Genotyping and phenotyping of platelet function disorders


  • S. P. Watson,

    Corresponding author
    1. Centre for Cardiovascular Sciences, Institute of Biomedical Research, College of Medical and Dental Sciences, University of Birmingham, Birmingham, UK
    • Correspondence: Steve Watson, Centre for Cardiovascular Sciences, Institute of Biomedical Research, College of Medical and Dental Sciences, University of Birmingham, Birmingham B15 2TT, UK.

      Tel.: +44 121 445 1654; fax: +44 121 415 8817.


    Search for more papers by this author
  • G. C. Lowe,

    1. Centre for Cardiovascular Sciences, Institute of Biomedical Research, College of Medical and Dental Sciences, University of Birmingham, Birmingham, UK
    Search for more papers by this author
  • M. Lordkipanidzé,

    1. Centre for Cardiovascular Sciences, Institute of Biomedical Research, College of Medical and Dental Sciences, University of Birmingham, Birmingham, UK
    Search for more papers by this author
  • N. V. Morgan,

    1. Centre for Cardiovascular Sciences, Institute of Biomedical Research, College of Medical and Dental Sciences, University of Birmingham, Birmingham, UK
    Search for more papers by this author
  • The GAPP consortium


This article is corrected by:

  1. Errata: Corrigendum Volume 11, Issue 9, 1790, Article first published online: 12 September 2013


The majority of patients with platelet function disorders (PFDs) have normal platelet counts and mild day-to-day bleeding symptoms, but are at risk of major hemorrhage at times of trauma, surgery, or childbirth. This group is challenging to investigate, because the assays are often time-intensive and labour-intensive, and interpretation is difficult, especially in patients with mild disorders. In addition, interuser variability in performance of the assays, including the currently accepted gold standard, light transmission aggregometry, makes the results difficult to compare between laboratories. Furthermore, a similar pattern of mucocutaneous bleeding is seen in disorders in other components of the hemostatic pathway, including type 1 von Willebrand disease (VWD). We have undertaken an extensive investigation of patients with clinically diagnosed excessive bleeding, using a genotyping and platelet phenotyping approach based on lumi-aggregometry, and other specialist tests of platelet function, in combination with Sanger and next-generation sequencing (NGS). We found a functional defect in ~ 60% of patients, the majority being associated with feedback pathways of platelet activation. Function-disrupting mutations were identified in known and novel genes, and coinheritance with other genetic disorders of hemostasis, including type 1 VWD, was shown. A significant number of mutations are heterozygous and unlikely to cause extensive bleeding in isolation, consistent with incomplete penetrance of inheritance of bleeding disorders and a multifactorial etiology for excessive bleeding in many patients. Mucocutaneous bleeding is a complex trait, and this has important implications for NGS in the assessment of a PFD.


Platelet function disorders (PFDs) make up a significant proportion of bleeding diatheses, but remain poorly understood and are often difficult to diagnose, especially in mild cases. Patients with severe platelet disorders, such as Glanzmann's thrombasthenia and Bernard–Soulier syndrome (BSS), caused by mutations in integrin αIIbβ3 (glycoprotein [GP]IIbIIIa) and the GPIb–IX–V complex, respectively, are usually identified early in life as a consequence of extensive bruising and other signs of mucocutaneous bleeding [1-3]. Furthermore, the diagnosis is readily made by absent aggregation and ristocetin-induced agglutination, respectively, with confirmation by flow cytometry for platelet surface GPs. However, the majority of patients with platelet disorders have a less severe bleeding pattern and a normal platelet count [1-3]. This group is challenging to diagnose, as excessive bleeding, which can occasionally be life-threatening, is only seen in response to an appropriate challenge, such as surgery, injury, menstruation, and childbirth. As a consequence, these conditions are often not recognized until the teenage years or adulthood. Nevertheless, repeated minor bleeds throughout life (e.g. intermittent epistaxis, extensive bruising, and menorrhagia) can have a significant effect on quality of life and activities of daily living.

A mild platelet disorder is first suspected from the pattern of bleeding, with disproportionate bruising for injury severity, bruising not associated with trauma, and excessive bleeding from mucocutaneous membranes, such as epistaxis and menorrhagia [1-3]. Whereas this pattern is usually distinguishable from that of coagulation disorders such as hemophilia, the symptoms are similar to those of von Willebrand disease (VWD) and collagen or connective tissue disorders [1-3]. A definitive diagnosis of a PFD is usually made by the exclusion of VWD and other hemostatic disorders, such as inherited coagulation factor defects and connective tissue disorders, and subsequent laboratory demonstration of a platelet defect. Some patients may have more than one defect contributing to the overall clinical phenotype.

A proportion of platelet disorders are associated with syndromes and systemic disease. One example is Wiskott–Aldrich syndrome (WAS), which is caused by mutations in the cytoskeletal protein Wasp. Affected individuals with this X-linked recessive disorder exhibit eczema, frequent episodes of infection, and a mild bleeding diathesis [4]. Hermansky–Pudlak syndrome (HPS) is a group of 15 distinct gene disorders in mouse, nine of which have been found in humans [5]. The genes code for proteins in a series of specialist organelles, which include melanosomes and platelet dense granules. Patients with HPS have loss of pigmentation, notably in the hair and eyes, and a mild bleeding diathesis. The mild nature of the bleeding and other causes of pigment loss mean that HPS often remains undiagnosed for many years, with an extreme example being the identification of the second patient with a mutation in dysbindin (HPS7) at the age of 77 years, despite an impressive bleeding history [6]. In contrast, HPS1 is the most common recessive disorder in Puerto Rico, with a frequency in the order of 1 : 2000 [5]. Arthrogryposis, renal dysfunction and cholestasis is a syndrome that is associated with a high mortality rate in the first year after birth. It is characterized by multiple defects, including a bleeding diathesis caused by an absence of platelet α-granules [7]. This very rare recessive disorder is caused by mutations in the trafficking proteins VPS33b and VIPAR.

