Su

KLa

b

c

a

ARRAA

KDAHSUT

1

pAaDoevic

Aet

mH

h0

Journal of Pharmaceutical and Biomedical Analysis 148 (2018) 238–244

Contents lists available at ScienceDirect

Journal of Pharmaceutical and Biomedical Analysis

journa l homepage: www.e lsev ier .com/ locate / jpba

imultaneous quantification of direct oral anticoagulants currentlysed in anticoagulation therapy

athrin I. Foerster a, Andrea Huppertz a, Oliver J. Müller b, Timolaos Rizos c,isa Tilemann b, Walter E. Haefeli a, Jürgen Burhenne a,∗

Department of Clinical Pharmacology and Pharmacoepidemiology, Im Neuenheimer Feld 410, University Hospital Heidelberg, GermanyDepartment of Cardiology, Angiology and Pneumology, Im Neuenheimer Feld 410, University Hospital Heidelberg, GermanyDepartment of Neurology, Im Neuenheimer Feld 400, 69120, University Hospital Heidelberg, Germany

r t i c l e i n f o

rticle history:eceived 30 June 2017eceived in revised form 10 October 2017ccepted 12 October 2017vailable online 16 October 2017

eywords:irect oral anticoagulantnticoagulation therapyuman plasma

a b s t r a c t

Direct oral anticoagulants (DOACs) are among the most effective options to prevent serious thromboem-bolic events in patients with atrial fibrillation. Coagulation assays are used to assess DOAC activity, butlack the possibility to quantify drugs with concurrent pharmacodynamic effect. We developed a selectivemulti-drug assay to analyze apixaban, betrixaban, dabigatran, edoxaban, edoxaban M4, and rivaroxabanwith ultra-performance liquid chromatography coupled to tandem mass spectrometry (UPLC/MS/MS) inplasma fulfilling all requirements of the FDA und EMA guidelines for bioanalytical method validation.Plasma samples were extracted using solid phase extraction in a 96-well micro volume format. Chro-matographic separation was performed on a Waters BEH Phenyl 1.7 �m column coupled to tandem massspectrometry. Extraction recoveries exceeded 80 %. Concentrations of 1–1000 ng/ml can be precisely

olid phase extractionltra-performance liquid chromatographyandem mass spectrometry

quantified (correlation coefficient of >0.99) using 100 �L plasma volume. Intra-day and inter-day accu-racies ranged between 91.0 % and 116 %. Precisions at low and high concentrations were below 13.3 %.The method was applied within a clinical drug trial and eight short pharmacokinetic profiles of patientsunder DOAC therapy were analyzed. The assay allows for highly sensitive and selective simultaneousquantification of DOACs in patient plasma samples.

© 2017 Elsevier B.V. All rights reserved.

. Introduction

Atrial fibrillation is a global disease with growing incidence andrevalence [1], associated with 20–30 % of all ischemic strokes [2].nticoagulant drugs including direct oral anticoagulants (DOACs)re remarkably successful in avoiding embolic complications [3].ue to their immediate inhibitor effect, DOACs have a rapid onsetf action, which is, however, short-lasting because of terminallimination half-lives that are considerably shorter than those of

itamin K antagonists. Therefore, to warrant adequate protectiont is crucial that patients strictly adhere to therapy [4]. Adherencean be verified by concentration monitoring. Apart from adher-

Abbreviations: DOAC, direct oral anticoagulant; EMA, European Medicinesgency; FDA, Food and Drug Administration; IS, internal standard; SPE, solid phasextraction; (UP)LC/MS/MS, (ultra-performance) liquid chromatography coupled toandem mass spectrometry; QC, quality control.∗ Corresponding author at: Department of Clinical Pharmacology and Phar-acoepidemiology, Heidelberg University Hospital, Im Neuenheimer Feld 410,eidelberg 69120, Germany.

E-mail address: juergen.burhenne@med.uni-heidelberg.de (J. Burhenne).

ttps://doi.org/10.1016/j.jpba.2017.10.011731-7085/© 2017 Elsevier B.V. All rights reserved.

ence, monitoring DOAC concentrations has further advantages. Asdemonstrated for dabigatran and edoxaban, blood concentrationscorrelate well with the risk of bleeding and thromboembolism[5–7]. Monitoring can further support dose selection in complexclinical situations such as impaired renal function or polypharmacywith interacting co-medication [8,9].

