Journal of Molecular Medicine https://doi.org/10.1007/s00109-018-1661-6

ORIGINAL ARTICLE

Targeted resequencing of a locus for heparin-induced thrombocytopenia on chromosome 5 identified in a genome-wide association study Anika Witten 1 & Juliane Bolbrinker 2 & Andrei Barysenka 1 & Matthias Huber 2 & Frank Rühle 1 & Ulrike Nowak-Göttl 3 & Edeltraut Garbe 2,4 & Reinhold Kreutz 2 & Monika Stoll 1,5 Received: 15 September 2017 / Revised: 5 June 2018 / Accepted: 12 June 2018 # Springer-Verlag GmbH Germany, part of Springer Nature 2018

Abstract Immune-mediated heparin-induced thrombocytopenia (HIT) is the clinically most important adverse drug reaction (ADR) in response to heparin therapy characterized by a prothrombotic state despite a decrease in platelet count. We conducted a genomewide association study in 96 suspected HIT cases and 96 controls to explore the genetic predisposition for HIT within a case-control pharmacovigilance study followed by replication in additional 86 cases and 86 controls from the same study. One single nucleotide polymorphism (SNP, rs1433265, P = 6.5 × 10−5, odds ratio (OR) 2.79) from 16 identified SNPs was successfully replicated (P = 1.5 × 10−4, OR 2.77; combined data set P = 2.7 × 10−8, OR 2.77) and remained the most strongly associated SNP after imputing locus genotypes. Fine mapping revealed a significantly associated risk-conferring haplotype (P = 4.9 × 10−6, OR 2.41). In order to find rare variants contributing to the association signals, we applied a targeted resequencing approach in a subgroup of 73 HIT patients and 23 controls for the regions with the 16 most strongly HIT-associated SNPs. C-alpha testing was applied to test for the impact of rare variants and we detected two candidate genes, the discoidin domain receptor tyrosine kinase 1 (DDR1, P = 3.6 × 10−2) and the multiple C2 and transmembrane domain containing 2 (MCTP2, P = 4.5 × 10−2). For the genes interactor of little elongation complex ELL subunit 1 (ICE1) and a disintegrin-like and metalloproteinase with thrombospondin type 1 motif, 16 (ADAMTS16) nearby rs1433265, we identified several missense variants. Although replication in an independent population is warranted, these findings provide a basis for future studies aiming to identify and characterize genetic susceptibility factors for HIT. Key messages & & &

We identified and validated a HIT-associated locus on chromosome 5. Targeted NGS analysis for rare variants identifies DDR1 and MCTP2 as novel candidates. In addition, missense variants for ADAMTS16 and ICE1 were identified in the locus.

Anika Witten, Juliane Bolbrinker, Reinhold Kreutz and Monika Stoll contributed equally to this work. Electronic supplementary material The online version of this article (https://doi.org/10.1007/s00109-018-1661-6) contains supplementary material, which is available to authorized users. * Monika Stoll [email protected] 1

Department of Genetic Epidemiology, Institute of Human Genetics, University Hospital Münster, Münster, Germany

2

Charité - Universitätsmedizin Berlin, corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Institute of Clinical Pharmacology and Toxicology, Berlin, Germany

3

Thrombosis and Hemostasis Unit, Department of Clinical Chemistry, University Hospital of Kiel and Lübeck, Kiel, Germany

4

Department of Clinical Epidemiology, Leibniz Institute for Prevention Research and Epidemiology – BIPS, Bremen, Germany

5

Department of Biochemistry, Cardiovascular Research Institute Maastricht, Maastricht University, Maastricht, The Netherlands

J Mol Med

Keywords ADR . HIT . GWAS . ADAMTS16 . ICE1

Introduction Apart from the increased bleeding risk associated with anticoagulant treatment, immune-mediated heparin-induced thrombocytopenia (HIT) is the clinically most important serious adverse drug reaction (ADR) of heparin therapy [1–4]. HIT represents an immune reaction initiated by antibodies directed to complexes between heparin and platelet factor 4 (PF4) [1–4]. By binding to platelet FcγIIa receptors (FcγRIIa), some of these antibodies lead to platelet activation with release of procoagulant microparticles and enhanced thrombin generation, thereby generating a prothrombotic state in a subset of patients, despite a decrease in platelet count [1–4]. Hence, unlike other drug-induced thrombocytopenias, HIT is characterized by an increased risk of venous and arterial thrombosis, which can lead to devastating clinical outcomes in affected patients. Interestingly, PF4/heparinspecific antibodies are also found in the general population [5] and it was recently demonstrated that heparin administration in healthy subjects without a history of HIT induces PF4/heparin antibody formation without a concomitant drop in platelet count or HIT [6]. A lower risk of HIT has been reported for low-molecular-weight heparins (LMWH) compared to unfractionated heparin (UFH) in postoperative patients [7], whereas in medical patients, no difference in the frequency of HIT between LMWH and UFH has been reported [8, 9]. In addition to the type of heparin and context of administration, the risk of HIT is modified by dose, indication for heparin, duration of treatment, underlying medical condition, and individual factors such as age and sex [10–13]. Following a case report in 1985 on HIT complicated by severe thrombosis in two brothers [14], a genetic predisposition to develop HIT has been suggested and genetic variants in several candidate genes with a possible role in the pathogenesis of HIT have been previously investigated, however, with inconsistent results [15–24]. More recently, a genome-wide association study (GWAS) in HIT patients was reported [25] observing single nucleotide polymorphisms (SNPs) near the T cell death-associated gene 8 (TDAG8) to be associated with HIT on a genome-wide level [25]. The identification of genetic variants conferring an increased risk for HIT is important, because it could prevent the development of this adverse drug reaction and its potential catastrophic outcome by considering alternative treatments [4, 26]. In this report, we explored the genetic predisposition for HIT by conducting a GWAS within a case-control pharmacovigilance study program. A candidate region on chromosome 5 was identified and further characterized by fine mapping and targeted deep

resequencing of 12 genomic regions comprising the 16 most strongly HIT-associated SNPs.

Methods Study population Cases HIT cases were recruited within the Berlin Case-Control Surveillance Study in our pharmacovigilance center at the Charité - Universitätsmedizin Berlin, Germany as described [27]. Validation of suspected HIT cases was reported recently [28]. Briefly, suspected HIT cases were identified by physicians of participating hospitals and laboratories based on newly diagnosed thrombocytopenia < 150 × 109/L and/or > 50% fall in platelet count while on treatment with UFH or LMWH for ≥ 5 days. Subsequently, suspected HIT cases reported to the pharmacovigilance center were thoroughly evaluated and a standardized interview was conducted by a trained interviewer of the pharmacovigilance center. In addition, a medical case report form (CRF) was filled in by the treating physician or interviewer with data of the patient’s records including—among others—laboratory tests on platelet count before and during heparin therapy and test results for HIT antibodies. The latter were performed at the discretion of the treating physicians in suspected HIT cases and included the functional heparin-induced platelet activation (HIPA) test [29], a heparin-PF4 enzyme-linked immunosorbent assay (ELISA) (Asserachrom HPIA, Diagnostica Stago), and the ID-Heparin/PF4 antibody test (ID-HPF4) [30]. In some patients, more than one test was documented. Suspected HIT cases were finally validated by pre-specified criteria as certain, probable, possible, or not confirmed based on information from laboratory and clinical information provided in the CRF and itemized interview. If necessary, the patient’s hospital records and consultation with the treating physician were also taken into account as well as review with an external hematological advisory board of the pharmacovigilance study program [28]. A more detailed description of the final validation criteria is given in the supplementary Methods S1. Further analysis was restricted to 209 HIT patients with a status of either certain (n = 131) or probable (n = 78) who gave written informed consent to participate in the pharmacovigilance study and completed the standardized interview.