The association of platelet disorders with syndromes and systemic disease has facilitated the identification of platelet defects by several groups. For example, van Geet and Freson in Leuven have reported defects in platelet responses in individuals affected by a variety of neurodevelopmental disorders [8], including autism [9]. They have also identified several other rare conditions associated with platelet dysfunction, such as Albright's hereditary osteodystrophy, which is characterized by short stature, shortened fourth and fifth fingers, and cognitive disability. This disease is inherited in an autosomal dominant manner, and is caused by epigenetic genomic imprinting leading to reduced expression of the G-protein G subunit [10]. Changes in platelet morphology, ultrastructure or function have been reported in several other disease states, including cancer, liver disease, asthma, HIV disease, and diabetes, although, in many of these cases, it is unclear whether the change is secondary to the disorder or a part of the underlying pathogenesis. Thus, the study of platelet morphology and function offers a powerful approach for the investigation of complex systemic diseases, especially as platelets are easily accessible.

Mild PFDs with normal platelet counts have a frequency of 1 : 10 000, and are thus considered to be rare diseases [1]. For example, in the UK, there are ~ 1500 patients registered at Haemophilia Care Centres with PFDs in a population of 60 million ( Causative mutations in patients with PFDs have been identified in a very small proportion of patients and gene families, as summarized in Table 1. For example, only two patients (both compound heterozygotes) with function-disrupting mutations in the stimulatory receptor for collagen, GPVI, have been described [11, 12], and only three with function-disrupting mutations in the Gq-coupled thromboxane receptor, all of whom are heterozygous for the mutation [13-15]. The number of patients with mutations in the Gi-coupled P2Y12 ADP receptor is now into double figures [16, 17], with several also being heterozygous. There are no patients with mutations in the genes encoding the two protease-activated receptors (PARs) PAR1 and PAR4, although it is notable that, in mice, deletion of either gene causes embryonic lethality. Many of the patients with excessive bleeding and heterozygous mutations in the thromboxane and P2Y12 receptors are likely to have a second abnormality, as some of their relatives who also carry the heterozygous change do not have a history of bleeding. Furthermore, inhibitors of these two pathways are used widely as prophylaxis and treatment for arterial thrombosis, but only induce excessive bleeding in a minority of treated patients [18]. It should also be borne in mind that the second abnormality may not necessarily be a platelet defect, as illustrated by the presence of heterozygous P2Y12 mutations in patients with type 1 VWD [19, 20].

Table 1. Brief reference guide on inherited platelet disorders
Platelet abnormalityDiseaseInheritanceDefective geneLaboratory and other findings
  1. GP, glycoprotein; MPV, mean platelet volume; VWF, von Willebrand factor. This table summarizes the genetic basis of known platelet disorders. All of the listed disorders are rare, and, in some cases, < 10 patients have been reported. The table is not intended to be fully comprehensive – the genetic causes of other platelet disorders have been described but are not included, owing to the very low number of reported patients.

Platelet adhesionPlatelet-type von Willebrand diseaseAutosomal dominantGP1BA (17p13.2)


Diminished or absent large VWF multimers

Enhanced ristocetin agglutination (occurs at low concentrations), which is corrected when donor platelets and patient plasma are used in mixing studies

Bernard–Soulier syndromeAutosomal recessive

GP9 (3q21.3)

GP1BA (17p13.2)

GP1BB (22q11.21)

Thrombocytopenia with increased MPV

Anomalies in components of the GPIb–V–IX complex

Platelet aggregation: absent ristocetin-induced agglutination

Platelet receptor defectsP2Y12 ADP receptorAutosomal recessive (mild phenotype in heterozygotes)P2Y12 (3q25.1)

Platelet count normal

Platelet aggregation: normal P2Y1 receptor-driven responses; shape change and transient aggregation

GPVI collagen receptorAutosomal recessiveGP6 (19q13.42)

Platelet count normal

Platelet aggregation: absent with GPVI-specific agonists, e.g. convulxin and collagen-related peptide; and marked reduction with collagen

Thromboxane A2 receptorAutosomal recessive (mild phenotype in heterozygotes)TBXA2R (19p13.3)

Platelet count normal

Platelet aggregation in response to arachidonic acid and U44619 reduced in heterozygotes to and presumed to be absent in homozygotes

GPIIbIIIa (αIIbβ3)

(Glanzmann's thrombasthenia)

Autosomal recessive

ITGA2B (17q21.32)

ITGB3 (17q21.32)

Normal platelet count, size, and morphology

Presents with severe bleeding symptoms in early life

Absent platelet aggregation with all agonists; agglutination in response to ristocetin is normal

Flow cytometry with CD41 and CD61 antibodies may show reduced levels of either GPIIb or GPIIIa

Platelet secretionHermansky–Pudlak syndromeAutosomal recessive

HPS1 (10q24.2)