Specific coagulation assays for DOAC concentration monitoringare already established in clinical practices [10], but none is specificenough to enable quantitative detection of all DOACs in a singleassay. Because these assays require calibration to the individualcompound, the specific DOAC taken by the patient must be knownto allow reliable quantification. A number of liquid chromatographymass spectrometry (LC/MS) methods for a simultaneous and selec-tive quantification of all DOACs have been reported [11–13], butthey all lack the possibility of an easily applicable, fast, and sensitivequantification. In one assay, DOAC plasma samples prepared withprotein precipitation were precisely but unconventionally quanti-

fied with high resolution orbitrap mass spectrometry [11]. Anothermethod using the more common tandem mass spectrometry lackedsensitivity to reliably quantify trough concentrations (calibrationrange: 23–750 ng/ml) [12]. A third method finally used a sensitive

al and

bfclatdsdMTFAn

2

2

bbFES(Aubp(%hpG

2

stdoshB◦

2

tC4t(viam

2

wm

K.I. Foerster et al. / Journal of Pharmaceutic

ut less common and expensive method of paramagnetic beadsor sample preparation [13]. We therefore aimed to establish oneonvenient protocol to analyze all DOACs in daily routine usingow plasma volumes, solid phase extraction in 96-well format,nd ultra-performance liquid chromatography tandem mass spec-rometry (UPLC/MS/MS) for quantification. Therefore, this paperescribes a specific and precise UPLC/MS/MS multi-drug assay toimultaneously quantify the currently approved DOACs apixaban,abigatran, edoxaban, rivaroxaban, the active metabolite edoxaban4, and the developing DOAC betrixaban in 100 �L plasma volume.

he method was validated according to current guidelines of theood and Drug Administration (FDA [14]) and European Medicinesgency (EMA [15]) and was applied to quantify short pharmacoki-etic profiles of eight patients regularly treated with a DOAC.

. Materials and methods

.1. Chemicals and reagents

13C2H8-apixaban, betrixaban, its internal standard (IS) 13C6-etrixaban, dabigatran, 13C6-dabigatran, edoxaban M4, rivaroxa-an, and 13C6-rivaroxaban were purchased from ALSACHIM (Illrich,rance). Apixaban was obtained from BIOZOL (Eching, Germany).doxaban and 2H6-edoxaban were kindly provided by Daiichi-ankyo (Munich, Germany). Acetonitrile, methanol, and waterBiosolve, Valkenswaard, Netherlands) were of UPLC/MS grade.mmonium formate (Sigma-Aldrich, Steinheim, Germany) wassed to prepare the aqueous UPLC/MS/MS eluent. Sodium acetateuffer (pH = 5.0; 0.2 M) used for solid phase extraction samplereparation was made by dissolving sodium acetate from MerckDarmstadt, Germany) in water and adjusted to pH 5.0 with 100

acetic acid from Roth (Karlsruhe, Germany). Analyte-free pooleduman plasma for validation, calibration, and quality control sam-les was supplied by the local blood bank (IKTZ, Heidelberg,ermany).

.2. Patient blood samples

Patient blood samples were collected in the course of a clinicaltudy for analytical method comparison, which was approved byhe responsible Ethics Committee of the Medical Faculty of Hei-elberg University (S-575/2015). Written informed consent wasbtained from all patients before participating in the study. Bloodamples from patients under regular DOAC therapy were takenourly over three hours after drug intake in lithium heparin tubes.lood samples were immediately centrifuged (2000g, 10 min, 20C) and plasma was stored at −25 ◦C until analysis.

.3. Calibrators, quality controls, and IS solution

Separate calibrator and internal quality control (QC) stock solu-ions were solved in acetonitrile/water and stored at −25 ◦C.alibration working solutions were prepared at 1, 5, 10, 50, 150,50, 750, and 1000 ng/ml by diluting the stock solutions with ace-onitrile/water (1:1 v/v). QC working solutions were prepared at 3QC A), 500 (QC B), and 800 ng/ml (QC C) in acetonitrile/water (1:1/v). Calibration and QC working solutions were stored at −25 ◦Cn light-protecting glass vials. Isotope-labeled IS were dissolved incetonitrile/water (1:1 v/v) and aliquots of IS stock solutions wereixed to a working solution and stored at −25 ◦C.

.4. Sample preparation

Calibrators and QCs were prepared by adding calibration or QCorking solutions (25 �L) to blank plasma. For volume adjust-ent, acetonitrile/water (1:1 v/v; 25 �L) was added to every patient

Biomedical Analysis 148 (2018) 238–244 239

plasma sample (100 �L). Calibrators, QCs, and patient samples werespiked with 150 �L acetate buffer and 25 �L IS solution; the mix-ture was vortexed (5 s), and centrifuged (16,000g, 3 min) beforesolid phase extraction (SPE).

