP Epi [17]2.5e-7EA, AA101 
PRNP20p13PRP Epi [17]1.7e-6EA, AA> 5001HS [97], P [98]
TRIM276p22.1PRP Epi [17]2.4e-6EA, AA540 
SGCZ8p22PRP Epi [17]4.5e-6EA, AA20 
ST3GAL411q24.2PRP Epi [17]3.0e-6EA, AA51AM [61,63], FD [62], HS [61]
LPAR19q31.3PRP Epi [54]5.8e-3EA110FD, HS [99]
SHH7q36.3PRP ADP [17]4.5e-8EA, AA> 50034MK [100]
MST15110p13PRP ADP [17]6.7e-7EA, AA00 
KCNQ111p15.5-p15.4PRP ADP [17]5.9e-6EA, AA2000 
RAPGEF24q32.1PRP ADP [17]9.1e-7EA, AA196S [101,102]
JAK29p24.1WB ADP [44], PRP ADP [54]6.0e-4, P < 0.05EA> 500499AM, FD, HS, R, P, S
COMMD720q11.21WB ADP [50], PRP ADP [54]1.2e-3, P < 0.05EA00AM, R [50]
GP619q13.42PRP collagen lag time [17]8.4e-14EA, AA23717FD, R [103], P [104]
PTPRD9p24.1PRP collagen lag time [17]1.2e-7EA, AA2003 
HSD17B612q13.3PRP collagen lag time [17]1.1e-6EA, AA60 
FCER1G1q23.3WB coll.(44], PRP coll. lag [17]9.6e-6, 1.6e-5EA120 
UGT1A102q37.1PRP collagen lag time [17]1.2e-5EA, AA1260 
RAP1B12q15PRP collagen lag time [17]1.5e-5EA, AA15052AM [64], R [105,106], S [107], P [108]
MYO5B18q21.1PRP collagen [54]3.0e-5EA340 
WDR6612q24.31MPV [51,55]2.7e-44EA00 
ARHGEF33p21-p13MPV [51,55]5.5e-31EA60R [109]
TMCC21q32.1MPV [51]1.4e-20EA10 
BET1L11p15.5MPV [51]1.3e-14EA100 
DNM31q24.3MPV [51]2.1e-14EA61MK [110], R [50]
EHD32p21MPV [51]3.2e-11EA210R [51]
SIRPA20p13MPV [51]7.7e-11EA1932 
CD22618q22.3MPV [51]1.4e-10EA1130MK [111]
TPM115q22.1MPV [51]1.9e-8EA1581MK [112], R [51,51,112]
AK39p24.1-p24.3PLT [51]8.5e-17EA510 
ATXN212q24PLT [51]2.2e-13EA230R [51]
PTPN1112q24PLT [51]7.7e-12EA200108HS [113], R [51], S [113,114]
THPO3q27.1PLT [34]5.4e-11AS9412MK [115], FD, HS [33,116]
BAK16p21.3PLT [51]3.7e-10EA> 50010 
NRG310q23.1PLT [54]3.6e-5EA380 

Human GWAS of platelet traits

A small number of human GWAS studies have been conducted for platelet traits [17,34,36,51–57]. The major findings from these studies with consistency in two or more population samples are summarised in Tables 3 and 4, along with supporting functional information and citations (e.g., transcriptome, proteome, platelet signalling pathways, animal model results, human disease associations). Table 3 contains loci with evidence for two or more platelet phenotypes. Table 4 contains loci mainly associated with a single platelet phenotype. Specific SNP results are not given in Tables 3 and 4 as these can vary across studies due to differences in SNP coverage, population and linkage disequilibrium. Readers interested in further SNP-specific details underlying results in Tables 3 and 4 are referred to Table S1 (where > 450 significant SNP associations and related SNP identifiers (rsIDs) and mappings are given with reference to the cited GWAS studies).