HPS2/AP3B1 (5q14.1)

HPS3 (3q24), HPS4 (22q12.1)

HPS5 (11p14)

HPS6 (10q24.32)

HPS7/dysbindin (6p22.3)

HPS8 (19q13.32)

HPS9 (15q21.1)

Platelet count normal

Skin and hair hypopigmentation

Reduced/absent δ-granules on electron microscopy

Lumiaggregometry: reduced/absent ATP release

Chediak–Higashi syndromeAutosomal recessiveCHS1/LYST (1q42)

Platelet count normal

Skin and hair hypopigmentation


Giant inclusions in granulocytes and their precursors

Reduced or irregular α-granules

Lumiaggregometry: reduced/absent ATP release

Gray platelet syndromeAutosomal recessiveNBEAL2 (3p21.31)


Increased MPV with platelet anisocytosis

Platelets grey in colour on blood film

Absent α-granules

X-linked dyserythropoietic anemia and thrombocytopeniaX-linkedGATA1 (Xp11.23)

Thrombocytopenia with increased MPV

Reduced α-granules


 Arthrogryposis, renal dysfunction and cholestasis syndromeAutosomal recessive

VPS33B (15q261)

VIPAS39 (14q24.3)

Thrombocytopenia with increased MPV

Severe multisystem syndrome, leading to fatal complications very early in life

Absent α-granules

 Paris–Trousseau/Jacobsen syndromeAutosomal dominantFLI1 (11q24.1-24.3)

Thrombocytopenia with increased MPV

Developmental delay and facial abnormalities

 Quebec platelet disorderAutosomal dominantPLAU (10q22.2)

Platelet count at low end of normal range

α-Granule protein degradation

Increased urokinase-type plasminogen activator storage in platelets

Platelet cytoskeletonMYH9-related disorders (May–Hegglin anomaly, also known as Sebastian/Fechtner/Epstein syndrome)Autosomal dominantMYH9 (22q12-13)

Thrombocytopenia with increased MPV

Döhle-like inclusions in neutrophils

Nephritis and hearing loss in some forms

Wiskott–Aldrich syndrome/X-linked thrombocytopeniaX-linkedWAS (Xp11.23)

Thrombocytopenia with small platelets

Immunodeficiency and eczema (in Wiskott–Aldrich syndrome)

Filamin A disorders (periventricular nodular heterotopia/otopalatodigital syndrome)X-linkedFLNa (Xq28)

Thrombocytopenia with raised MPV and abnormal platelet morphology

Abnormal distribution of platelet filamin on confocal microscopy

Platelet procoagulation activityScott syndromeAutosomal recessiveTMEM16F (12q12-13.11)

Platelet count normal

Anomalies in flippases translocating negatively charged phospholipids on the plasma membranes

Impaired annexin A5 binding with flow cytometry

Other thrombocytopeniasCongenital amegakaryocytic thrombocytopeniaAutosomal recessiveMPL (1p34)

Severe thrombocytopenia


Absent megakaryocytes in bone marrow

Increased plasma thrombopoietin levels

Thrombocytopenia with absent radius syndromeAutosomal recessiveRBM8A (1q21.1)

Severe thrombocytopenia

Normal platelet morphology

Shortened/absent radii in forearm

THC2Autosomal dominantMASTL, ACBD5, ANKRD26 (all 10p12.1)

Mild to moderate thrombocytopenia with normal MPV

Platelets deficient in GPIa and α-granules

Platelet aggregation normal

Possible dysmegakaryopoiesis

Familial platelet disorder with predisposition to acute myelogenous leukemiaAutosomal dominantRUNX1 (21q22.12)

Mild thrombocytopenia, with possible raised MPV

Abnormal aggregation in response to multiple agonists

There is considerable redundancy in the receptors and signaling pathways underlying platelet activation, and bleeding disorders are therefore likely to arise from additive mutations in more than one platelet pathway or in other parts of the hemostatic system, such as coagulation factors, vascular endothelium, and fibrinolysis. Some of these areas are also not well understood (e.g. vascular causes of bleeding [21, 22]), and this makes it more difficult to discriminate between the different origins of the bleeding phenotype. The complexity of platelet signaling and the need to perform assays on fresh samples present further obstacles in investigating platelet disorders, and explain why research activity in this area is relatively underdeveloped, such that the frequency of platelet disorders is probably heavily underestimated. For example, it has been estimated that just over 50% of women with menorrhagia have an undiagnosed PFD [23], although this estimate is dependent on the platelet function test used and the criteria used to identify the defect. For example, using lumi-aggregometry with a panel of agonists and our own reference ranges, we found the figure to be closer to 25% in a relatively small sample of women with menorrhagia but with no structural uterine defects or coagulophathy [24]. However, the proportion reached 60% when the women were recruited from hemophilia centers, where concomitant mild bleeding symptoms increased the likelihood of a platelet defect [24].

The major route of identification of PFDs has been through laboratory investigation of patients with suspected platelet bleeding disorders, with a well-established platelet function assay, light transmission aggregometry (LTA), which requires specific expertise [25, 26]. More recently, additional technologies have been used, including flow cytometry for the measurement of platelet glycoproteins and P-selectin, a marker of α-granule secretion [26]. Certain groups have also made particular use of electron microscopy to identify defects in platelet morphology [27]. There have been several recent initiatives from individual investigators [28] and national/international organizations to standardize experimental conditions in the use of LTA, including reference ranges for agonists [25, 26]. The ISTH SSC on Platelet Physiology is leading in the standardization of methodology, and is now directing its efforts towards the evaluation of patients with PFDs.