J Mol Med

Controls Control patients were recruited in the same hospitals as cases. Crucial criterion for selection as a control was exposure to the same class of heparin (UFH or LMWH) for at least 5 days without clinical evidence of HIT. Matching did not account for indication for heparin therapy, age, and sex. Finally, 231 controls with written informed consent to participate in the pharmacovigilance study and who completed the standardized interview were included. Pharmacogenetic study An independent written informed consent for additional blood withdrawal and participation in the pharmacogenetic study was essential for inclusion. Of the 209 eligible certain or probable HIT cases, 27 were excluded (9 refused to participate, 15 no blood available, 3 not of European ancestry). Selfreported information on ancestry (place of birth, nationality, and birthplace of parents and grandparents) was retrieved during the face-to-face interview conducted during the pharmacovigilance study. The remaining 182 eligible cases were subsequently reevaluated, retrospectively applying two clinical probability scoring models for HIT, namely the 4Ts score [31] and the HIT Expert Probability (HEP) score [32]. Of the 231 eligible controls, 49 were excluded (18 refused to participate, 28 no blood available, 3 matched to the 3 cases not of European ancestry) leaving 182 control patients for genetic analysis. Clinical characteristics of cases and controls included in the pharmacogenetic study were derived from the CRFs and itemized interviews (Table 1). Results of the tests for HIT antibodies, if performed, for cases included in the pharmacogenetic study are given in supplementary Table S1. The study protocol was approved by the ethics committee of the Charité - Universitätsmedizin Berlin, Germany (file number 1369/ 2000). Genotyping Genotyping of the discovery samples was performed using Illumina HumanCNV370-Quad v3.0 (Illumina, Inc.) bead arrays. DNA samples (96 cases and 96 controls, discovery set, n = 192) were prepared, processed, and scanned using the iScan™ System and iScan™ Control software v2 following the manufacturer’s instructions. Genotype data were exported from Illumina GenomeStudio™ 2010 and all further analyses were performed using PLINK v1.07 [33]. Two control samples (one female, one male) missed a chip-wide call rate of ≥ 95%, resulting in an initial set of 351,507 markers in 190 individuals with an average call rate of 97.51%. Individual SNPs were excluded having an overall minor allele frequency (MAF) of ≤ 3%, missing SNP genotyping rate of more than 10% or a Hardy-Weinberg equilibrium (HWE) test P value ≤

0.001 in control samples. In addition, a chi-square test of missing genotypes by case/control status was applied using a threshold P value of ≥ 0.05 to control genotype errors. Sex estimated by genotype data was matched to clinical data using Illumina GenomeStudio™ 2010 to avoid sample errors. Genotyping of the replication samples (86 cases and 86 controls) was performed either with the ABI PRISM® 7000 SDS instrument in conjunction with TaqMan® Genotyping Master Mix using pre-developed assays by Applied Biosystems or by direct DNA sequencing using the Big-Dye Terminator v1.1 Cycle Sequencing Kit on a 3130 Genetic Analyzer (Applied Biosystems). Genotyping of the 16 selected genome-wide associated SNPs was performed in the replication set (n = 172) and, subsequently, in all samples (n = 364). All genotyping assays showed a call rate of ≥ 99% and all SNPs were in HWE (P ≥ 0.001). Validation of the TaqMan® genotyping results was obtained in selected samples by direct DNA sequencing.

Statistical analysis Data quality (MAF, HWE, missingness test) was checked using the PLINK software package v1.07 [33]. To determine the genome-wide threshold of significance, empirical P values were estimated by adaptive permutation testing, as implemented in PLINK. Logistic regression models were implemented in PLINK, adjusted for age and gender and using additive genetic models, and genomic control corrected P values were calculated; SNPs surpassing a threshold of P ≤ 1 × 10−04 for suggestive genome-wide association were forwarded to subsequent analyses and replication. LocusZoom plot was generated using the LocusZoom tool [34]. Tagging SNPs were selected based on the CEU data for a European ancestry population from the International HapMap Project [35] (HapMap Data Rel. 24, Phase II, Noc08, on NCBIassembly, dbSNP b126) using the HaploView software package [36]. SISA two-by-two table analysis was used to calculate haplotype odds ratios (OR) and 95% confidence intervals (CI). For genotype analysis, Ensembl human reference build GRCh38 was used to determine SNP and gene locations. Haplotype estimation and haplotype/phenotype association in the replication data set were performed using Haploview as well [36]. To impute additional genotypes on chromosome 5, the array data was phased using the ShapeIT [37] software tool (http://www. shapeit.fr/). Imputation was performed with IMPUTE2 [38] (http://mathgen.stats.ox.ac.uk/impute/impute_v2.html#top) using HapMap 3 and 1000 Genomes Pilot haplotypes (NCBI build 36 coordinates) as reference panel. Association analyses of imputed genotypes, adjusted for age and gender, were calculated assuming an additive genetic model using SNPTEST v2 [39]. Genotype uncertainty for imputed SNPs was incorporated using a missing data likelihood score test, as

J Mol Med Table 1

Clinical characteristics of HIT cases and controls included in the study Discovery set

n (female) Age, mean ± SD, years Type of heparin, n UFH LMWH Both Heparin exposure, mean ± SD, daysb Thromboembolic complications, n (%) Indication for heparin, n (%) Preventive Therapeutic Clinical context of heparin administration, n (%) Cardiovascular surgery Orthopedic surgery General surgery Cardiovascular disease Treatment of thromboembolism Other (infections, falls, internal medicine) Comorbiditiesc , n (%) Diabetes mellitus Arterial hypertension Coronary artery disease/myocardial infarction Atherosclerosis/peripheral artery disease Stroke Cancer Clinical probability scoring modelsd 4 Ts scoree HEP scoref

Replication set

Resequencing seta

Total

Cases

Controls Cases

Controls

Cases

Controls

Cases

Controls

96 (47) 64.9 ± 13

96 (47) 86 (41) 64.5 ± 14 63.8 ± 15

86 (36) 59.9 ± 13

182 (88) 64.4 ± 14

182 (83) 62.3 ± 13

73 (36) 64.4 ± 13

23 (9) 65.7 ± 15

96 – – 7.1 ± 5 25 (26)

96 – – 11.9 ± 9* –

31 37 18 10.0 ± 7 23 (27)

44 42 – 15.2 ± 14* –

127 37 18 8.5 ± 6 48 (26)