An aliquot of the supernatant (250 �L) was transferredonto �Elution PRIME HLB 96-well cartridges (Waters, Eschborn,Germany) and slowly pushed through the cartridges using a 96-well positive pressure processor (Waters, Eschborn, Germany).Subsequently, wells were washed by adding methanol/water (5:95v/v; 200 �L). After drying with nitrogen (20 psi, 30 s) the ana-lytes were eluted with methanol (100 �L) into a 96-well samplecollection plate (Waters, Eschborn, Germany). The eluates wereevaporated to dryness under a stream of nitrogen (40◦ C, 6.5min; UltravapPorvair Sciences Limited, Norfolk, UK). Residues werereconstituted in UPLC/MS/MS eluent, vortexed, and injected (20 �L)into the UPLC/MS/MS system.

2.5. Instrumental analysis

Analysis was performed on an Acquity UPLC system coupledto a TQD triple quadrupole mass spectrometer (Waters, Eschborn,Germany). The autosampler was kept at 15 ◦C. Separation wasdone on a 2.1 x 50 mm Acquity UPLC BEH Phenyl 1.7 �m col-umn (Waters, Eschborn, Germany) at 40 ◦C. The eluents consistedof 95 % 5 mM ammonium formate in water with 5 % acetoni-trile (A) and 100 % acetonitrile (B). The UPLC gradient separationwas performed at a flow rate of 0.5 ml/min and started with100 % A/0 % B (0.5 min). The ratio was changed to 5 % A/95 %B within 3.0 min (0.5–3.5 min) and kept until 4.0 min. Withinthe next 0.5 min, the system returned to its initial conditions.The analytes were positively ionized using heated electrosprayionization (Z-spray) with ionization parameters as follows: cap-illary voltage 1 kV, cone voltage 38 V, source temperature 150◦C, cone gas flow (N2) 50 L h−1, desolvation gas flow (N2) 850 L h−1, and desolvation temperature 500 ◦C. Mass analysis was per-formed by multiple reaction monitoring (MRM) using argon ascollision gas. The ion transitions monitored were m/z 460.4 → m/z443.2 for apixaban, m/z 469.5 → m/z 452.4 for 13C2H8-apixaban, m/z452.2 → m/z 324.1 for betrixaban, m/z 458.3 → m/z 330.1 for 13C6-betrixaban, m/z 472.4 → m/z 289.1 for dabigatran, m/z 478.4 → m/z295.1 for 13C6-dabigatran, m/z 548.4 → m/z 152.1 for edoxaban,m/z 521.3 → m/z 339.2 for edoxaban M4, m/z 554.5 → m/z 158.1for 2H6-edoxaban, m/z 436.2 → m/z 144.9 for rivaroxaban, andm/z 442.3 → m/z 144.8 for 13C6-rivaroxaban. Peaks were inte-grated automatically by MassLynx 1.4 procedures. Batches withcorrelation coefficients r2 > 0.99 and accurately quantified QC con-centrations (±15 %) were accepted.

2.6. Method validation

This multi-compound assay was validated according to FDA [14]and EMA guidelines [15]. Accuracy and precision were determinedsix-fold within three analytical batches at QC concentrations 1, 3,500, and 800 ng/ml. Both parameters were evaluated within (intra)and between (inter) runs with an acceptance range of ± 15 % (exceptfor LLOQ, 1 ng/ml: ± 20 %). Selectivity was tested by analyzingsix blank human plasma samples from different sources. A pos-sible effect of hyperlipemic or hemolytic plasma on accuracy wasdetermined at 500 ng/ml. Stability tests were performed at QC lev-els (3, 500, and 800 ng/ml) in triplicate determinations. Long-termstability was tested for 1, 2, and 5 weeks at 4 ◦C and −25 ◦C. Bench-top stability (short-term stability) was assessed by analyzing QC

plasma samples stored at 20 ◦C on a lab bench for 1 and 2 weeks.Carry-over was tested by analyzing blank eluent samples injecteddirectly after the analysis of the highest calibration level. Recov-ery was determined at each QC level (3, 500, and 800 ng/ml) in

240 K.I. Foerster et al. / Journal of Pharmaceutical and Biomedical Analysis 148 (2018) 238–244

Fig. 1. UPLC-/MS/MS chromatograms of a processed plasma sample at LLOQ level (1 ng/ml) with corresponding mass transitions and intensities of each analyte.

Table 1Intra-assay and inter-assay accuracies and precisions of all DOACs.