A small number of research groups have used Sanger sequencing methodology to identify the genetic basis of PFDs. This includes the identification of patients with function-disrupting mutations in P2Y12 [16, 17] and filamin [29]. The very small number of patients with identified genetic defects, however, emphasizes the need for a more coherent and powerful strategy that takes advantage of next-generation sequencing (NGS) methodology to identify genetic mutations, and biological assays to establish the functional consequences of these mutations. In this context, we are undertaking an extensive genotyping and platelet phenotyping (GAPP) study (ISRCTN 77951167) to investigate PFDs throughout the UK [30]. In this article, we summarize the current status of the GAPP study, and compare this with other approaches for the identification and genotyping of platelet disorders.

Phenotyping of platelets

The first suspicion of a PFD arises from the clinical history, but a definitive diagnosis requires demonstration of a laboratory defect in platelet function [1-3]. Historically, the most widely used test was the bleeding time, but this is not only invasive, but is poorly reproducible and no longer recommended [26, 31]. The modern day ‘gold standard’ test for monitoring platelet activation is LTA, which was first described by Born [32]. The basic principle has not changed in over 50 years, even though modern-day aggregometers are much smaller and easier to use [33]. The test uses platelet-rich plasma (PRP) or washed platelets, and monitors the change in light transmission upon agonist addition. The response can consist of several components, including an initial increase in OD brought about by a change in platelet shape, followed by a primary, biphasic, sustained or reversible increase in light transmission, which is normally mediated by αIIbβ3-dependent aggregation [34]. The nature of the response is dependent on the agonist, agonist concentration, and the feedback action of ADP and thromboxane A2 (TxA2). LTA, however, does require an experienced operator, and does not pick up many aspects of platelet activation, including procoagulant activity and platelet spreading.

A variation of LTA is the addition of luciferin–luciferase reagent to simultaneously monitor dense granule secretion [35]. Concern has been raised that addition of this reagent may alter the pattern of platelet aggregation in certain groups of patients [36], although, in a systematic investigation of platelet activation by adrenaline in 100 participants within the GAPP study, we have not seen a significant change in response (Lordkipanidzé et al., submitted). Diagnosis of HPS and other dense granule secretory disorders cannot be achieved on the basis of aggregation alone, as all of the commonly used platelet agonists, including ADP and adrenaline, can elicit full aggregation in the absence of secretion [26, 37]. It should also be borne in mind that a defect in ATP secretion could reflect either abnormal granule formation or defective trafficking of granules. Additional tests, such as electron microscopy, serotonin uptake, and total ATP content, are therefore required to distinguish between these two possibilities.

A wide range of other platelet function tests are also available, some of which focus on responses to specific agonists and some of which measure a global output of platelet activation. The variety of available tests have been described in several excellent reviews [26, 33, 38-42]. These will not be discussed in further detail, other than to highlight the fact that each places a different emphasis on the role of primary and secondary aggregation and feedback pathways, which explains the subtle distinctions in their results.

A major challenge in the use of lumi-aggregometry in the investigation of PFDs is the wide variation observed in responses of healthy volunteers [34, 43]. Thus, the common practice of comparing results in the patient with those in a healthy volunteer analyzed at the same time can often be misleading in the absence of locally determined normal ranges [26]. In the GAPP study, we have now compiled data on responses to a minimum of three concentrations of nine platelet agonists from ~ 100 healthy volunteers, and have established normal ranges with cut-offs at the 5th and 95th percentiles for a range of parameters, including the delay to onset of response, the maximal levels of aggregation and ATP secretion, and the presence of biphasic and transient aggregation (Lowe and Lordkipanidzé, unpublished). We have found that certain concentrations of platelet agonists are highly discriminatory for PFDs, including 10 μm ADP, 1 mm arachidonic acid, and 3 μg mL−1 collagen, which ordinarily give a sustained aggregation response in platelets from drug-free healthy volunteers. Other factors that can influence the response include the platelet count, although we have shown that this is not significant in PRP over the range 1.5–6.0 × 10mL−1 [34]. This is in line with recent literature suggesting the use of native PRP to be superior to that of PRP diluted to a standard platelet count [44, 45]. However, the level of ATP secretion is platelet count-dependent, making normalization to platelet count necessary [46]. Below a platelet count of 1.5 × 10mL−1, a flow cytometry assay for assessment of platelet function can be successfully used to assess whether a qualitative defect is present in addition to the thrombocytopenia [47].

Despite the availability of locally determined normal ranges, diagnosing a defect in platelet function often remains a challenge, because of overlap with the response of healthy volunteers. There is also the further challenge of interpreting the likely mechanism of the defect, because of the feedback effects of ADP and TxA2, and the fact that many PFDs are associated with partial rather than complete loss of a particular receptor or signaling pathway [46]. Assessment of the overall pattern of response by an experienced investigator is therefore essential to determine whether there is a defect in platelet activation [46].