140 42 – 13.4 ± 12* –

54 14 5 8.3 ± 4 30 (41)

15 8 – 14.3 ± 11* –

27 (28) 69 (72)

34 (35) 62 (65)

38 (44) 48 (56)

32 (37) 54 (63)

65 (36) 117 (64)

66 (36) 116 (64)

23 (32) 50 (69)

11 (48) 12 (52)

42 (44) 1 (1) 9 (9) 6 (6) 15 (16) 23 (24)

58 (60)* 9 (9)* 7 (7) 5 (5) 5 (5)* 12 (13)

31 (36) 11 (13) 13 (15) 1 (1) 12 (14) 18 (21)

38 (44) 14 (16) 18 (21) 7 (8) 2 (2)* 7 (8)*

73 (40) 12 (7) 22 (12) 7 (4) 27 (15) 41 (23)

96 (53)* 23 (13) 25 (14) 12 (7) 7 (4)* 19 (10)*

34 (47) 6 (8) 9 (12) 2 (3) 14 (19) 8 (11)

9 (39) 3 (13) 4 (17) 4 (17)* 1 (4) 2 (9)

30 (31) 66 (69) 33 (34) 16 (17) 11 (12) 15 (16)

28 (29) 54 (56) 44 (46) 39 (41)* 12 (13) 13 (14)

23 (27) 40 (47) 34 (40) 14 (16) 10 (12) 21 (24)

17 (20) 58 (67)* 43 (50) 12 (14) 8 (9) 15 (17)

53 (29) 106 (58) 67 (37) 30 (17) 21 (12) 36 (20)

45 (25) 112 (62) 87 (48)* 51 (28)* 20 (11) 28 (15)

18 (25) 45 (62) 28 (38) 15 (21) 11 (15) 14 (19)

4 (17) 15 (65) 13 (57) 7 (30) 1 (4) 3 (13)

4.8 (3–8) – 3.8 (− 6 to 13) –

4.9 (3–8) – 4.7 (− 3 to 14) –

4.9 (3–8) – 4.2 (− 6 to 14) –

5.7 (4–8) – 6.3 (0–14) –

HIT heparin-induced thrombocytopenia, n number of samples, SD standard deviation, UFH unfractionated heparin, LMWH low-molecular-weight heparin, HEP score HIT Expert Probability score a

Cases and controls of the resequencing set were selected from the discovery and replication set (cases: n = 45 discovery set, n = 28 replication set; controls: n = 12 discovery set, n = 11 replication set)

b

In cases defined as days after initiation of heparin until index date (index date defined as date of platelet fall > 50% of baseline, if this was not applicable, date of positive antibody test result was defined as the index date)

c

Data on comorbidities were derived from CRF and interview combined. For two controls, no comorbidity was documented

d

Clinical probability scoring models for HIT were retrospectively applied to cases eligible for the pharmacogenetic study

e

According to Warkentin (2003 #277)

f

According to Cuker (2010 #51)

*P < 0.05 compared to cases using t tests or chi-square tests as appropriate

implemented in SNPTEST (for quality control setting, see above). Target enrichment and next generation sequencing data analysis Target candidate regions were selected for subsequent next generation sequencing (NGS) based on linkage disequilibrium (LD) information and association P values using the postgwas package [40]. A sequencing library was prepared using 1 μg sonicated DNA (S220, Covaris) and the TruSeqDNA LT Sample Prep Kit (Illumina) according to the manufacturer’s instructions. The solution-based SeqCap EZ Choice Library (Roche NimbleGen) capture method was applied for custom

target enrichment of the defined regions (supplementary Table S5). Library quality control and final quantification for subsequent pooling were performed using the LapChipGX system (PerkinElmer). The library was paired-end sequenced (2 × 101 cycles) on a HiScanSQ system (Illumina) using SBS chemistry v3. The CASAVA software v1.8.2 (Illumina) was used for demultiplexing of the sequencing reads and conversion to fastq data for further analysis. The reads were mapped against the human reference genome (GRCh37/hg19) using the BWA v0.7.12 aligner [41]. SNP and short insertions/ deletions (indel) calling were performed using the GATK v3.5.0 software package [42] followed by variant annotation using SnpEff v4.2. SIFT and PROVEAN scores were predicted using SnpSift v4.2 and dbNSFP v2.9 [43, 44]. Single

J Mol Med

variants detected by NGS in more than six samples were tested using logistic regression (additive model, adjusted for age and sex). P values were corrected for multiple testing using false discovery rate (FDR). Rare-variant association analysis of variants within the target regions was performed on SNPs with an overall minor allele frequency (MAF) of ≤ 2%, a call rate of > 95%, and a minor allele count of 1 to 3 using a Calpha test statistic [45]; both tests are implemented in PLINK/ SEQ (http://atgu.mgh.harvard.edu/plinkseq). Empirical P values were estimated using 10,000 permutations per gene.

Results Genome-wide association study An exploratory GWAS was conducted by selecting 96 HIT cases and 96 matched controls from our pharmacovigilance study; both groups were treated with UFH (discovery set, Table 1). The genomic inflation factor based on the median chi-squared statistic was 1.03, indicating absence of population stratification (QQ plot, Fig. S1 in the supplementary appendix). The absence of population stratification is also reflected by only slightly adjusted GC P values (Table 2). In total, 16 HIT-associated SNPs with P ≤ 1 × 10−4 (unadjusted for multiple testing) were identified on 10 chromosomes through logistic regression analysis and an additive model adjusted for age and sex (Table 2 and supplementary Fig. S2). Empirical P values from adaptive permutation testing were in the same range of significance, suggesting that model assumptions did not produce spurious results.

Replication analysis To obtain replication of the results, these 16 SNPs were validated in the remaining 86 HIT cases and 86 controls (replication set, Table 1). In this analysis, one SNP on chromosome 5 (rs1433265) was significantly associated with HIT in the replication set (P = 1.5 × 10−4) with an OR of 2.77 (95% CI, 1.64–4.68). The replication results for all 16 SNPs are listed in Table 2. After imputing genotypes on chromosome 5, rs1433265 remained the most strongly associated SNP among all typed and imputed SNPs (P = 3.5 × 10−5, supplementary Table S2 and Fig. S3). Figure 1 gives a detailed view on the locus. Genotyping in the combined analysis set (n = 364) revealed an association P value of 2.7 × 10−8 with an OR of 2.77 (95% CI, 1.93–3.96) (supplementary Table S3). As 31 of the 182 HIT cases included in the genetic study had a documented negative HIPA test result (supplementary Table 1), we conducted a sensitivity analysis on the 16 initially HITassociated SNPs restricted to the 124 HIT cases with a positive HIPA test result (see supplementary Table S4).