Apixaban Betrixaban Dabigatran

LLOQ QC A QC B QC C LLOQ QC A QC B QC C LLOQ QC A QC B QC C1 3 500 800 1 3 500 800 1 3 500 800

Within batch1 Mean [ng/ml] 1.1 3.2 521 819 1.1 3.1 509 811 1.0 2.8 474 763

Accuracy [%] 105 106 104 102 114 104 102 101 103 91.5 94.8 95.4Precision [%] 1.7 2.0 0.8 0.9 0.6 0.5 1.2 1.0 3.1 2.2 0.8 0.6

2 Mean [ng/ml] 1.1 3.3 531 842 1.2 3.1 517 823 1.1 2.9 484 775Accuracy [%] 111 109 106 105 115 105 103 103 107 96.8 96.9 96.8Precision [%] 2.0 2.1 4.3 0.4 2.0 1.6 0.9 0.7 6.6 2.6 0.8 0.6

3 Mean [ng/ml] 1.1 3.2 531 843 1.2 3.1 519 827 1.1 2.9 486 779Accuracy [%] 113 107 106 105 116 103 104 103 112 95.9 97.2 97.4Precision [%] 2.5 1.4 1.4 1.2 2.3 1.0 0.7 0.6 3.4 2.7 0.7 1.3

Between batchMean [ng/ml] 1.1 3.2 527 834 1.2 3.1 514 820 1.1 2.8 481 772Accuracy [%] 110 107 105 104 115 104 103 103 107 94.7 96.2 96.5Precision [%] 3.8 2.0 1.3 1.6 1.8 1.3 1.2 1.1 5.8 3.4 1.4 1.3

Edoxaban Edoxaban M4 Rivaroxaban

LLOQ QC A QC B QC C LLOQ QC A QC B QC C LLOQ QC A QC B QC C1 3 500 800 1 3 500 800 1 3 500 800

Within batch1 Mean [ng/ml] 1.1 3.3 528 839 1.0 2.7 517 794 1.1 3.2 562 893

Accuracy [%] 112 111 106 105 97.8 91.0 103 99.2 107 106 112 112Precision [%] 5.0 2.3 0.8 0.5 4.6 3.5 1.6 1.2 10.0 3.2 1.3 0.8

2 Mean [ng/ml] 1.1 3.1 530 861 1.0 3.0 525 807 1.0 3.3 545 897Accuracy [%] 106 103 106 108 104 101 105 101 100 109 109 112Precision [%] 7.7 8.9 2.1 1.4 7.4 1.3 1.1 1.9 2.6 4.9 2.2 1.0

3 Mean [ng/ml] 1.0 3.1 539 859 1.0 2.8 497 772 0.9 3.1 563 903Accuracy [%] 98.9 105 108 107 100 93.7 99.3 96.5 91.6 102 113 113Precision [%] 10.2 5.6 0.8 1.6 8.5 5.8 0.9 1.3 13.3 3.8 0.9 1.0

Between batch

tsw

Mean [ng/ml] 1.1 3.2 532 853 1.0Accuracy [%] 106 106 106 107 101Precision [%] 8.8 6.5 1.6 1.7 7.2

riplicate determinations by comparing peak areas from regularlypiked QCs with peak areas obtained from blank plasma spikedith QCs after sample preparation (representing a 100 % value).

2.9 514 791 1.0 3.2 557 89895.2 103 98.8 99.5 106 111 1125.8 2.6 2.4 8.5 3.9 1.4 0.9

Matrix effects were analyzed for all analytes and IS in triplicatedeterminations at each QC level (3, 500, and 800 ng/ml). Peak areasof blank plasma spiked with analyte and IS after sample preparation

K.I. Foerster et al. / Journal of Pharmaceutical and Biomedical Analysis 148 (2018) 238–244 241

Fig. 2. UPLC/MS/MS chromatograms of plasma samples at 500 ng/ml (processed QC B sample) with corresponding mass transitions and intensities of each analyte.

Table 2Accuracies during stability tests of all DOACs.