Following the publication of normal ranges of platelet responses to nine stimulatory agonists [34], we established an extensive multicenter study of patients with clinically diagnosed PFDs registered at over 20 Haemophilia Comprehensive Care Centres in the UK [30]. This study is backed by the National Institute of Health Research as making a difference to the management of patients (UKCRN ID 9858). For the purpose of this study, a diagnosis of excessive clinical bleeding with the possibility of a PFD was made by the referring consultant on the basis of the patient's bleeding history, normal results in coagulation assays, and measurement of coagulation factor levels. In some cases, the referring center had evidence for a defect in platelet function through tests such as LTA or the PFA-100. We included only patients with normal and platelet counts in this group; patients with thrombocytopenia were recruited to a distinct arm of the GAPP study. We excluded patients with recognized platelet disorders, including Glanzmann's thrombasthenia, BSS, May–Hegglin anomaly, and WAS. Each patient with a normal platelet count was subjected to platelet phenotyping by lumi-aggregometry, with the nine platelet agonists being tested alongside PRP from a healthy volunteer. In some cases, we used inhibitory agonists and additional tests, including flow cytometry and measurement of cAMP levels, to gain further information. Participants with thrombocytopenia were functionally analyzed with a flow cytometry assay, as lumi-aggregometry is unreliable below a platelet count of 1.5 × 108 mL−1 [34].

We analyzed samples from approximately 350 patients and 100 healthy volunteers, and reported our findings based on the first 111 patients [46]. In this study, we identified a defect in platelet function in 58% of patients, and subdivided these according to the pattern of response. Over 70% of patients with a defective response showed an impairment in dense granule secretion, Gi receptor signaling, or arachidonic acid metabolism (cyclooxygenase pathway). In addition, several patients were shown to have function-disrupting mutations in the P2Y12 ADP receptor and thromboxane receptor. The remaining patients had a complex phenotype that included loss of response to collagen, Gq-receptor agonists, or multiple agonists. The distribution of platelet defects is shown in Fig. 1.

Figure 1.

Pie chart showing the distribution of classification of platelet function disorders (PFDs) in patients studied within the genotyping and platelet phenotyping (GAPP) study. Participants with a suspected PFDs were referred from Comprehensive Haemophilia Care Centres throughout the UK, and evaluated by lumiaggregometry as part of the GAPP study [46]. Participants whose platelets were observed to have a functional defect were subdivided on the basis of the defect, as shown in the pie chart. Sixty-five of the 111 (58%) participants in this analysis showed a defect on lumiaggregometry (Adapted from [46]).

An important finding from this work is that the vast majority of patients present with impairments in activation by the major feedback pathways underlying platelet activation, namely the two feedback agonists, ADP and TxA2, and in dense granule secretion (which releases ADP). Interestingly, in many of these cases, the effect is partial, with the defect in aggregation being overcome at high agonist concentrations, and with secretion being impaired but not abolished in all patients diagnosed with a secretion disorder other than those with HPS [46].

The absence of a defect in aggregation and/or secretion in just over 40% of the patients studied may be attributable to several factors, including:

  1. limited sensitivity in lumi-aggregometry testing;
  2. overlap of response with that of controls in mild cases – although the establishment of reference ranges has helped in the classification of whether a defect is present;
  3. the defect only being revealed in adhesion assays, notably under flow, or in blood;
  4. excessive bleeding being caused by a defect in another part of the hemostatic pathway.

In view of these considerations, on a subset of patients we performed impedance aggregometry (Multiplate, Roche Products Ltd, Burgess Hill, UK), platelet aggregation on collagen at arterial shear (1000 s−1), clot retraction assays, and assessment of inhibitory platelet pathways in lumi-aggregometry with prostacyclin and sodium nitroprusside. These assays, however, did not increase the number of patients identified as having a PFD over that achieved with lumi-aggregometry (unpublished). Although these studies do not rule out the possibility that defects in these assays could be detected in patients with a normal pattern of lumi-aggregation, they demonstrate that this is not a likely occurrence, making it difficult to justify these additional assays on the grounds of sample volume, cost, and time.

On the basis of work carried out in the first 111 patients tested within the GAPP study, we have developed a streamlined panel of agonists to reduce the sample volume and time required for platelet function testing [46]. A retrospective analysis has shown that this rationalized panel of agonists has high reliability in diagnosing a platelet function defect by lumi-aggregometry [46]. Nevertheless, the relatively large blood sample volume and considerable amount of time needed to perform the studies, alongside the absence of a defect in > 40% of patients, has encouraged us to investigate other assays of platelet function with increased throughput.

With throughput in mind, we have investigated the use of a 96-well plate aggregation assay (Optimul), using plates with lyophilized platelet agonists, in collaboration with T. Warner in London [48, 49]. This assay requires a total PRP sample volume of < 3 mL, measures changes in light absorbance in response to various agonists in a conventional plate reader, and can be performed and analyzed within 30 min following the preparation of PRP. It can also be used for the measurement of TxA2 and other secreted agents by analysis of the supernatant. However, this is an endpoint assay, and thus does not provide kinetic information, and nor does it measure platelet shape change. In general, we have found good agreement for identification of a platelet defect between the Optimul assay and lumi-aggregometry [50]. However, for reasons that are unclear, the majority of platelet disorders are associated with a reduction in aggregation in response to arachidonic acid on Optimul, and it thus does not allow diagnosis of the nature of the platelet defect [50]. This result serves to emphasize that, despite similarities in the way in which platelet aggregation is measured between conventional LTA and Optimul, each platelet function test has a different sensitivity to the pathways that underlie platelet activation.