The first platelet-related GWAS was a moderate density 100 000 SNP scan in FHS which displayed modest findings for PRP aggregation traits with no replication attempted (n ≤ 724, [57]). This study is largely super-ceded by a higher density 550 000 SNP scan in FHS in a larger sample (n ≤ 2753) with imputation to 2.33 million variants which was further combined in a meta-analysis for similar agonists with imputation-based results from the GS study (n ≤ 2075) [17]. This study represents the only large GWAS study of platelet function, and identified regions strongly associated with PRP ADP response (PEAR1, MRVI1, SHH), PRP epinephrine response (ADRA2A, PEAR1, PIK3CG, JMJD1C) and collagen lag time (GP6). Discovery was conducted in two European ancestry samples with replication in one African ancestry sample. Thus, this study represents the first platelet-related GWAS study to compare results across ancestry groups. Notably, there is general consistency in signals and direction of effects among many of the loci in both European and African ancestry samples suggesting that most of the platelet function loci may be shared across ancestries [17].

We identified additional loci with consistent associations in two or more populations with replication evidence in African ancestry samples but which did not reach genome-significance thresholds. Several loci were associated with platelet responses to two or more agonists (Table 3) with the alleles having the same direction of effect in each case, including loci near or containing ADAMTS2, ATP6VOD2, CTCFL, CUBN, FBXL7, FLJ39743, GMDS, HMG1L1, KIAA0802, KLHL31, MIPOL1, MRVI1, NUP210, PEAR1, PCK1, PIP5K1B, PSKH2, RGS18, SETBP1, STMN4, SVIL, THSD4, and WBSCR17. Overall these results suggest many loci contribute to platelet function, with most contributing to a small proportion of the total population variance.

Many of the identified loci contain genes that were not previously studied in relation to platelet biology (see PubMed search column results for each gene in Tables 3 and 4), indicating the value of GWAS in uncovering novel targets for investigation. Other loci from this study are notable for past links to platelet function including the prion protein PRNP, small G-protein related platelet genes including RAP1B, RAPGEF2, RGS18, and PIK3CG, or for their role in animal models for thrombosis, bleeding or healing (MRVI1[58], PIK3CG [59,60], ST3GAL4 [61–63], RAP1B [61–64]), or megakaryocyte development or function (JMJD1C [65]). The GS cohort published results separately from GWAS of pre- and post-aspirin platelet responses in whole blood and PRP in European ancestry and African ancestry samples [36]. These results suggest several potential loci of interest (e.g., the endopeptidase MME) but await attempts at replication.

Larger GWAS sample sizes have been achievable for PLT and MPV due to the easier nature of collecting these measures. The first published study conducted GWAS on 335 152 SNPs in 1606 individuals with MPV measurements, and follow-up replication in up to 8443 individuals, finding and replicating three loci (ARHGEF3, TAOK1, WDR66) that account for approximately 4–5% of the variance in MPV [55]. By further investigation of WDR66 within previously collected microarray results for leukocyte samples, WDR66 transcript levels were inversely correlated with MPV suggesting a functional link. Soranzo et al. followed soon with a MPV GWAS in samples that were used in the replication phase of Meisinger et al., discovering and replicating an additional MPV locus containing PIK3CG, and providing evidence that this locus further influences platelet responses to a collagen mimetic as assessed by annexin V binding [56]. Consultation with microarray data suggested that PIK3CG and another nearby transcript are expressed in megakaryocytes and platelets.

In late 2009 a meta-analysis of MPV and PLT GWAS with up to 4627 discovery samples and imputation to 2.1 million SNPs and replication in up to 9316 samples was published [51]. This was the first study to conduct GWAS of PLT or to report a platelet-related GWAS that employed SNP imputation to increase genomic coverage. The authors replicated 12 loci for MPV, which included the four prior loci, and additionally found and replicated four novel loci for PLT (summarised in Tables 3 and 4). Nine of the 12 MPV loci were also associated with PLT; in each case the MPV-raising alleles were associated with an inverse PLT-lowering effect consistent with the inverse correlation of PLT and MPV.