1 week 2 weeks 5 weeks

QC A [%] QC B [%] QC C [%] QC A [%] QC B [%] QC C [%] QC A [%] QC B [%] QC C [%]

−25 ◦CApixaban 1.0 −0.3 1.8 2.1 1.5 0.7 −3.7 −1.1 −2.3Betrixaban −0.9 −2.5 2.2 0.3 −1.1 −1.4 −2.1 −1.7 −0.2Dabigatran −11.2 −11.0 −8.7 −9.3 −10.1 −10.4 −7.0 −6.3 −11.3Edoxaban 0.9 −1.0 −1.8 −2.4 −6.8 −0.8 −4.0 −4.3 −3.7Edoxaban M4 −11.7 −10.1 −5.5 −11.3 −6.6 −12.0 −12.7 −12.4 −11.3Rivaroxaban −5.8 −3.9 −1.8 −5.6 −3.5 −4.6 −2.0 −5.0 0.2+4 ◦CApixaban 8.0 7.6 5.4 −2.8 3.9 5.0 −3.8 −1.1 −4.0Betrixaban 1.6 −1.5 −3.4 0.8 −0.1 −1.9 0.1 −1.5 −3.1Dabigatran −9.3 −7.8 −9.8 −11.7 −8.3 −10.3 −3.7 −7.1 −7.8Edoxaban 0.3 1.0 −0.8 −8.7 −9.1 −9.6 −9.7 −12.6 −14.2Edoxaban M4 −21.2 −15.2 −19.6 −21.0 −21.1 −24.0 −27.8 −24.3 −25.4Rivaroxaban −5.6 −4.9 −4.7 −4.7 −4.6 −5.5 −3.6 −2.3 −2.9+ 20 ◦CApixaban 7.8 4.6 5.8 −5.8 1.1 1.4Betrixaban 3.3 −1.6 −2.1 −7.2 −7.3 −4.9Dabigatran −8.1 −11.3 −12.6 −13.3 −9.9 −7.5

wcNm

3

3

B0ia

Edoxaban −6.0 −8.9 −7.6 −48.6

Edoxaban M4 −24.7 −21.9 −23.2 −33.9

Rivaroxaban −4.3 −5.6 −3.9 −6.8

ere compared with peak areas gained from pure eluent solutionontaining all analytes and IS at the respective QC concentrations.ormalized matrix factors were calculated by dividing the analytes’atrix effects by IS’ matrix effects.

. Results and discussion

.1. Chromatography and mass spectrometry

After preliminary tests with different columns (Acquity UPLC

EH C18, CSH C18, Waters, Eschborn, Germany) and eluents (e.g..01 % aqueous formic acid solution), the best separation and most

ntense [M+H]+ ions were achieved on a phenyl UPLC column with gradient program composed of an aqueous 5 mM ammonium

−40.0 −44.4−37.3 −43.1−7.5 −6.3

formate solution and acetonitrile. Fast and sharp peaks with peakwidth at baseline of 3–6 s depending on the analyte were achieved(Figs. 1–3). With a cycle time of 100 ms our method resulted in atleast 20 data points per peak.

3.2. Sample preparation

A routine monitoring assay requiring only minimal plasma vol-ume (100 �L) and offering the possibility of automation is certainlyfavorable. Different from an earlier method using paramagnetic

micro-particles for sample extraction [13], we chose �Elution SPEwhich is more convenient and can easily be scaled up to automa-tion. Triplicate determinations of recovery rates for all analytesand their internal standards at QC A to C level showed that the

242 K.I. Foerster et al. / Journal of Pharmaceutical and Biomedical Analysis 148 (2018) 238–244

Fig. 3. UPLC-/MS/MS chromatograms of selected patient plasma samples with apixaban (calculated concentration 1.9 ng/ml), dabigatran (190 ng/ml), edoxaban (140 ng/ml),edoxaban M4 (4.5 ng/ml), and rivaroxaban (230 ng/ml).

F aban (( = 2, pap was p

utb%f

ig. 4. Pharmacokinetic profiles of patients regularly treated with 2 × 5 mg/d apixn = 2, patient 1 depicted in squares, patient 2 in circular), 1 × 60 mg/d edoxaban (n

atient 1 depicted in squares, patient 2 in circular). In one patient no edoxaban M4

sed hydrophilic-lipophilic material retained all analytes effec-

ively. Good recovery rates from plasma were achieved and rangedetween 85.5 % to 93.1 % for apixaban and 13C2H8-apixaban, 109.7

to 120.1 % for betrixaban and 13C6-betrixaban, 80.6 % to 84.8 %or dabigatran and 13C6-dabigatran, 98.9 % to 100.9 % for edoxaban

n = 2, patient 1 depicted in squares, patient 2 in circular), 2 × 110 mg/d dabigatrantient 1 depicted in squares, patient 2 in circular), or 1 × 20 mg/d rivaroxaban (n = 2,resent at baseline.