We have further investigated, in collaboration with N. Dovlatova and S. Fox in Nottingham, a remote screening tool that relies on flow cytometry analysis of the platelet activation markers P-selectin (α-granules) and CD63 (dense granules) in blood [51]. The addition of blood to four tubes containing activating agents followed by fixation with a patented fixative solution developed by Platelet Solutions (Nottingham, UK) allows subsequent platelet function analysis at a remote site. Preliminary data indicate good agreement between this remote assay and lumi-aggregometry in diagnosing a platelet defect, indicating that this test offers a straightforward and easy-to-conduct approach for determining which patients may benefit from more extensive testing of platelet function. The utility of such an assay, in association with the ISTH bleeding assessment tool [52], could be as a screening tool to identify patients who are unlikely to have a major platelet function defect, especially in cases with a relatively mild clinical history of excessive bleeding.

Genotyping of PFDs

The most likely explanation for a defect in platelet function is a mutation in one or more of the genes involved in megakaryocyte development, platelet formation, or platelet function. In most cases, the identification of the causative mutations is extremely challenging, because of the overwhelming number of candidate genes, and increasing evidence that, in most cases, platelet disorders are complex traits that are influenced by a combination of inherited and acquired defects, such as ingestion of an antiplatelet agent. Thus, a disorder may only be revealed following exposure to a suitable challenge, such as surgery or childbirth, and it may not be possible to ascribe the disorder to a single, causative mutation. The incomplete penetrance of bleeding between family members further hampers genetic studies, although the genetic analysis of family members without a history of excessive bleeding can help to rule out candidate genes.

Owing to these considerations, we have used phenotyping to direct the search for candidate mutations. The nature of the search for mutations is governed by a number of factors, including the presence of parents who are blood relatives of one another (referred to as consanguinuity when they are cousins, which is the most commonly encountered scenario), evidence of autosomal dominant inheritance patterns, and the nature of the platelet function defect. At present, the greatest success has been achieved in identifying mutations in patients with a clear pattern of inheritance, such as thrombocytopenia, and conditions associated with other phenotypic changes, as exemplified by HPS.

Sanger sequencing of candidate mutations

The selective loss of platelet activation by a single agonist is strong evidence for a defect at the level of its cell surface receptor. In principle, this is a relatively straightforward scenario, but in reality it is complicated by the feedback action of ADP and TxA2, and, in many cases, by the partial nature of the defect as a result of heterozygous expression of the mutation or a function-impairing rather than function-abrogating mutation. These points are illustrated by the following examples.

  1. A homozygous mutation in the ADP P2Y12 receptor was identified in the GAPP study in a 30-year-old female whose parents were cousins [46]. She had required blood transfusions during two pregnancies. Her brother had suffered from prolonged bleeding after a minor injury. The patient's platelets underwent shape change and transient aggregation in response to a high concentration of ADP, and showed normal biphasic aggregation in response to adrenaline. This response is representative of that observed in the presence of a P2Y12 receptor antagonist. Sanger sequencing of the P2Y12 receptor identified a base pair shift early in the coding sequence that prevented receptor expression [46].
  2. A heterozygous function-disrupting point mutation in the thromboxane receptor was identified in the GAPP study in a 14-year-old boy who required frequent nasal packing to treat severe epistaxis [15]. The patient's platelets showed a marked reduction in aggregation in response to intermediate concentrations of arachidonic acid and the stable thromboxane mimetic U46619, with recovery at higher concentrations. The marked impairment in response to U46619, but not in response to other Gq-linked receptor agonists, indicated a defect at the level of the thromboxane receptor, as platelet activation in response to U46619 is unaffected by cyclooxygenase inhibition. A D304N point mutation was identified in transmembrane 7, and was shown to prevent ligand binding in a stably transfected cell line [15].

In the second example, the heterozygous mutation is unlikely to be the cause of the increase in bleeding, as the father of the index case, who is also heterozygous for the mutation, does not have a history of excessive bleeding. On the other hand, his mother, who does not have the mutation, has a mild bleeding history. Thus, it seems likely that the index case has an as-yet unidentified second mutation inherited from his mother, and that this, in combination with the thromboxane receptor mutation, gives rise to excessive bleeding.


The development of NGS technology has revolutionized the sequencing of large amounts of genomic and transcriptomic material [53]. Thus, we are already in an era when it is financially viable to sequence whole exomes, and even whole genomes. The sheer volume of genetic information, however, means that a rate-limiting step will be the processing and interpretation of the data and the comparison of DNA from patients with that of appropriate datasets to identify candidate function-disrupting mutations [54]. Potential limitations also exist both within the reference databases themselves and in the sequencing methodology [55]. Sanger sequencing of the index case and, where possible, additional family members will strengthen the case for the association with the bleeding by examination of phenotype–genotype segregation. In some cases, however, several candidate mutations may remain, or the DNA of other family members may not be available, necessitating further investigations to identify the causative gene. In many cases, the functional significance of the mutation may be uncertain, or the role of the implicated protein in platelet function may be unknown. Thus, it may be necessary to express the protein in a suitable cell line model and evaluate its function. This step is particularly important in cases of uncertainty, as even healthy volunteers are heterozygous for almost 100 function-disrupting mutations that have little or no associated phenotype [56]. The challenge is therefore to distinguish the causative mutation(s) that contribute to the platelet phenotype and clinical bleeding diathesis from other mutations that do not alter platelet function.