An independent GWAS on up to 6015 individuals for blood cell traits with imputation to 2.1 million variants found the HBS1L locus was strongly associated with PLT [52]. Platelet count was the only platelet trait analysed and no PLT SNPs were carried forward to the replication stage. The PLT association with HBS1L is consistent with a prior candidate gene study in twins [66], and the strong pleiotropy of this locus including effects on PLT was also observed in prior GWAS [51,53]. Given strong association of HBS1L with several blood counts it may not be highly specific to platelet function. Ferreira et al. [52] identified several additional suggestive PLT loci in univariate and bivariate analyses including ARHGEF3, PHACTR1, and CUBN. The CUBN locus is notable since it was subsequently associated with both ADP and epinephrine PRP responses [17] and recently with albuminuria [67]. The PHACTR1 locus is notable since it is one of the published MI GWAS regions [68].

Kamatani and colleagues [34] conducted GWAS of blood measures including PLT in 14 806 Japanese individuals finding support for prior loci (HBS1L, SH2B3, BAK1, RCL1) and additionally noting strong associations at 3q27.1 (THPO) and 17p13.2 (GP1Bα). The THPO locus encodes thrombopoietin, a known growth factor for MK cells, with the identified GWAS SNP (rs6141) near the translation stop site in the 3′UTR. This region and specific variant were also identified in a prior linkage scan for PLT in Asian Indian families [33]. The peak SNP at GP1Bα (rs6065) encodes a Thr161Met change previously studied as a candidate functional SNP, which was also modestly associated in a GWAS of epinephrine-induced aggregation in PRP [17]. Finally, a recent GWAS measured a range of > 20 platelet traits in a cohort of children (n = 75) with limited replication attempted [54]. Despite a sample size that limited statistical power, they discovered and replicated three new loci (LPAR1 for PRP Epi, MYO5B for PRP collagen, NRG3 for PLT), as well as providing support for several prior loci.

Platelet gene stories new and old

In the dawning ‘omics’ era of platelet studies we remarkably find that some of the strongest candidates of prior work now re-emerge among the strongest signals of ‘omics’ work (e.g., ADRA2A, GP6, CYP2C19). A number of prior candidate genes also re-emerge with weaker, yet supportive evidence in a recent PRP platelet function GWAS (e.g., P2RY12, GP1bα) [17]. With more than 50 loci for platelet traits identified, and more than 50 additional PLT/MPV loci promised soon [69], we have considerably advanced in identifying novel candidates for platelet function, as well as ear-marking others that have some prior evidence but have not been the focus of in depth genetic or functional studies. Some have criticised GWAS as not identifying functional variants, but there is evidence that several of the strongest variants from GWAS may be functional (e.g., GP6, THPO, GP1bα). Synergy between RNA, protein, animal model, candidate gene and GWAS studies is emerging to varying degrees to identify the likely causal genes within loci and in some cases to identify the potential functional variants.

Among the newly highlighted loci are suggested functional receptors (PEAR1, ST3GAL4, PRNP), potential regulators of cytoskeleton dynamics and platelet morphology (DNM3, TPM1, ARHGEF3), MK development (JMJD1C), and intracellular platelet signalling (PIK3CG, MRVI1 (also known as IRAG), RGS18, RAPGEF2, RAP1B, PTPN11). These candidates propel us beyond a focus on cell surface receptors and their ligands and invite new functional platelet and megakaryocyte studies even without respect to specific alleles. It remains to be seen if any of these loci provide new therapeutic targets, or contribute to a deeper understanding of common or rare thrombosis or haemostasis conditions.