and 2H6-edoxaban, 83.9 % to 86.8 % for edoxaban M4, and 101.613

% to 104.2 % for rivaroxaban and C6-rivaroxaban. Results from

matrix effect investigations revealed that the used sample prepara-tion protocol effectively removed co-eluting biological compounds.Acceptable ion suppression was present for betrixaban (87.9 %)

al and

wm%im0be%tac

3

Usrc0MCAar1p[ofrfdarihspShamwwbptfps

3

tsds[lp

K.I. Foerster et al. / Journal of Pharmaceutic

hereas all other analytes were subject to slight ion enhance-ent (apixaban: 112.0 %, dabigatran: 101.9 %, edoxaban: 114.2

, and rivaroxaban: 108.3 %). With the use of an isotope-labelednternal standard this effect is controlled as shown in normalized

atrix factor values ranging between 0.94 (apixaban, edoxaban),.98 (dabigatran), 1.01 (betrixaban), and 1.02 (rivaroxaban). Edoxa-an M4 showed reproducible, but concentration-dependent matrixffects. At QC A (3 ng/ml) ion enhancement was present (124.9), whereas at higher concentrations (500 and 800 ng/ml) ioniza-ion was suppressed (86.0 %). These varying matrix effects do notffect the quantification of edoxaban M4, as proven by valid qualityontrols analyzed within each batch.

.3. Method validation

The combined protocol using solid phase extraction andPLC/MS/MS resulted in a method fulfilling all FDA and EMA

tandards for bioanalytical method validation [14,15]. Calibrationanging between 1 and 1000 ng/ml resulted in good correlationoefficients (r2) of ≥ 0.9988 (apixaban), ≥ 0.9991 (betrixaban), ≥.9975 (dabigatran), ≥ 0.9983 (edoxaban), ≥ 0.9968 (edoxaban4), and ≥ 0.9935 (rivaroxaban). Inter-assay accuracies (QC A-

) ranged from 91.0 % to 113 % and intra-assay accuracies (QC-C) varied between 94.7 % and 112 %. A maximum inter-assaynd intra-assay precision deviation of 8.9 % and 6.5 % was presentespectively. Accuracies at LLOQ level were between 91.6 % and16 % for all analytes (Table 1). Similar to other multi-drug assaysublished before, DOACs can be analyzed down to the ng/ml range11–13]. With a maximum precision deviation of 13.3 % (Table 1),ur assay allows a precise quantification of at least 1 ng/ml (LLOQ)or apixaban, betrixaban, dabigatran, edoxaban, edoxaban M4, andivaroxaban. Selectivity was confirmed by the absence of inter-ering peaks in six different lots of pooled human blank plasmaonated by healthy volunteers. Samples spiked into hyperlipemicnd hemolytic plasma showed acceptable accuracies for all DOACsanging between 86.2-105 % and 94.0-104 %, respectively, prov-ng that the assay performance is also valid for hyperlipemic andemolytic plasma samples. Chromatograms of a spiked plasmaample at LLOQ level (Fig. 1), at QC B level (Fig. 2), and a humanlasma sample for each DOAC (Fig. 3) are depicted in Figs. 1–3.tability was tested at lower (3 ng/ml), medium (500 ng/ml), andigh (800 ng/ml) QC concentration and revealed good long-termnd short-term stability for all analytes except edoxaban and itsetabolite. In accordance with earlier publications [11,13], bothere unstable when stored at temperatures above −25 ◦C for 5eeks (Table 2). If the analyte is unknown and samples must

e stored for more than one week, we thus recommend freezinglasma DOAC samples at ≤−25 ◦C. Likely due to its high responseo the TQD mass spectrometer, slight carry-over was observedor betrixaban, which makes it necessary to inject a blank sam-le between patient samples to reduce carry-over. Other DOACshowed no carry-over.

.4. Method application to patient samples

Short pharmacokinetic profiles in eight patients regularlyreated with apixaban, dabigatran, edoxaban, and rivaroxaban arehown in Fig. 4. Therapeutic ranges for DOACs have not beenefined by any manufacturer, but plasma concentrations under

teady state conditions have been reported from Phase III trials5,7,16,17]. The quantified concentrations are comparable to pub-ished values, suggesting that our method qualifies to monitoratient plasma samples.

Biomedical Analysis 148 (2018) 238–244 243

3.5. Limitations

With focus on future prospects, if DOAC monitoring becomesa routine analysis, an external quality control system with realpatient samples should be introduced. An external quality controlsystem (round robin tests) would guarantee that every laboratoryreliably quantifies DOACs in patient samples for a better compa-rability between the laboratories. A limitation of this assay is thatthe turnaround time is about two hours from sample receipt toresult reporting and it lacks to quantify the active metabolites ofdabigatran (dabigatran acylglucuronides).