NGS has significantly improved the rate of discovery of causative mutations in patients whose parents are blood relatives of one another through autozygosity mapping. The greater challenge, however, is to identify causative mutations in unrelated patients, and here there are few success stories.

NGS and autozygosity mapping

Autozygosity mapping has been used for many years in the identification of autosomal recessive mutations in families with consanguinity, as illustrated by the identification of the eighth form of HPS in a Birmingham family of Pakistani origin [57]. Autozygosity mapping is based on the premise that affected members within each family share autozygosity over a region of DNA encompassing the disease-causing mutation. In conditions where there is more than one candidate gene, such as HPS, the mapping is facilitated by the use of microsatellite markers [6]. The use of NGS techniques greatly reduces the time needed to perform the mapping studies, as illustrated by the identification of the recessive mutation in the neurobeachin-like 2 gene (NBEAL2) as being causative for gray platelet syndrome. Three independent studies using whole exome sequencing, RNA sequencing and classic positional cloning discovered the causative gene [58-60].

The identification of ANKRD18A as the causative mutation of a severe form of thrombocytopenia and platelet dysfunction in a consanguineous family illustrates this powerful approach [47] as shown in Fig. 2. In consanguineous families, the average number of nonsense single-nucleotide polymorphisms (SNPs) per individual is 45, whereas the average number of missense SNPs is almost 8000. However, confining SNP analysis to regions of autozygosity quickly excludes the majority of the novel variants, increasing the speed of identifying causative genes.

Figure 2.

Whole exome sequencing data from a consanguineous family with severe thrombocytopenia. The number of candidate sequence variants can be reduced greatly in a cousinship, as shown.

NGS and syndromic PFDs

NGS is very powerful for the identification of causative genes in conditions with a clear phenotype. This is illustrated by the identification of the site of a second causative mutation in thrombocytopenia with absent radii syndrome, which is characterized by abnormal forearm development and a mild bleeding tendency. A heterozygous microdeletion on the long arm of chromosome 1 in the gene RBM8A was identified in affected individuals by the use of conventional mapping techniques [61], but this alone could not be the cause of the syndrome, as family members with the same heterozygous mutation did not have the disorder. Recently, NGS was used to identify a second mutation in the regulatory region of RBM8A in individuals with this disease. This mutation in the second allele was found in 53 of 55 cases, and was associated with reduced expression of the protein [62].

NGS and autosomal dominance

In non-consanguineous families with a strong family history of bleeding over multiple generations, autosomal dominant inheritance should be suspected. In these cases, the power of NGS can be increased by the sequencing of multiple affected and unaffected individuals, and searching for novel variants that cosegregate with the phenotype of excessive bleeding. The number of causative genes that are known to have a dominant pattern of inheritance is limited to a few examples, including the genes encoding the transcription factors Runx1 and Fli1, which cause mild thrombocytopenia and impairment of platelet activation in response to multiple agonists [63, 64]. There are presently no published examples of the use of NGS in the identification of autosomal dominant mutations for PFDs, although we have had several example of success with this approach in the GAPP study, including the identification of novel variants of Fli1 (Stockley and Daly, personal communication).

NGS and platelet phenotyping

The subdivision of PFDs according to their phenotype provides an important guide to the identity of the underlying gene defect. However, many platelet traits are complex, and this approach will therefore only identify gene mutations associated with the predominant phenotype. Furthermore, the phenotyping studies can only identify potential mutations in combinations of genes that lie on the same function pathway, with the exception of selective loss of receptor responses, as illustrated for the P2Y12 and thromboxane receptors. Nevertheless, the identification of a clear phenotype narrows the search to genes involved in a particular process. The power of this approach can be further increased by investigating pooled, unrelated groups of patients with the same phenotype, on the assumption that several of the patients will have mutations in the same gene.

As an illustration of this, we generated an array of the coding domains of > 200 platelet genes that we consider to be strong candidates for defective platelet activation [65]. The list of genes includes those encoding receptors and signaling, cytoskeletal, trafficking and secretory proteins. We then interrogated the arrays as guided by the phenotype of the affected individual. As proof-of-principle, we identified a novel mutation in the gene encoding HPS4 in a patient with a clinical presentation and platelet phenotyping suggestive of HPS [65]. Furthermore, we have gone on to identify several candidate novel mutations in other participants in the GAPP study, and these are currently being verified (Stockley, Jones, Daly and Mumford, personal communication). We have now moved to the use of whole exome sequencing arrays in the GAPP study, as this is more cost-efficient than the use of targeted arrays, but we have continued to focus on the candidate genes based on the platelet phenotype. The follow-up of candidate mutations necessitates Sanger sequencing of affected and unaffected family members for verification, and, in the case of point mutations, demonstration of loss of protein function in a transfected cell line (Fig. 3).

Figure 3.

Genotyping and platelet phenotyping (GAPP) workflow. Illustrative workflow for samples from patients recruited to the GAPP study.

NGS and large populations

A distinct approach is being used in the multinational Bridge-BPD study (UKCRN ID 11131) led by W. Ouwehand, investigating the genetic basis of clinical bleeding conditions of suspected platelet origin but in the absence of detailed phenotypic information ( The power of this study comes from the very large number of patients and the prediction that common variants that underlie bleeding will be found in a significant number of patients. The rarity of PFDs necessitates international collaboration to achieve sufficient numbers. Nevertheless, the major challenge, in the absence of phenotypic information, is to establish whether novel variants are related to the defect in bleeding.

