Novel GNE Gene Variants Associated with Severe Congenital Thrombocytopenia and Platelet Sialylation Defect

Abstract

The GNE gene encodes an enzyme that initiates and regulates the biosynthesis of N-acetylneuraminic acid, a precursor of sialic acids. GNE mutations are classically associated with Nonaka myopathy and sialuria, following an autosomal recessive and autosomal dominant inheritance pattern. Reports show that single GNE variants cause severe thrombocytopenia without muscle weakness. Using panel sequencing, we identified two novel compound heterozygous variants in GNE in a young girl with life-threatening bleedings, severe congenital thrombocytopenia, and a platelet secretion defect. Both variants are located in the nucleotide-binding site of the N-acetylmannosamin kinase domain of GNE. Lectin array showed decreased α−2,3-sialylation on platelets, consistent with loss of sialic acid synthesis and indicative of rapid platelet clearance. Hematopoietic stem cell transplantation (HSCT) normalized platelet counts. This is the first report of an HSCT in a patient with an inherited GNE defect leading to normal platelet counts.

Introduction

The GNE gene codes for uridine diphosphate-N-acetylglucosamine 2-epimerase/N-acetylmannosamin kinase (GNE), a bidirectional enzyme that initiates and regulates the N-acetylneuraminic acid (Neu5Ac, or sialic acid) synthesis.¹ Sialyltransferases catalyze the addition of sialic acid to terminate underlying glycan structures. Given their location and ubiquitous distribution, sialic acids can mediate or modulate physiological and pathological processes, including prevention of rapid platelet and plasma glycoprotein clearance. Sialic acids, diversified by its C-2 anomeric carbon addition to the C-3 or C-6 position of galactose (Gal) resulting in α−2,3- or α−2,6-linked sialic acid, respectively, prevent rapid platelet and plasma glycoprotein clearance.² Platelet surface glycoproteins bear highly sialylated glycans.³ Loss of α−2,3-sialic acid correlates with thrombocytopenia^4,5 due to increased platelet clearance via Ashwell–Morell receptor-expressing hepatocytes and the recently identified macrophage galactose lectin on liver macrophages.^6,7 However, the specific sialic acid alterations, i.e., α−2,3- or α−2,6-, associated with increased platelet clearance remain to be determined.

Genetic variants in GNE are autosomal recessive associated with Nonaka myopathy (MIM#605820), a progressive muscular disorder and autosomal dominant with sialuria (MIM#269921), a metabolic disorder. Initial symptoms in patients with GNE myopathy appear during early adulthood⁸ and may exhibit thrombocytopenia.^9,10 Sialuria is characterized by massive free sialic acid excretion, coarse facial features, hepatosplenomegaly, and variable developmental delays.¹¹ These patients have heterozygous missense mutations affecting the allosteric site of the GNE enzyme causing a loss of feedback inhibition of GNE-epimerase activity by CMP-sialic acid, resulting in excessive sialic acid synthesis.¹² Another GNE phenotype causing autosomal recessive, isolated macrothrombocytopenia without associated muscle wasting even during adulthood (at age 24–42 years) has been recently reported, indicating a third distinct clinical phenotype.¹³

We report two novel compound heterozygous GNE variants identified in a young girl with life-threatening bleedings, severe congenital thrombocytopenia, and thrombocytopathy. Platelet glycan expression analyses by lectin arrays showed alteration in sialic acids with decreased α−2,3-sialylation and increased α−2,6-sialylation. β-Galactose exposure was increased indicating increased platelet clearance by β-galactose-specific lectins. Hematopoietic stem cell transplantation (HSCT) in this patient with inherited GNE defect normalized platelet counts.

Methods

Platelet Functional and Molecular Genetic Analyses

Platelet count, platelet aggregometry, and flow cytometry analyses were performed as previously described.¹⁴ Flow cytometry data of patient and controls were analyzed using GraphPad Prism software (version 8, San Diego, California, United States). A control cohort of six individual controls (20 independent measurements from 6 controls as mean±standard error of the mean) was established. In addition, a day control of a control person was performed together with each platelet function analysis of the patient.