4. Conclusions

Multi-drug assays offer the possibility to analyze different DOACsamples in parallel, which saves time and cost. So far this is anadvantage of mass spectrometry, since coagulation assays do notdistinguish between drugs revealing identical pharmacodynamiceffects and have to be calibrated to the respective compound mak-ing them difficult to use in unconscious patients with unknowndrug histories. A number of mass spectrometry methods for thequantification of DOACs have been published in the past, but onlyfew were multi-drug assays [11–13] and to our knowledge noneincludes betrixaban, which was recently FDA approved and is cur-rently under EMA investigation for market approval. By using�Elution SPE, a LLOQ of 1 ng/ml was reached, which allows reli-able quantification of plasma trough concentrations present undertherapeutic DOAC maintenance doses. Furthermore, �Elution SPEis convenient and can be scaled up to automated SPE methods,which makes our method of advantage if it should be establishedfor routine measurements.

Conflict of interest statement

JB, KIF, AH, and LT, report no financial or personal relation-ships that could potentially be perceived as influencing the researchdescribed herein.

OJM, TR, and WEH received consulting honoraria, speaker fees,or travel support from BMS, Boehringer Ingelheim, Bayer Health-Care, Daiichi Sankyo, and Pfizer, outside the submitted work.

This study was supported in part by a grant from Daiichi-Sankyo(Munich, Germany).

Acknowledgements

The study was supported in part by a grant from Daiichi-Sankyo(Munich, Germany). Daiichi-Sankyo provided the reference stan-dards of edoxaban and 2H6-edoxaban. The authors would like tothank Andrea Deschlmayr, Kevin Jansen, Magdalena Longo, and Dr.Lukas Witt for their valuable assistance during assay development.

References

[1] S.S. Chugh, R. Havmoeller, K. Narayanan, D. Singh, M. Rienstra, E.J. Benjamin,R.F. Gillum, Y.H. Kim, J.H. McAnulty Jr., Z.J. Zheng, M.H. Forouzanfar, M.Naghavi, G.A. Mensah, M. Ezzati, C.J. Murray, Worldwide epidemiology ofatrial fibrillation: a global burden ofdisease 2010 study, Circulation 129(2014) 837–847.

[2] P. Kirchhof, S. Benussi, D. Kotecha, A. Ahlsson, D. Atar, B. Casadei, M. Castella,H.C. Diener, H. Heidbuchel, J. Hendriks, G. Hindricks, A.S. Manolis, J. Oldgren,B.A. Popescu, U. Schotten, B. van Putte, P. Vardas, S. Agewall, J. Camm, G.Baron Esquivias, W. Budts, S. Carerj, F. Casselman, A. Coca, R. de Caterina, S.Deftereos, D. Dobrev, J.M. Ferro, G. Filippatos, D. Fitzsimons, B. Gorenek, M.Guenoun, S.H. Hohnloser, P. Kolh, G.Y. Lip, A. Manolis, J. McMurray, P.

Ponikowski, R. Rosenhek, F. Ruschitzka, I. Savelieva, S. Sharma, P. Suwalski, J.L.Tamargo, C.J. Taylor, I.C. van Gelder, A.A. Voors, S. Windecker, J.L. Zamorano,K. Zeppenfeld, 2016 ESC guidelines for the management of atrial fibrillationdeveloped in collaboration with EACTS, Eur. J. Cardiothorac. Surg. 50 (2016)e1–e88.

2 al and

[

[

[

[

[

[

[

44 K.I. Foerster et al. / Journal of Pharmaceutic

[3] C.T. Ruff, R.P. Giugliano, E. Braunwald, E.B. Hoffman, N. Deenadayalu, M.D.Ezekowitz, A.J. Camm, J.I. Weitz, B.S. Lewis, A. Parkhomenko, T. Yamashita,E.M. Antman, Comparison of the efficacy and safety of new oralanticoagulants with warfarin in patients with atrial fibrillation: ameta-analysis of randomised trials, Lancet 383 (2014) 955–962.

[4] S. Shore, E.P. Carey, M.P. Turakhia, C.A. Jackevicius, F. Cunningham, L. Pilote,S.M. Bradley, T.M. Maddox, G.K. Grunwald, A.E. Barón, J.S. Rumsfeld, P.D.Varosy, P.M. Schneider, L.N. Marzec, P.M. Ho, Adherence to dabigatrantherapy and longitudinal patient outcomes: insights from the veterans healthadministration, Am. Heart J. 167 (2014) 810–817.