Few of the loci discovered to date have genetic alleles with strongly validated roles in human disease. One exception is GP6, which has been independently associated with venous thrombosis in two separate studies [70,71]. Notably, the peak GP6 SNP from the VT GWAS and a PRP collagen lag GWAS are the same nonsynonymous SNP which was previously characterised for functional effects. Another exception, CYP2C19, is associated with outcomes in anti-platelet treatment trials [29,72–74], though with conflicting results in another set of trials, which could be due to demographic variation among the clinical populations [75]. Beyond CYP2C19 there is no other compelling locus associated with response to aspirin or other anti-platelet therapies at this time [13,76], though ABCB1 may play a role [73,74]. Finally, an extremely large linkage disequilibrium block on chromosome 12q24 has been linked to multiple phenotypes including PLT, blood pressure and coronary artery disease but the functional aetiology and causal gene(s) remains to be worked out [51].

Although most known platelet loci have not been convincingly linked to a platelet-related human condition, evidence from animal models does link some of them to thrombosis or bleeding effects (e.g., GP6, ST3GAL4, MRVI1, PIK3CG, ADRA2A, P2RY12, RAP1B). Human disease studies with larger samples and a deeper examination of results with moderate effect sizes may eventually reveal additional platelet-related genes that make modest contributions to disease risk or treatment.

Limitations of current findings and future directions

Current genomic findings are limited both by the constraints of the phenotypes available, the ability to ascertain large populations, and by our current understanding and assessment of human population variation. While variation maps have improved substantially, we are still beginning to survey alleles at the low allele frequency end of the spectrum. Future directions of research include resequencing of platelet loci for additional rare variants that may have larger effect sizes and potential clinical relevance (e.g., in affected families with plateletopathies), functional experiments aimed at uncovering the mechanisms underlying genetic associations, and further study of platelet-related variants in relevant clinical samples which provide adequate statistical power.

While the validation, functional and clinical study of individual variants remains important, locus level views of overlapping trait scans suggests extensive pleiotropy for some platelet-related regions (e.g., ABO, SH2B3, HBS1L). With a growing number of published GWAS and examples of pleiotropy this is an active area of future development since pleiotropic signals may provide important clues to the functions of genes in the associated loci [31,77]. As more data become available data mining, pathway analysis and other bioinformatics approaches may also prove useful to characterising important loci for platelet function. We previously assembled a catalogue of results from 118 published GWAS demonstrating the potential to discover novel suggestive patterns [31].

Using a similar, updated catalogue of results from > 700 GWAS (unpublished results) Table S1 was created with > 450 significant associations related to platelet traits from all available GWAS results. This provides a quick overview of loci consistent across studies (e.g., HBS1L) but could also suggest loci to prioritise for further study. For example, RAPGEF2 is seen to be associated with PRP response to ADP in [17] and clopidogrel response assessed by ADP in [29], though this locus has not yet been the focus of replication studies for clopidogrel response. Another potential example is observed when PRP aggregation GWAS, venous thrombosis GWAS and platelet gene expression results are considered. Three distinct loci that encode subunits of a V-type proton ATPase were highlighted: for PRP aggregation associations (ATP6V0D2, 8q21.2-q21.3, [17]), gene expression (ATP6V1F, 7q23.1, [50]), and in a GWAS follow-up study on venous thrombosis (ATP6V1B2, 8p21.2, [78]). These loci were not among the strongest in any of these studies individually. Notably, they are all subunits of a single transporter that has received little study. Evidence indicates this vacuolar ATPase is involved in bone homeostasis [79] and thus it may influence megakaryopoeisis or other platelet-related functions. Additional examples of such hypotheses are likely to emerge as further ‘omics’ studies are published for platelet and haemostasis related traits.

It is important to note that the loci identified thus far which contain common variation contributing to variability in platelet traits individually account for a small proportion of the population variance. Large, well-designed prospective trials of variants are needed to determine their potential relevance in clinical practice. While CYP2C19 is of potential clinical significance, a position statement does not support routine genotyping or platelet function testing in practice [80].Thus, at this time none of the specific markers highlighted here should be considered to have clear and direct utility in clinical practice.


A. Johnson drafted the manuscript and is solely responsible for its content.


A. Johnson was supported by a National Institutes of Health Intramural Research Training Fellowship.

Disclosure of Conflict of Interest

The author states that he has no conflict of interest.