After informed consent was obtained we used next-generation sequencing for molecular genetic analysis. Panel (targeting) enrichment of 95 genes (Supplementary Panel gene list, available in the online version) associated with inherited platelet disorders/bleeding disorders (Nextera Rapid Custom Enrichment, Illumina) followed by sequencing on a MiSeq (Illumina) was performed. Data were analyzed with SeqPilot (JSI medical systems) and variants were filtered according to minor allele frequency (MAF<1% in gnomAD) and serious consequences (nsSNV, essential splice site, nonsense, indel). Supporting software Alamut Visual, pathogenicity prediction (SIFT, MutTaster, PolyPhen2, CADD), and public version of the Human Gene Mutation Database were used. Family genotyping and variant verification was done using direct sequencing.

Lectin Array

Platelet Preparation

Whole blood was collected under an institutional review board-approved protocol in a 3 mL citrate tube and centrifuged for 15 minutes at 110×g with no brake within 15 minutes of collection. The platelet-rich plasma fraction was centrifuged with 1 μg/mL M PGE₁ at 850×g for 5 minutes with no break. To remove residual plasma from platelets, the platelet pellet was suspended with a modified platelet wash buffer (10 mM trisodium citrate, 140 mM NaCl, 1 mM EDTA, pH 6.0, and 1 μg/mL M PGE₁) and platelets were collected by centrifugation (850× g, 5 minutes) and lysed using NP-40 lysis buffer (150 mM NaCl, 50 mM Tris, pH 8.0, 1% NP-40, 1 mM EGTA, 1 mM Na3VO4).¹⁵ Glucose and sucrose were removed from the platelet wash buffer to prevent inhibition in downstream lectin analysis.

Lectin Array Acquisition and Analysis

The total protein concentration from each extracted cell fraction was determined using a Micro BCA Protein Assay kit (Thermo Fisher Scientific, Massachusetts, United States). Phosphate-buffered saline was used to dilute cell lysate fractions to 50 μg/mL with 100 μg of Cy3 monoreactive dye pack (GE Healthcare, Fairfield, Connecticut, United States). Samples were incubated for 1 hour at room temperature in the dark and with excess Cy3 removed using a Zeba Spin Desalting Column (Thermo Fisher Scientific, Boston, Massachusetts, United States). Optimized concentration (0.5 μg) of glycoprotein was added to an individual lectin microarray with 45 printed lectins (Supplementary Table S1, available in the online version) each in triplicate (GlycoTechnica, Yokohama, Japan). Fluorescent microarray images were acquired and analyzed using an evanescent-field fluorescence scanner using Signal Capture and GlycoStation Tools Pro (v1.5, GlycoTechnica, Yokohama, Japan). Optimal fluorescence was determined from images with minimal background and nonsaturated lectin well fluorescence.

Statistical Analysis

Raw fluorescence intensity data in .csv file were imported into R (v. 1.1.463).¹⁶ The triplicate values from each lectin were summed and lectins with fluorescence below the negative control spot were deemed to have no fluorescence and were excluded from the analysis. To account for lot-to-lot variation, data were quantile-normalized to align overall distributions.¹⁷ Quantile normalization was accomplished on each lectin so that the scaled values were not influenced by the entire range of values within the array. Each lectin was set to a maximum and minimum value of 2 and 2 for scaling, respectively. Normalized fluorescence data were then visualized with heatmaps (unscaled and row capped z-scaled) with unsupervised hierarchical clustering using Ward’s error sum of squares hierarchical clustering method.^18,19 Normalized values in red or closer to a normalized value of 2 were higher than the compared values within the same lectin. Principal component analyses were plotted with the first two dimensions with 95% confidence interval ellipses.^20,21 Fluorescence was fit to a linear model using the empirical Bayes method. Principal component analysis is a qualitative tool in observing the clustering of data or samples and has the ability to capture the variability of the data in two dimensions. It was used to investigate the overall similarity of lectin binding in the samples. Statistically significant differences were calculated as adjusted p-values <0.05. Based on the pathobiology of the genetic glycan lesion, sialic acid and galactose were the most likely affected glycans on platelets. Lectins distinguishing sialic acid and galactose were used in creating a linear model and establishing significance.