To further characterize the replicated HIT-associated locus, fine mapping was performed in the combined discovery and replication sets (n = 364) using eight haplotype-tagging SNPs. Five out of eight tagging SNPs on chromosome 5 demonstrated significant associations, with P values ranging between 4.0 × 10−5 and 6.4 × 10−6 (supplementary Table S3). These SNPs are located in a 33.7-kb intergenic region on the short arm of the chromosome comprising the lincRNA AC106799.2. The replicated SNP rs1433265 represents an intronic variant of this lincRNA. The closest annotated genes are the iroquois homeobox protein 1 (IRX1) on the 5′ side of the locus and a disintegrin-like and metalloproteinase with thrombospondin type 1 motif, 16 (ADAMTS16), and interactor of little elongation complex ELL subunit 1 (ICE1) on the 3′ end of the locus. LD structure was determined (supplementary Fig. S4) and haplotype analysis with all tagging SNPs, including rs1433265, revealed five common haplotypes with a frequency > 5% of which the two most frequent haplotypes were significantly associated with HIT (P = 5.6 × 10−3 and P = 4.9 × 10 − 6 , respectively). One of these haplotypes (CCACTCACG) exhibited a risk-conferring effect with an OR of 2.41 (95% CI, 1.64–3.55) (supplementary Fig. S4).

Targeted resequencing We applied a targeted resequencing approach of 12 genomic regions comprising the 16 most strongly HIT-associated SNPs to discover potential rare coding variants contributing to the association signal (3.8 Mb, supplementary Table S5). Further, a gene network analysis of those 12 genomic regions in postgwas [40] (data not shown) identified ADAMTS1 as a candidate gene for HIT, which was thus incorporated in the target design. Due to its impact on platelet and thrombus regulation and important role in thrombotic diseases such as thrombotic thrombocytopenic purpura (TTP), ADAMTS13 was added to the target design. NGS-based resequencing was performed in a subgroup of 73 HIT patients and 23 control samples selected from the discovery and replication set (Table 1). All cases in the subgroup had a positive HIPA test result and exhibited scores of at least four in the 4Ts score model, thus complying with the current expert recommendations that diagnosis of HIT should be based on a combination of intermediate to high probability score and laboratory test results for anti-PF4/heparin antibodies [4, 26, 46]. On average, 70.2% of sequenced base pairs had a quality score ≥ 30 and 3.5 million single reads per sample were mapped to the reference genome, resulting in a mean target region specificity of 68%. Mean coverage values are given in supplementary Table S5. Seven samples (four cases, three controls) were excluded from the analysis due to an on-target rate of < 10%. Overall, 24,834 SNPs and 3039 indels were detected, 206 of them were missense. No HIT-associated variant was detected in the case-control design by logistic regression analysis (data not

1

1 5 6 6 7 10 12

12

12 12 12 14 15 16 19

rs2982364

rs1883109 rs1433265 rs2256919 rs2535319 rs798332 rs9416637 rs2036288

rs11147217

rs2088685 rs905227 rs595241 rs3748316 rs10520749 rs878340 rs10402917

133,051,977 133,052,317 133,077,993 94,387,527 94,364,459 14,301,301 4,211,259

133,030,455

34,427,312 4,474,476 29,972,973 30,746,702 78,279,611 57,828,823 132,971,958

11,372,687

Position

ZNF84 ZNF84 ZNF140, AC073911 SERPINA1 MCTP2 MIR193B, MIR193BHG ANKRD24

PTP4A1P2, AC073911.1, ZNF26

C1orf94 AC106799.2 MICD, HCG9 IER3, HCG20 MAGI2 LOC105378314 NANOGNBP2

LOC105376740

(Associated) Gene

9.05 × 10 9.05 × 10−06 4.59 × 10−05 1.38 × 10−05 1.59 × 10−05 6.60 × 10−05 6.08 × 10−05

−06

1.91 × 10−05

6.19 × 10 6.47 × 10−05 6.28 × 10−05 7.53 × 10−06 8.49 × 10−05 5.65 × 10−05 1.86 × 10−05

−05

8.96 × 10−05

P value

0.36 (0.22–0.57) 0.34 (0.21–0.55) 0.34 (0.21–0.55) 2.54 (1.62–3.97) 0.25 (0.13–0.47) 2.97 (1.81–4.88) 2.58 (1.62–4.11) 3.71 (1.95–7.03)

0.39 (0.25–0.63) 2.81 (1.69–4.65) 2.79 (1.69–4.61) 2.54 (1.61–4.01) 0.33 (0.2–0.54) 0.4 (0.26–0.63) 2.65 (1.65–4.25) 0.35 (0.22–0.57)

OR (95% CI)

8.00 × 10−06 9.00 × 10−06 9.00 × 10−06 6.38 × 10−03 2.00 × 10−06 6.29 × 10−05 3.27 × 10−03 3.20 × 10−05

3.00 × 10−05 1.62 × 10−03 3.20 × 10−05 1.09 × 10−03 3.96 × 10−04 6.71 × 10−05 1.20 × 10−05 2.00 × 10−06

Perm. P value

2.50 × 10−05 1.21 × 10−05 1.21 × 10−05 5.87 × 10−05 1.83 × 10−05 2.10 × 10−05 8.37 × 10−05 7.72 × 10−05

1.13 × 10−04 7.85 × 10−05 8.20 × 10−05 7.97 × 10−05 1.01 × 10−05 1.07 × 10−04 7.18 × 10−05 2.44 × 10−05

GC P value

0.35 0.31 0.31 0.42 0.08 0.47 0.48 0.24

0.31 0.41 0.41 0.48 0.29 0.27 0.45 0.31

MAF cases

0.43 0.46 0.46 0.37 0.27 0.26 0.28 0.1

0.49 0.22 0.22 0.27 0.48 0.47 0.25 0.47

MAF controls

0.13 0.14 0.10 0.16 0.16 0.21 0.08 0.31

0.16 0.65 1.5 × 10−4 0.85 0.65 0.89 0.32 0.42

P value

1.40 (0.91–2.17) 0.72 (0.46–1.11) 0.69 (0.45–1.08) 1.37 (0.88–2.12) 0.67 (0.38–1.17) 1.34 (0.85–2.11) 0.66 (0.41–1.05) 1.36 (0.75–2.45)

0.72 (0.45–1.14) 1.12 (0.69–1.80) 2.77 (1.64–4.68) 0.96 (0.63–1.47) 1.11 (0.71–1.71) 1.03 (0.66–1.62) 1.25 (0.80–1.96) 0.84 (0.55–1.28)

OR (95% CI)

Replication (86 vs. 86)

Chr chromosome, Position chromosomal position in bp, OR odds ratio, CI confidence interval, GC P value, genomic control corrected P value, MAF minor allele frequency, MICD MHC class I polypeptide-related sequence D, HCG HLA complex group, IER immediate early response, MAGI membrane associated guanylate kinase, ZNF zink finger, SERPINA Serpin family a member, MCTP multiple C2 and transmembrane domain containing, ANKRD ankyrin repeat domain

Chr

Genome-wide association study (96 vs. 96)

Significant findings identified in the GWAS (discovery set) and corresponding replication P values, assessed using logistic regression (additive model) adjusted for age and sex