Future developments

Although genetic studies provide great scope for developing the field of PFDs in a seemingly cost-effective manner, the challenge will be in processing all of the data generated and finding methods to identify the causative gene among other variations. Before this can be done, the normal variations must be identified, which is not a simple task. An essential direction in the future of platelet genetic research will therefore be the use of additional approaches, including proteomics and systems biology, to enable us to filter and combine genetic data.

Functional verification of mutations and novel biological insights

In many cases, expression of the mutated protein in an immortalized or primary cell line is essential to establish the functional consequence of the mutation. This is particularly challenging in the context of PFDs, owing to the need to study function following expression in primary megakaryocytes or platelets, although, for some mutations, expression in a non-megakaryocyte cell may be sufficient. A high titer of infection in hematopoietic progenitors or primary megakaryocytes can only be achieved by viral technology, and this is both time-consuming and challenging. Furthermore, it is still not possible to generate large numbers of platelets in vitro, and, at the present time, platelet generation is best achieved by bone marrow transplantation in mice. However, this is not appropriate for the investigation of many human genes, especially when there is no counterpart in the mouse or the variant is in a non-conserved region. Thus, on some occasions, it may be necessary to rely on structure–function prediction programs and cosegregation to verify an association with platelet function.

The analysis of point mutations has the potential to provide new insights into platelet function, as illustrated by the discovery of a patient with a proline to alanine mutation in the PDZ-binding domain in the C-terminus of the P2Y12 ADP receptor [20]. This mutation has been shown to prevent recycling of the receptor back to the membrane following exposure to ADP in a transfected cell line, and is associated with reduced surface expression in the patient. We therefore speculate that, ordinarily, the P2Y12 receptor is internalized in platelets following contact with ADP released from damaged cells in the circulation, and is subsequently recycled to the membrane. In the patient with the P341A mutation, the receptor becomes trapped in an internal pool, and so cannot be recycled [20].


The partnership between the clinicians caring for and managing patients with PFDs and research scientists working on platelet function, signaling and biology is becoming stronger. Translational research will improve the diagnosis and classification of patients with these disorders, and may impact on treatment for future generations. Much fundamental work remains to be done on platelet-based bleeding disorders and finding ways to accurately diagnose them. There is an urgent need for the development of new, user-friendly, standardized and high-throughput functional tests that can be performed on small volumes of blood. The 96-well plate format in combination with flow cytometry has the potential to offer this, subject to validation, as it has the power to measure a wide variety of parameters, including aggregation, secretion, procoagulant activity, and surface GPs, with just a few milliliters of blood. The reliance on flow cytometry also makes these assays less dependent on platelet count, and thus makes them potentially more applicable to patients with thrombocytopenia, as illustrated in a recent study [66]. On the other hand, the flow cytometry-based 96-well plate assay does not measure platelet spreading, and nor does it mimic the high shear conditions of the arterial circulation. Recently, there have been considerable developments in high-throughput microfluidics in the monitoring of platelet activation, and several commercial and in-house systems are now available [66, 67]. Although it is likely that these systems will be limited to specialist testing centers, they have the potential to identify defects in platelet function that are only revealed under flow conditions. Moreover, further developments in microfluidics are ongoing, including the development of specialist biosensors, e.g. for the detection of thrombin generation [68], and the use of systems biology to identify complex, multifactorial patterns of platelet dysfunction, as may be the norm in PFDs [69].

The gaps in our current understanding of the genetics underlying platelet physiology and function are recognized by major stakeholders, including funders such as the NHLBI working group of the NIH (USA) [70], who encourage research into the identification and characterization of the full spectrum of genetic variants associated with platelet biology, hemostasis, and thrombosis phenotypes, including bleeding. The challenge, however, in the complex trait of bleeding will be to link the genetic changes to the underlying bleeding disorder, and, in many cases, it may not be possible to do this, because of the multifactorial nature of the disorder. On the other hand, there is the real danger of classifying an individual as having a genetic predisposition for a PFD, when in reality they have little phenotypic change in platelet function. Thus, it will be important to recognize that, in many cases, especially those associated with heterozygosity, the gene mutation should be regarded as a risk factor rather than a causative agent in giving rise to excessive bleeding. The era of personalized medicine offers enormous potential phenotype–genotype-guided diagnoses and potential therapies, but requires both phenotypic and genetic information in complex traits such as PFDs.


Many of the ideas and concepts in this review have been developed through the GAPP consortium ( and through a Bleeding Disorders Workshop, held on 1–2 December 2011 in Birmingham, UK, with partial funding from the European Regional Development Fund. The GAPP study is chaired by S. P. Watson, and has principal investigators in Bristol (A. Mumford and S. Mundell), London (P. Gissen), and Sheffield (M. Daly), with key collaborators in London (T. Warner) and Nottingham (S. Fox). We would like to acknowledge the support of all of the GAPP team and consultant hematologists at our collaborating UK Haemophilia Comprehensive Care Centres, in particular J. Wilde and M. Makris, who have played a pivotal role in the GAPP study since its inception. The GAPP study is supported by the British Heart Foundation (RG/09/007; PG/06/038; PG/11/31/28835) and Wellcome Trust (093994).

Disclosure of Conflict of Interests

The authors state they have no conflicts of interest.