Results

Severe Thrombocytopenia in a Pediatric Patient of Unknown Etiology

After birth, the index patient developed intracranial bleeding requiring multiple platelet transfusions because of severe thrombocytopenia (lowest platelet count: 5 G/L). The girl needed weekly platelet transfusions (platelet counts: 15–40 G/L). At the age of 6 months, the girl was referred to our hospital. Besides hematomas the internal examination remained unobtrusive (no hepatosplenomegaly or dysmorphic stigma). Bone marrow analysis indicated slightly increased megakaryopoiesis with mature megakaryocytes of normal morphology (Fig. 1A) and hypolobulated, immature cells (Fig. 1B). In addition, only a few dysplastic megakaryocytes such as small binucleated forms were observed (Fig. 1C).

Fig. 1.

Giemsa staining of bone marrow (magnification 400×). (A) Bone marrow morphology revealed that most megakaryocytes were of normal morphology. (B) Immature, hypolobulated cells were increased in numbers. (C) The arrow indicates a small binucleated megakaryocyte.

Neonatal alloimmune thrombocytopenia was excluded by testing for antiplatelet antibodies in both parents and the index patient. Auto- or alloantibodies against platelets were not detected in the patient (exclusion of idiopathic thrombocytopenic purpura and neonatal alloimmune thrombocytopenia). Comprehensive testing for other platelet disorders has been performed and the following classic inherited thrombocytopenia/pathias have been excluded using genetic and biochemical analyses: CAMT (MPL gene), TAR syndrome, TPO/THPO-gene defect, Wiskott–Aldrich syndrome, MYH9-associated disorders, Glanzmann thrombasthenia, Bernard–Soulier syndrome, Fanconi syndrome, dyskeratosis congenita, ADAMTS13 deficiency, platelet-type von Willebrand disease, and type 2B von Willebrand disease. In blood smear analysis, inclusion bodies in leukocytes were not detectable, which would hint to MYH9-associated disorders. Platelet aggregometryand flow cytometryanalyses excluded Glanzmann thrombasthenia and Bernard–Soulier syndrome. Normal values for ADAMTS13 antigen and ADAMTS13 activity excluded thrombotic thrombocytopenic purpura. Von Willebrand factor (VWF):antigen, VWF:activity, VWF multimeric analysis, FVIII-activity, protein C, protein S, antithrombin, D-dimers, and C-reactive protein were normal (no sign for infection or activated coagulation). Genetically, we further investigated a large number of genes associated with megakaryopoiesis and platelet biogenesis. THPO, MPL, JAK2, and STAT3 are involved in regulation of hematopoietic stem cell proliferation. Other genes like HOXA11, MECOM, FLI1, RUNX1, GATA1, GFI1B, ETV6, MYH9, DIAPH1, FLNA, TUBB1, GP1BA, GP1BB, GP9, ITGA2B, and ITGB3 are involved in endoreplication, MK maturation, proplatelet formation, and platelet function. Additionally, we investigated the genes ACTN1, ANKRD26, CYCS, LYST, MASTL, MLPH, MYO5A, NBEA, NBEAL2, NFE2, ORAI1, STIM1, SLFN14, UNC13D, VIPAS39, VPS33B, and WIPF1 without any pathological findings (supplementary genes investigated). Microarray analysis excluded deletions.

Lupus antibodies were negative. Virus serology (immunoglobulin G) showed borderline values for rubella virus, parvovirus B19, varicella-zoster virus, and herpes simplex virus-1/2, and negative results for human herpesvirus-6, human immunodeficiency virus, and cytomegalovirus. Microbacteriological investigations of serum were also negative. Endocrinopathies, defects of amino acid metabolism, organoacidopathies, and fatty acid oxidation disorders were excluded via the German newborn screening. Magnetic resonance imaging (MRI) of the whole body and cranial MRI did not show any sign for vessel malformation or Kasabach–Merritt syndrome. Heart ultrasound did not reveal any signs for ventricular septum defect, aortic stenosis, or persistent ductus arteriosus. At the age of 24 months the girl showed normal statomotoric development and has not developed any signs of myopathy (normal muscular tonus). Therapy with immunoglobulin, prednisolone, romiplostim, eltrombopag, and oseltamivir did not improve platelet counts. The girl is the first child of Caucasian, non-consanguineous parents. The parents do not suffer from myopathy, thrombocytopenia, or bleeding symptoms.