SNP

Table 2

J Mol Med

J Mol Med

Fig. 1 Plot of the region surrounding the most strongly associated SNP rs1433265 on chromosome 5. –log10 genome-wide association P values (determined by logistic regression, additive model adjusted for age and sex) and the recombination rate in centi Morgan per Mega base (cM/Mb) are plotted against the chromosomal position. The genomic region comprising the eight haplotype-tagging SNPs used for fine mapping is labeled in purple. Pairwise LD (r2) per SNP is color coded

shown) following an FDR-adjusted Q value of < 0.05, likely due to power constraints. Therefore, we performed a C-alpha test statistic, a collapsing test that identifies accumulations of rare variants between cases and controls to improve statistical power in the current setting. The test yielded significant associations for two genomic regions (see supplementary Table S6) comprising the discoidin domain receptor tyrosine kinase 1 (DDR1) on chromosome 6 (P = 3.6 × 10−2) and the multiple C2 and transmembrane domain containing 2 (MCTP2) on chromosome 15 (P = 4.5 × 10−2). On chromosome 6, we identified three missense variants with high (rs143367160) and moderate effects (rs1264319, rs140012475). In MCTP2, four missense variants were found: rs61737195, rs77311454, rs7178698, and a new variant at position 95,013,576. In the primary chromosome 5 locus coming from the initial GWAS, including rs1433265 and nearby genes ICE1 and ADAMTS16, we identified several missense variants in these two genes (see supplementary Table S7). The combined C-alpha P value of the rare variants within ADAMTS16 is 8.5 × 10−2 (see supplementary Table S6) curtly missing the threshold of significance. For IRX1, all detected variants were synonymous.

Discussion Here, we report the identification and validation of a HITassociated interval on chromosome 5 in HIT cases and

controls recruited from a single-center pharmacovigilance study program [28]. To correctly diagnose HIT in general clinical practice is a challenge [3, 4, 47]. Nonetheless, establishing a rapid diagnosis in patients with suspected HIT is critically important due to the necessity to discontinue heparin and the increased risk for thromboembolic complications mandating the initiation of alternative treatment [26]. To guide the clinical assessment process in suspected HIT patients, several diagnostic algorithms have been developed [31, 32, 48]. Evaluation of the 4Ts score [31] and the HEP score [32] particularly demonstrated their negative predictive value to rule out the presence of HIT, while their positive predictive value remains rather low [2, 49, 50]. None of the scoring models was in clinical use during the recruitment of cases and controls for the pharmacovigilance study. Nevertheless, suspected HIT cases underwent a thorough validation procedure and only cases with a classification status of certain or probable were put forward into the pharmacogenetic substudy. In an effort to additionally characterize our study population according to current established HIT probability models, the 4Ts [31] and HEP [32] scores were retrospectively applied to all cases on the basis of the available information in case report forms. This procedure might explain in part the findings of relatively low scores for some patients included in the pharmacogenetic study, because the models are usually applied in the acute setting. Thirty-one of our cases exhibited a negative HIPA test result; yet, a positive immunoassay test was documented in 29 of these cases and thus taken into consideration during the individual case evaluation in the pharmacovigilance study. Of note, none of the laboratory tests available to date has a sensitivity and specificity of 100% [4, 51, 52]. Especially immunoassays exhibit a sensitivity above 95% [51, 53] and it was shown that in patients with an intermediate clinical score for HIT—which applies to 23 of our cases with a negative HIPA test result—a positive immunoassay considerably increases the likelihood for HIT [54–56]. This underlines the importance of a combined consideration of clinical and laboratory data [4, 26, 46] as was done in our thorough evaluation in the pharmacovigilance study [28]. In our explorative GWAS, we failed to reach the commonly accepted genome-wide threshold for statistical significance of P < 5 × 10−8 in the discovery set. This is in our opinion attributable to the small sample size, a well-known challenge for GWAS on pharmacological traits [57] especially in the context of rare ADR such as HIT [58]. In addition, study controls were selected on a one to one basis in a nested study design, although a 1 to 2 ratio would have been more powerful for association analysis from a genetic-epidemiological perspective. One SNP, rs1433265, from initial 16 GWAS suggestive SNP associations was replicated and we are aware that the design bears the lack of an independent study population and, above all, a small replication set size. Nevertheless, the

J Mol Med

consistent association of rs1433265 in the discovery and replication samples and the strong association in the combined analysis of all 364 patients with a combined P value of 2.7 × 10−8, reaching genome-wide significance thresholds if we had combined the samples from the start in favor of a two-stage design, support a true association signal. Further, by imputation of additional SNPs to HAPMAP3 and 1000-GenomesProject data, we could also show that neighboring SNPs are at least moderately associated as expected for true associations. An independent replication of our association findings in another study population would strengthen the impact of our findings. The HIT-associated SNPs identified by GWAS and fine mapping are located in an intergenic region devoid of protein coding genes. The next annotated genes are IRX1 at the 5′ side and ADAMTS16 at the 3′ side of the associated interval. Both genes have not yet been linked to platelet function. In agreement with other gene products of the ADAMTS family of proteases, ADAMTS16 is mainly located in the extracellular matrix and described to be upregulated in tissue of aortic aneurysms [59]. Interestingly, ADAMTS18 [60], located on the same phylogenetic arm as ADAMTS16 [61], has been implicated in platelet fragmentation and platelet thrombus clearance via platelet oxidative fragmentation induced by thrombin cleavage of ADAMTS18 [60], suggesting that ADAMTS16 may exert similar functional properties. The SNP rs1433265 represents an intronic variant of AC106799.2, a lincRNA without any known functional annotation. Regardless of the potential functional relevance of noncoding RNAs, there are no current implications for any coexpression with eQTL linked mRNAs, ontology, or trait associations for AC106799.2 (fantom.gsc.riken.jp/cat). Variants in FCGR2A and particularly the nonsynonymous SNP rs1801274 (H131R) have previously been analyzed as candidates for HIT, however, with inconsistent results [24]. A non-significant result (P = 0.83) was detected for rs1801274 in our analysis in the discovery set (see supplementary Table S8). Moreover, several additional SNPs in FCGR2A and other previously postulated HIT candidate genes were analyzed, none of which yielded P values < 1 × 10−4 in the discovery set and were thus not further pursued (see supplementary Table S8). Recently, a GWAS and candidate study for HIT was published by Karnes et al. [25] using a comparable number of HIT cases for association analysis in a similar two-stage approach. The authors reported a locus on chromosome 14 near TDAG8 comprising several significantly HIT-associated SNPs with a genome-wide significance level when using a recessive model [25]. However, we were not able to confirm these findings in our study. Since Karnes et al. selected the HIT samples regardless of ethnicity, we cannot rule out that the two studies are not comparable as our study was strictly restricted to individuals of Caucasian descent. Also, the reported associations for