[5] P.A. Reilly, T. Lehr, S. Haertter, S.J. Connolly, S. Yusuf, J.W. Eikelboom, M.D.Ezekowitz, G. Nehmiz, S. Wang, L. Wallentin, The effect of dabigatran plasmaconcentrations and patient characteristics on the frequency of ischemicstroke and major bleeding in atrial fibrillation patients: the RE-LY Trial(Randomized Evaluation of Long-Term Anticoagulation Therapy), J. Am. Coll.Cardiol. 63 (2014) 321–328.

[6] J.I. Weitz, S.J. Connolly, I. Patel, D. Salazar, S. Rohatagi, J. Mendell, H.Kastrissios, J. Jin, S. Kunitada, Randomised parallel-group, multicentre,multinational phase 2 study comparing edoxaban, an oral factor Xa inhibitor,with warfarin for stroke prevention in patients with atrial fibrillation,Thromb. Haemost. 104 (2010) 633–641.

[7] C.T. Ruff, R.P. Giugliano, E. Braunwald, D.A. Morrow, S.A. Murphy, J.F. Kuder, N.Deenadayalu, P. Jarolim, J. Betcher, M. Shi, K. Brown, I. Patel, M. Mercuri, E.M.Antman, Association between edoxaban dose, concentration, anti-Factor Xaactivity, and outcomes: an analysis of data from the randomised double-blindENGAGE AF-TIMI 48 trial, Lancet 385 (2015) 2288–2295.

[8] T. Baglin, D. Keeling, S. Kitchen, Effects on routine coagulation screens and

assessment of anticoagulant intensity in patients taking oral dabigatran orrivaroxaban: guidance from the British Committee for Standards inHaematology, Br. J. Haematol. 159 (2012) 427–429.

[9] T. Baglin, A. Hillarp, A. Tripodi, I. Elalamy, H. Buller, W. Ageno, Measuring OralDirect Inhibitors (ODIs) of thrombin and factor Xa: a recommendation from

[

Biomedical Analysis 148 (2018) 238–244

the subcommittee on control of anticoagulation of the scientific andstandardisation committee of the international society on thrombosis andhaemostasis, J. Thromb. Haemost. 11 (2013) 756–760.

10] B.J. Dale, N.C. Chan, J.W. Eikelboom, Laboratory measurement of the directoral anticoagulants, Br. J. Haematol. 172 (2016) 315–336.

11] T. Gous, L. Couchman, J.P. Patel, C. Paradzai, R. Arya, R.J. Flanagan,Measurement of the direct oral anticoagulants apixaban, dabigatran,edoxaban, and rivaroxaban in human plasma using turbulent flow liquidchromatography with high-resolution mass spectrometry, Ther. Drug Monit.36 (2014) 597–605.

12] EM, K. Schmitz, D.J. Boonen, van den Heuvel, M.W. van Dongen, J.M.Schellings, L. van der Graaf, D. van de Kerkhof, Determination of dabigatran,rivaroxaban and apixaban by ultra-performance liquid chromatography –tandem mass spectrometry (UPLC-MS/MS) and coagulation assays for therapymonitoring of novel direct oral anticoagulants, J. Thromb. Haemost. 12 (2014)1636–1646.

13] C. Wiesen, T. Blaich, G. Streichert, Paramagnetic micro-particles as a tool forrapid quantification of apixaban, dabigatran, edoxaban and rivaroxaban inhuman plasma by UHPLC-MS/MS, Clin. Chem. Lab. Med. 55 (2017) 1349–1359.

14] Food and Drug Administration, Guidance for Industry Bioanalytical MethodValidation. http://www.fda.gov/downloads/Drugs/Guidance/ucm070107.pdf,2001 (Accessed 10, October 2017).

15] European Medicines Agency, Guideline on bioanalytical method validation.http://www.ema.europa.eu/docs/en GB/document library/Scientificguideline/2011/08/WC500109686.pdf, 2011 (Accessed 10, October 2017).

16] Eliquis, Summary of product characteristics (SPC). http://www.ema.europa.eu/docs/en GB/document library/EPAR - Product Information/human/

002148/WC500107728.pdf, Feb 2017 (Accessed 10, October 2017).

17] I.G. Girgis, M.R. Patel, G.R. Peters, K.T. Moore, K.W. Mahaffey, C.C. Nessel, J.L.Halperin, R.M. Califf, K.A. Fox, R.C. Becker, Population pharmacokinetics andpharmacodynamics of rivaroxaban in patients with non-valvular atrialfibrillation: results from ROCKET AF, J. Clin. Pharmacol. 54 (2014) 917–927.