Molecular Genetic Analysis Showed Novel Heterozygous Variants in the GNE Gene

We sequenced 99% of the 1,631 target regions at a minimum coverage depth of 20, and 91% with a minimum coverage depth of 100 for each nucleotide base-pair position of interest. The index patient carried two previously undescribed heterozygous variants, c.1250C>T (p.Thr417Met) and c.1259G>A (p. Arg420Gln), in the GNE gene (NM_005476.6) (Table 1), verified by Sanger sequencing. No candidate variants were identified in the other known thrombocytopenia-causing genes. The pedigree analysis confirmed segregation within the family per an autosomal recessive pattern. The mother was the carrier of the variant c.1250C>T, and the father presented with the variant c. 1259G>A.

Table 1.

Variants in GNE (NM_005476.6) CADD ≥20 indicates the 1% most deleterious substitutions

	Variant 1	Variant 2
Nucleotide	1250C>T	1259G>A
Protein	Thr417Met	Arg420Gln
Zygosity	Heterozygous	Heterozygous
Frequency (MAF%)
dbSNP	–	rs780092539
gnomAD	ALL: 0%	ALL: 0.00080%, NFE: 0.0018%
ESP	–	–
Mutation database
HGMD (public)	–	–
In silico pathogenicity prediction
Align GVGD	Class C65 (GV: 0.00, GD: 81.04)	Class C35 (GV: 0.00, GD: 42.81)
SIFT	Deleterious (score: 0)	Deleterious (score: 0)
MutationTaster	Disease causing (prob: 1)	Disease causing (prob: 1)
PolyPhen2	Probably damaging	Probably damaging
CADD score	26.9	32

Abbreviations: CADD, Combined Annotation Dependent Depletion; bdSNP, Single Nucleotide Polymorphism Database; ESP, Exome Sequencing Project.

Patient’s Platelets Have Decreased α-2,3-Sialic Acid and Increased Terminal Galactose and α-2,6-Sialic Acid Moieties

To measure alterations in sialic acid expression on platelets, we performed lectin array analysis of platelet glycoproteins. After platelet isolation and Cy3 labeling (Fig. 2A), platelet lysates from the father, mother, the index patient, and three healthy individuals were analyzed on a 45-probe lectin array and subjected to quantile normalization to account for biological and lot-to-lot variability of lectins (Fig. 2B). Data were analyzed based on known glycan deficits from GNE mutations in sialic acid and underlying structures including galactose which would be exposed with desialylation.² Consistent with this notion, a statistically significant decrease in α−2,3-sialic acid with ACG and WGA lectins (p=0.008 and 0.003, respectively) was noted in the child (Fig. 2C). A reciprocal global increase in terminal galactose residues was measured (ABA, ACA, and RCA 120). The Ricinus communis lectin (RCA120), a terminal galactose (Galβ1–4) detecting lectin, was statistically significant (p=0.04; Fig. 2D). Principal component analysis demonstrated that 94.5% of lectin fluorescence data were visualized in two dimensions (x- and y-axes). Controls clustered together, as expected, and the lectin phenotypes of each family were distinct from the controls (Fig. 2E). In contrast, α−2,6-sialic acid was increased in the child with the GNE mutations compared with healthy controls observed with SNA, SSA, and TJA-I lectins (p=0.00007, 0.005, and 0.000007, respectively) (Fig. 2C). Although all three lectins bind sialic acid, SSA and TJA-I recognize sialic acid moieties on O-linked glycans. The parents were only significant for a loss of SNA (p=0.000007) and enhanced TJA-I (p=0.00007) binding compared with the controls.

Fig. 2.

Platelet glycan expression profile using lectin arrays. (A) Workflow for lectin microarray. (B) Platelet lysate lectin microarray quantile normalization capped Z-scale. (C) Statistically significant decrease (blue squares) in α−2,3-sialic acid with ACG and WGA lectins (p = 0.008 and 0.003, respectively) in the index patient. (D) Reciprocal global increase (red squares) in terminal galactose residues (ABA, ACA, and RCA 120), but only the Ricinus communis lectin (RCA120), a terminal galactose (Galβ1–4) detecting lectin, was statistically significant (p=0.04). In contrast, α−2,6-sialic acid was increased in the child (red squares [B]) compared with healthy controls observed with SNA, SSA, and TJA-I lectins (p = 0.00007, 0.005, and 0.000007, respectively). The parents were only significant for a loss of SNA (p = 0.000007) and enhanced TJA-I (p = 0.00007) binding compared with the controls. (E) Principal component analysis of lectin microarray fluorescence data visualized in two dimensions (x- and y-axes)