SNPs in the candidate gene HLA-DRA [25] were not associated with HIT in our analysis either, when applying a significance threshold of P ≤ 1 × 10−4. Nevertheless, other HLA-DR variants represented on the array show a weak association in comparison to other candidate gene SNPs and might be interesting targets for a candidate gene approach in the future (see supplementary Table S8). We next performed targeted NGS allowing for the detection of rare, private mutations and novel variants in HIT-associated regions implicated through our initial GWAS. In this analysis, we were initially not able to confirm our association signal for rs1433265, or any other GWAS locus for that matter, likely due to power constraints as rare variants are simply too infrequent for a conventional association analysis. However, when testing the impact of rare variants in a gene-based method utilizing the C-alpha test statistic, we identified two new candidate genes for the development of HIT: DDR1 and MCTP2. Both genes were not previously connected to platelet biology or thrombosis but bear several interesting missense mutations. Future functional studies are warranted to elucidate their potential roles. Despite its marginal significant association signal of the combined rare variants for ADAMTS16 in the C-alpha test statistic, we are confident that the initial genetic association of the chromosome 5 locus with HIT is a true finding but requires work-up to elucidate the functional consequences of several missense variants discovered in ADAMTS16 and a second gene, ICE1. Particularly, ADAMTS16 is a promising new candidate due to the detected deleterious mutations and its phylogenetic relationships with other ADAMTS genes implicated in thromboembolic complications such as TTP and pediatric (thromboembolic) stroke [62]. Taken together, our study does not support a role for previously implicated candidate genes, e.g., in FCGR2A or the findings of a recently reported GWAS, in the genetic predisposition to HIT. Nevertheless, our findings on chromosome 5 and new candidate genes such as ADAMTS16 provide a basis for future studies aiming to identify and characterize genetic susceptibility factors for HIT. The addition of genetic markers into HIT scoring models could improve their power in establishing the diagnosis in the future. Moreover, the characterization of the molecular basis and role of the corresponding candidate genes contributing to the genetic predisposition for HIT may lead to novel insights into the mechanisms of the syndrome. A better handling of heparin therapy by avoiding exposure in patients at increased risk for HIT is of major clinical interest, since HIT is still a significant safety concern despite the indisputable value of UFH and LMWH for both prophylaxis and treatment of thrombotic disorders [4, 7, 26]. Acknowledgments The authors appreciate the technical assistance of Tanja Bauer, Marianne Jansen-Rust, Petra Pietsch, Silke Pollack, and

J Mol Med Karen Böhme. We thank all participating hospitals for contributing cases and controls to this study and all patients who participated. Funding The Berlin Pharmacovigilance Center Study was funded by the Federal Institute for Drugs and Medical Devices in Bonn, Germany, grant number V-5238/68502-68605.

References 1.

2.

3. 4. 5.

6.

7.

8.

9.

10. 11.

12.

13.

14.

15.

Warkentin TE, Levine MN, Hirsh J, Horsewood P, Roberts RS, Gent M, Kelton JG (1995) Heparin-induced thrombocytopenia in patients treated with low-molecular-weight heparin or unfractionated heparin. N Engl J Med 332(20):1330–1335 Ortel TL (2009) Heparin-induced thrombocytopenia: when a low platelet count is a mandate for anticoagulation. Hematol Am Soc Hematol Educ Program:225–232 Greinacher A (2015) Heparin-induced thrombocytopenia. N Engl J Med 373(19):1883–1884 Arepally GM (2017) Heparin-induced thrombocytopenia. Blood 129(21):2864–2872 Krauel K, Potschke C, Weber C, Kessler W, Furll B, Ittermann T, Maier S, Hammerschmidt S, Broker BM, Greinacher A (2011) Platelet factor 4 binds to bacteria-inducing antibodies crossreacting with the major antigen in heparin-induced thrombocytopenia. Blood 117(4):1370–1378 Kelton JG, Warkentin TE, Moore JC, Arnold DM, Nazi I, Arepally GM, Roach JM, Fier I (2012) A prospective study measuring the development of antibodies against platelet factor 4-heparin in healthy males after exposure to heparins. J Thromb Haemost 10(7):1446–1449 Junqueira DR, Zorzela LM, Perini E (2017) Unfractionated heparin versus low molecular weight heparins for avoiding heparin-induced thrombocytopenia in postoperative patients. Cochrane Database Syst Rev 4:Cd007557 Girolami B, Prandoni P, Stefani PM, Tanduo C, Sabbion P, Eichler P, Ramon R, Baggio G, Fabris F, Girolami A (2003) The incidence of heparin-induced thrombocytopenia in hospitalized medical patients treated with subcutaneous unfractionated heparin: a prospective cohort study. Blood 101(8):2955–2959 Prandoni P, Siragusa S, Girolami B, Fabris F (2005) The incidence of heparin-induced thrombocytopenia in medical patients treated with low-molecular-weight heparin: a prospective cohort study. Blood 106(9):3049–3054 Girolami B, Girolami A (2006) Heparin-induced thrombocytopenia: a review. Semin Thromb Hemost 32(8):803–809 Kato S, Takahashi K, Ayabe K, Samad R, Fukaya E, Friedmann P, Varma M, Bergmann SR (2011) Heparin-induced thrombocytopenia: analysis of risk factors in medical inpatients. Br J Haematol 154(3):373–377 Lubenow N, Hinz P, Thomaschewski S, Lietz T, Vogler M, Ladwig A, Junger M, Nauck M, Schellong S, Wander K, Engel G, Ekkernkamp A, Greinacher A (2010) The severity of trauma determines the immune response to PF4/heparin and the frequency of heparin-induced thrombocytopenia. Blood 115(9):1797–1803 Warkentin TE, Sheppard JA, Sigouin CS, Kohlmann T, Eichler P, Greinacher A (2006) Gender imbalance and risk factor interactions in heparin-induced thrombocytopenia. Blood 108(9):2937–2941 Kosfeld RE, Lansing AM, Masri Z, Liu YK (1985) Heparininduced thrombocytopenia and recurrent thromboembolism in siblings. Am J Hematol 18(4):421–423 Lee DH, Warkentin TE, Denomme GA, Lagrotteria DD, Kelton JG (1998) Factor V Leiden and thrombotic complications in heparininduced thrombocytopenia. Thromb Haemost 79(1):50–53

16.