GNE Patient Platelets Have Impaired Function

At admission, the patient’s platelet count was 90 G/L and the mean platelet volume was 10.6 fL (3 days after last platelet transfusion). Platelet aggregation (normal: 70–100%) was severely impaired after stimulation with 4 and 10 μmol/L adenosine diphosphate (ADP; max. aggregation 22 and 55%, respectively) and with 8 μmol/L epinephrine (max. aggregation 21%). Aggregation after stimulation with 2 μg/mL collagen was slightly reduced (46%) and with 10 μg/mL collagen it was in the normal range (81%). Agglutination after stimulation with ristocetin (1.2 mg/mL) was normal (91%) (Fig. 3A). Flow cytometry analyses revealed severely decreased expression of CD62P and CD63, respectively, after stimulation with thrombin, hinting to a platelet α and d granule secretion defect. Also, VWF binding was impaired after stimulation with ristocetin, whereas fibrinogen binding after stimulation with ADP was borderline (Fig. 3B). Expression of CD42a, CD42b, and CD41 was normal (not shown).

Fig. 3.

Platelet function analysis. (A) Light transmission aggregometry was performed with different agonists and concentrations. Shown are maximal platelet aggregations after stimulation with ADP, epinephrine, and collagen and agglutination after ristocetin compared with healthy control (norm ≥70%). (B) Flow cytometry analyses show moderately impaired VWF binding after stimulation with ristocetin (concentrations: 0, 0.2, 0.3, 0.5, 0.75, 1.0 mg/mL). Fibrinogen binding after stimulation with ADP (concentrations: 0, 0.25, 0.75, 2.0 μmol/L) was at the lowest range of the control group. Platelet granule secretion was stimulated with thrombin (concentrations: 0, 0.05, 0.1, 0.2, 0.5, and 1.0 U/mL) and showed moderately impaired α-granule and d-granule secretion indicated by reduced platelet CD62 and CD63 expression compared with the healthy day control/controls. Data of patient and controls (day control and 20 independent measurements from six controls as mean ±standard error of the mean, SEM) were analyzed using GraphPad Prism software (version 8, San Diego, California, United States). ADP, adenosine diphosphate; VWF, von Willebrand factor.

In Silico Simulation Suggests that Patient Mutations Affect ADP Binding

The index patient inherited two compound heterozygous mutations p.Thr417Met and p.Arg420Gln from each parent (Fig. 4A). These mutations are located in the initial nucleotide-binding site of the ManNAc kinase domain (residues 411–420). We used the crystal structure of GNE (PDB 2YI1) and PyMOL software to determine in silico how these mutations affect GNE function. In the wild-type protein, Thr417 and Arg420 form hydrogen bonds with the ADP β-and α-phosphates, respectively (Fig. 4B).²² In silico mutations show that p.Thr417Met (T417M) and p.Arg420Gln (R420Q) disrupt ADP binding (Fig. 4C).

Fig. 4.

T417M and R420Q mutations impair ADP binding of GNE. (A) Schematic diagram showing GNE domains and mutations in index patient and parents. (B) Crystal structure of GNE kinase domain in complex with ADP (PDB 2YI1) shows the hydrogen-bonding interactions of T417 and R420 with ADP. (C) In silico mutations of T417M and R420Q suggest disruption of ADP binding. The figures were generated with PyMOL. Hydrogen bonds are dashed lines in yellow. ADP, adenosine diphosphate.

Therapy with Hematopoietic Stem Cell Transplantation

Because it was unclear whether platelet transfusions will be continuously sufficient to ameliorate the severe thrombocytopenia, the girl has now received HSCT with a T cell reduced graft from a matched unrelated donor. The myeloablative-conditioning regimen comprised targeted busulfan, fludarabine, and thiotepa. Additionally, short-course MTX (methotrexate) and CSA (cyclosporine) were used for graft versus host disease (GvHD) prophylaxis. The girl displayed fast engraftment of neutrophils and platelets (100–200 G/L) without signs of GvHD. Chimerism analysis of peripheral blood showed a full donor type at day 24 posttransplant. Follow-up hematological studies showed normalizing platelet counts increasing from 95 G/L 1 month after HSCT to 245 G/L. Two months after HSCT, platelet count was stabilized above >200 G/L and platelet count at the last patient visit (11 months after HSCT) was 375 G/L. So far, the child has not presented with any complications.

Discussion

We identified a patient with two novel compound heterozygous variants p.Thr417Met and p.Arg420Gln in GNE. Both wild-type nucleotides and amino acids are phylogenetically highly conserved, and the variants are predicted to be pathogenic.

Both GNE variants p.Thr417Met and p.Arg420Gln localized in the initial nucleotide-binding site of the ManNAc kinase domain (residues 411–420), consistent with GNE variants associated with refractory thrombocytopenia in newborns or young children. In 2018, Futterer et al described two Pakistani children (cousins) with severe inherited thrombocytopenia (baseline count 10–20×10⁹/L) needing recurrent platelet transfusions.²³ One patient developed bilateral intraventricular bleeding shortly after birth. Exome Sequencing in both patients revealed an homozygous variant in GNE (NM_005476:c.1246G>A, p.Gly416Arg) located directly aside from one of the two GNE variants (p.Thr417Met) identified in our index patient. In 2020, Li et al described compound heterozygous GNE variants (NM_001128227.2: c.1330G>T (p.Asp444Tyr) and c.1351C>T (p.Arg451*)) in newborn twins with refractory thrombocytopenia.²⁴ In the shorter transcript (NM_005476.6) used in our study and the most publications, these alterations are located in p.Asp413Tyr and p.Arg420*, respectively.

Crystal structures analysis by Martinez et al showed among others that Asp413 is required for Mg²⁺ coordination and therefore crucial for adenosine triphosphate binding, while Thr417 stabilizes the ADP β-phosphate and Asn418 and Arg420 coordinate the α-phosphate.²² In silico mutations suggest that p.Thr417Met and p.Arg420Gln disrupt ADP binding, likely affecting the kinase activity of the enzyme. In support, site-directed mutagenesis of GNE p. Arg420Met results in the loss of the kinase activity, while the epimerase activity is retained.²⁵

Despite having heterozygous p.Thr417Met and p. Arg420Gln mutations, the parents of the index patient did not display thrombocytopenia and had a distinct sialic acid expression profile in the lectin array, suggesting compensation by the control allele. Consistent with earlier data, our findings indicate that mutations of two alleles in the kinase nucleotide binding site, either homozygous or compound heterozygous, are associated with thrombocytopenia.

GNE variants causing a reduction of sialic acids leading to a shortened platelet life span mainly affect the kinase domain.^13,23 The girl (compound heterozygous) showed a decreased level of α−2,3-sialylation on platelets with a previously unknown reciprocal increase in α−2,6-sialylation (lectin microarray). Loss of α−2,3-sialylation correlated with galactose reciprocal exposure of terminal β-galactose moieties. These data corroborated earlier findings that mutations in the kinase domain also affect platelet sialic acid levels which may lead to increased platelet clearance via galactose receptors. However, the significance of increased α−2,6-sialylation is currently unknown. It is tempting to speculate that the hepatic Ashwell–Morell receptor which recognizes α−2,6-sialylated moieties removes platelets with increased α−2,6-sialic acid moieties.²⁶ Thus, the compound heterozygous GNE variants in the girl appear to be the disease-causing variants. The severe thrombocytopenia may be due to increased clearance via lectin receptors, including the Ashwell–Morell receptor and the recently identified macrophage lectin.

The girl has not suffered from myopathy so far; however, she required weekly platelet transfusions before successful treatment with HSCT which normalized platelet counts. These data show that HSCT is a therapeutic option for patients with a GNE defect who present solely with thrombocytopenia. So far, the child has not presented with any complications, long-term follow-up will monitor if the patient will develop any further symptoms.

What is known about this topic?

• Sialic acids prevent rapid platelet and plasma glycoprotein clearance.

• GNE encodes a bidirectional enzyme that initiates and regulates the biosynthesis of N-acetylneuraminic acid, a precursor of sialic acids.

• Single patients have been described with GNE variants which only cause isolated thrombocytopenia.

What does this paper add?

• Two novel compound heterozygous variants in the GNE gene in a patient with severe congenital thrombocytopenia and a platelet secretion defect were identified (using next-generation sequencing).

• Lectin array analysis of platelet glycoproteins shows decreased α−2,3-sialylation and elevated α−2,6-sialylation with reciprocal increase in terminal galactose, associated with increase in platelet clearance.

• Hematopoietic stem-cell transplantation was successfully performed and led to normal platelet count.