Carlsson LE, Lubenow N, Blumentritt C, Kempf R, Papenberg S, Schroder W, Eichler P, Herrmann FH, Santoso S, Greinacher A (2003) Platelet receptor and clotting factor polymorphisms as genetic risk factors for thromboembolic complications in heparininduced thrombocytopenia. Pharmacogenetics 13(5):253–258 17. Horsewood P, Kelton JG (2000) Investigation of a platelet factor 4 polymorphism on the immune response in patients with heparininduced thrombocytopenia. Platelets 11(1):23–27 18. Harris K, Nguyen P, Van Cott EM (2008) Platelet PlA2 polymorphism and the risk for thrombosis in heparin-induced thrombocytopenia. Am J Clin Pathol 129(2):282–286 19. Pouplard C, Cornillet-Lefebvre P, Attaoua R, Leroux D, LecocqLafon C, Rollin J, Grigorescu F, Nguyen P, Gruel Y (2012) Interleukin-10 promoter microsatellite polymorphisms influence the immune response to heparin and the risk of heparin-induced thrombocytopenia. Thromb Res 129(4):465–469 20. Rollin J, Pouplard C, Gratacap MP, Leroux D, May MA, Aupart M, Gouilleux-Gruart V, Payrastre B, Gruel Y (2012) Polymorphisms of protein tyrosine phosphatase CD148 influence FcgammaRIIAdependent platelet activation and the risk of heparin-induced thrombocytopenia. Blood 120(6):1309–1316 21. Rollin J, Pouplard C, Leroux D, May MA, Gruel Y (2013) Impact of polymorphisms affecting the ACP1 gene on levels of antibodies against platelet factor 4-heparin complexes. J Thromb Haemost 11(8):1609–1611 22. Pamela S, Anna ML, Elena D, Giovanni M, Emanuele A, Silvia V, Carmen B, Andreas G, Fabrizio F (2013) Heparin-induced thrombocytopenia: the role of platelets genetic polymorphisms. Platelets 24(5):362–368 23. Slavik L, Svobodova G, Ulehlova J, Krcova V, Hlusi A, Prochazkova J, Hutyra M (2015) Polymorphism of the Fcgamma receptor II as a possible predisposing factor for heparin-induced thrombocytopenia. Clin Lab 61(8):1027–1032 24. Rollin J, Pouplard C, Gruel Y (2016) Risk factors for heparininduced thrombocytopenia: focus on Fcgamma receptors. Thromb Haemost 116(5):799–805 25. Karnes JH, Cronin RM, Rollin J, Teumer A, Pouplard C, Shaffer CM, Blanquicett C, Bowton EA, Cowan JD, Mosley JD, Van Driest SL, Weeke PE, Wells QS, Bakchoul T, Denny JC, Greinacher A, Gruel Y, Roden DM (2015) A genome-wide association study of heparin-induced thrombocytopenia using an electronic medical record. Thromb Haemost 113(4):772–781 26. Linkins LA, Dans AL, Moores LK, Bona R, Davidson BL, Schulman S, Crowther M (2012) Treatment and prevention of heparin-induced thrombocytopenia: Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest 141(2 Suppl):e495S–e530S 27. Garbe E, Andersohn F, Bronder E, Salama A, Klimpel A, Thomae M, Schrezenmeier H, Hildebrandt M, Spath-Schwalbe E, Gruneisen A, Meyer O, Kurtal H (2012) Drug-induced immune thrombocytopaenia: results from the Berlin Case-Control Surveillance Study. Eur J Clin Pharmacol 68(5):821–832 28. Bolbrinker J, Garbe E, Douros A, Huber M, Bronder E, Klimpel A, Andersohn F, Meyer O, Salama A, Kreutz R (2017) Immobilization and high platelet count are associated with thromboembolic complications in heparin-induced thrombocytopenia. Pharmacoepidemiol Drug Saf 26(10):1149–1155 29. Greinacher A, Michels I, Kiefel V, Mueller-Eckhardt C (1991) A rapid and sensitive test for diagnosing heparin-associated thrombocytopenia. Thromb Haemost 66(6):734–736 30. Meyer O, Salama A, Pittet N, Schwind P (1999) Rapid detection of heparin-induced platelet antibodies with particle gel immunoassay (ID-HPF4). Lancet 354(9189):1525–1526 31. Warkentin TE (2003) Heparin-induced thrombocytopenia: pathogenesis and management. Br J Haematol 121(4):535–555

J Mol Med 32.

33.

34.

35.

36.

37.

38.

Cuker A, Arepally G, Crowther MA, Rice L, Datko F, Hook K, Propert KJ, Kuter DJ, Ortel TL, Konkle BA, Cines DB (2010) The HIT expert probability (HEP) score: a novel pre-test probability model for heparin-induced thrombocytopenia based on broad expert opinion. J Thromb Haemost 8:2642–2650 Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, Maller J, Sklar P, de Bakker PI, Daly MJ, Sham PC (2007) PLINK: a tool set for whole-genome association and populationbased linkage analyses. Am J Hum Genet 81(3):559–575 Pruim RJ, Welch RP, Sanna S, Teslovich TM, Chines PS, Gliedt TP, Boehnke M, Abecasis GR, Willer CJ (2010) LocusZoom: regional visualization of genome-wide association scan results. Bioinformatics (Oxford, England) 26(18):2336–2337 Frazer KA, Ballinger DG, Cox DR, Hinds DA, Stuve LL, Gibbs RA, Belmont JW, Boudreau A, Hardenbol P, Leal SM, Pasternak S, Wheeler DA, Willis TD, Yu F, Yang H, Zeng C, Gao Y, Hu H, Hu W, Li C, Lin W, Liu S, Pan H, Tang X, Wang J, Wang W, Yu J, Zhang B, Zhang Q, Zhao H, Zhao H, Zhou J, Gabriel SB, Barry R, Blumenstiel B, Camargo A, Defelice M, Faggart M, Goyette M, Gupta S, Moore J, Nguyen H, Onofrio RC, Parkin M, Roy J, Stahl E, Winchester E, Ziaugra L, Altshuler D, Shen Y, Yao Z, Huang W, Chu X, He Y, Jin L, Liu Y, Shen Y, Sun W, Wang H, Wang Y, Wang Y, Xiong X, Xu L, Waye MM, Tsui SK, Xue H, Wong JT, Galver LM, Fan JB, Gunderson K, Murray SS, Oliphant AR, Chee MS, Montpetit A, Chagnon F, Ferretti V, Leboeuf M, Olivier JF, Phillips MS, Roumy S, Sallee C, Verner A, Hudson TJ, Kwok PY, Cai D, Koboldt DC, Miller RD, Pawlikowska L, Taillon-Miller P, Xiao M, Tsui LC, Mak W, Song YQ, Tam PK, Nakamura Y, Kawaguchi T, Kitamoto T, Morizono T, Nagashima A, Ohnishi Y, Sekine A, Tanaka T, Tsunoda T, Deloukas P, Bird CP, Delgado M, Dermitzakis ET, Gwilliam R, Hunt S, Morrison J, Powell D, Stranger BE, Whittaker P, Bentley DR, Daly MJ, de Bakker PI, Barrett J, Chretien YR, Maller J, McCarroll S, Patterson N, Pe'er I, Price A, Purcell S, Richter DJ, Sabeti P, Saxena R, Schaffner SF, Sham PC, Varilly P, Altshuler D, Stein LD, Krishnan L, Smith AV, Tello-Ruiz MK, Thorisson GA, Chakravarti A, Chen PE, Cutler DJ, Kashuk CS, Lin S, Abecasis GR, Guan W, Li Y, Munro HM, Qin ZS, Thomas DJ, McVean G, Auton A, Bottolo L, Cardin N, Eyheramendy S, Freeman C, Marchini J, Myers S, Spencer C, Stephens M, Donnelly P, Cardon LR, Clarke G, Evans DM, Morris AP, Weir BS, Tsunoda T, Mullikin JC, Sherry ST, Feolo M, Skol A, Zhang H, Zeng C, Zhao H, Matsuda I, Fukushima Y, Macer DR, Suda E, Rotimi CN, Adebamowo CA, Ajayi I, Aniagwu T, Marshall PA, Nkwodimmah C, Royal CD, Leppert MF, Dixon M, Peiffer A, Qiu R, Kent A, Kato K, Niikawa N, Adewole IF, Knoppers BM, Foster MW, Clayton EW, Watkin J, Gibbs RA, Belmont JW, Muzny D, Nazareth L, Sodergren E, Weinstock GM, Wheeler DA, Yakub I, Gabriel SB, Onofrio RC, Richter DJ, Ziaugra L, Birren BW, Daly MJ, Altshuler D, Wilson RK, Fulton LL, Rogers J, Burton J, Carter NP, Clee CM, Griffiths M, Jones MC, McLay K, Plumb RW, Ross MT, Sims SK, Willey DL, Chen Z, Han H, Kang L, Godbout M, Wallenburg JC, L'Archeveque P, Bellemare G, Saeki K, Wang H, An D, Fu H, Li Q, Wang Z, Wang R, Holden AL, Brooks LD, McEwen JE, Guyer MS, Wang VO, Peterson JL, Shi M, Spiegel J, Sung LM, Zacharia LF, Collins FS, Kennedy K, Jamieson R, Stewart J (2007) A second generation human haplotype map of over 3.1 million SNPs. Nature 449(7164):851–861 Barrett JC, Fry B, Maller J, Daly MJ (2005) Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics (Oxford, England) 21(2):263–265 Delaneau O, Marchini J, Zagury JF (2012) A linear complexity phasing method for thousands of genomes. Nat Methods 9(2): 179–181 Howie BN, Donnelly P, Marchini J (2009) A flexible and accurate genotype imputation method for the next generation of genomewide association studies. PLoS Genet 5(6):e1000529

39.

Marchini J, Howie B, Myers S, McVean G, Donnelly P (2007) A new multipoint method for genome-wide association studies by imputation of genotypes. Nat Genet 39(7):906–913 40. Hiersche M, Ruhle F, Stoll M (2013) Postgwas: advanced GWAS interpretation in R. PLoS One 8(8):e71775 41. Li H, Durbin R (2009) Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics (Oxford, England) 25(14):1754–1760 42. McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, Garimella K, Altshuler D, Gabriel S, Daly M, DePristo MA (2010) The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res 20(9):1297–1303 43. Cingolani P, Patel VM, Coon M, Nguyen T, Land SJ, Ruden DM, Lu X (2012) Using Drosophila melanogaster as a model for genotoxic chemical mutational studies with a new program, SnpSift. Front Genet 3:35. https://doi.org/10.3389/fgene.2012. 00035 44. Liu X, Wu C, Li C, Boerwinkle E (2016) dbNSFP v3.0: a one-stop database of functional predictions and annotations for human nonsynonymous and splice-site SNVs. Hum Mutat 37(3):235–241 45. Neale BM, Rivas MA, Voight BF, Altshuler D, Devlin B, OrhoMelander M, Kathiresan S, Purcell SM, Roeder K, Daly MJ (2011) Testing for an unusual distribution of rare variants. PLoS Genet 7(3):e1001322 46. Cuker A (2014) Clinical and laboratory diagnosis of heparininduced thrombocytopenia: an integrated approach. Semin Thromb Hemost 40(1):106–114 47. Tan CW, Ward CM, Morel-Kopp MC (2012) Evaluating heparininduced thrombocytopenia: the old and the new. Semin Thromb Hemost 38(2):135–143 48. Lillo-Le LA, Boutouyrie P, henc-Gelas M, Le BC, Gautier I, Aiach M, Lasne D (2004) Diagnostic score for heparin-induced thrombocytopenia after cardiopulmonary bypass. J Thromb Haemost 2(11): 1882–1888 49. Uaprasert N, Chanswangphuwana C, Akkawat B, Rojnuckarin P (2013) Comparison of diagnostic performance of the heparininduced thrombocytopenia expert probability and the 4Ts score in screening for heparin-induced thrombocytopenia. Blood Coagul Fibrinolysis 24(3):261–268 50. Cuker A, Gimotty PA, Crowther MA, Warkentin TE (2012) Predictive value of the 4Ts scoring system for heparin-induced thrombocytopenia: a systematic review and meta-analysis. Blood 120(20):4160–4167 51. Nagler M, Bachmann LM, ten Cate H, ten Cate-Hoek A (2016) Diagnostic value of immunoassays for heparin-induced thrombocytopenia: a systematic review and meta-analysis. Blood 127(5): 546–557 52. Cuker A, Cines DB (2012) How I treat heparin-induced thrombocytopenia. Blood 119(10):2209–2218 53. Sun L, Gimotty PA, Lakshmanan S, Cuker A (2016) Diagnostic accuracy of rapid immunoassays for heparin-induced thrombocytopenia. A systematic review and meta-analysis. Thromb Haemost 115(5):1044–1055 54. Linkins LA, Bates SM, Lee AY, Heddle NM, Wang G, Warkentin TE (2015) Combination of 4Ts score and PF4/H-PaGIA for diagnosis and management of heparin-induced thrombocytopenia: prospective cohort study. Blood. https://doi.org/10.1182/blood-201412-618165 55. Leroux D, Hezard N, Lebreton A, Bauters A, Suchon P, de ME, Biron C, Huisse MG, Ternisien C, Voisin S, Gruel Y, Pouplard C (2014) Prospective evaluation of a rapid nanoparticle-based lateral flow immunoassay (STic Expert(®) HIT) for the diagnosis of heparin-induced thrombocytopenia. Br J Haematol 166 (5):774– 782. doi:

J Mol Med 56.

57. 58. 59.

Nellen V, Sulzer I, Barizzi G, Lammle B, Alberio L (2012) Rapid exclusion or confirmation of heparin-induced thrombocytopenia: a singlecenter experience with 1,291 patients. Haematologica 97(1):89–97 Yang JJ, Plenge RM (2011) Genomic technology applied to pharmacological traits. JAMA 306(6):652–653 Negrini S, Becquemont L (2017) Pharmacogenetics of hypersensitivity drug reactions. Therapie 72(2):231–243 Gunes MF, Akpinar MB, Comertoglu I, Akyol S, Demircelik B, Gurel OM, Aynekin B, Erdemli HK, Ates M, Eryonucu B, Demircan K (2016) The investigation of a Disintegrin and Metalloproteinase with ThromboSpondin Motifs (ADAMTS) 1, 5 and 16 in thoracic aortic aneurysms and dissections. Clin Lab 62(3): 425–433

60.

61.

62.

Li Z, Nardi MA, Li YS, Zhang W, Pan R, Dang S, Yee H, Quartermain D, Jonas S, Karpatkin S (2009) C-terminal ADAMTS-18 fragment induces oxidative platelet fragmentation, dissolves platelet aggregates, and protects against carotid artery occlusion and cerebral stroke. Blood 113(24):6051–6060 Nicholson AC, Malik SB, Logsdon JM Jr, Van Meir EG (2005) Functional evolution of ADAMTS genes: evidence from analyses of phylogeny and gene organization. BMC Evol Biol 5:11 Arning A, Hiersche M, Witten A, Kurlemann G, Kurnik K, Manner D, Stoll M, Nowak-Gottl U (2012) A genome-wide association study identifies a gene network of ADAMTS genes in the predisposition to pediatric stroke. Blood 120(26):5231–5236