A

maeaitmswnt©

K

1

cmipt

oB

(

0d

Journal of Chromatography A, 1160 (2007) 3–12

High-throughput on-line solid-phase extraction–liquidchromatography–tandem mass spectrometry methodfor the simultaneous analysis of 14 antidepressants

and their metabolites in plasma

Ana de Castro a, Maria del Mar Ramırez Fernandez b,∗, Marleen Laloup b, Nele Samyn b,Gert De Boeck b, Michelle Wood c, Viviane Maes d, Manuel Lopez-Rivadulla a

a Forensic Toxicology Service, Institute of Legal Medicine, University of Santiago de Compostela, Santiago de Compostela, Spainb Federal Public Service Justice, National Institute of Criminalistics and Criminology, Brussels, Belgium

c Waters Corporation, MS Technologies Centre, Manchester, UKd Department of Clinical Chemistry-Toxicology, Academic Hospital, Free University of Brussels, Brussels, Belgium

Available online 8 February 2007

bstract

A rapid, sensitive and fully automated on-line solid-phase extraction–liquid chromatography–tandem mass spectrometry (SPE–LC–MS/MS)ethod was developed and validated for the direct analysis of 14 antidepressants and their metabolites in plasma. Integration of the sample extraction

nd LC separation into a single system permitted direct injection of the plasma without prior sample pre-treatment. The applied gradient ensured thelution of all the examined drugs within 14 min and produced chromatographic peaks of acceptable symmetry. The total process time was 20 minnd only 50 �L of plasma was required. Selectivity of the method was achieved by a combination of retention time and two precursor-producton transitions for the non-deuterated compounds. The use of SPE was demonstrated to be highly effective and led to significant decreases inhe interferences present in the matrix. Extraction was found to be both reproducible and efficient with recoveries >99% for all the analytes. The

ethod showed excellent intra-assay and inter-assay precision (relative standard deviation (RSD) and bias <20%) for quality control (QC) samplespiked at a concentration of 40, 200 and 800 �g/L and the r2 > 0.99 over the range investigated (10–1000 �g/L). Limits of quantification (LOQs)

ere estimated to be 10 �g/L. Furthermore, the processed samples were demonstrated to be stable for at least 48 h, except for clomipramine andorclomipramine, where a slight negative trend was observed, but did not compromise the quantification. The method was subsequently appliedo authentic samples previously screened by a routine HPLC method with diode array detection (DAD).

2007 Elsevier B.V. All rights reserved.

oci

g

eywords: Antidepressant; Plasma; On-line SPE; LC–MS/MS

. Introduction

Major depressive disorder (MDD) is a conditionharacterized by a prolonged depression of mood or by aarked loss of interest or pleasure. Depression has received

ncreased attention owing to the growing recognition of itsrevalence. For many years, the prevailing hypothesis has beenhat the condition is caused by (or associated with) a deficiency

∗ Corresponding author at: Federal Public Service Justice, National Institutef Criminalistics and Criminology, Chaussee de Vilvorde 100, 1120 Brussels,elgium. Tel.: +32 2 240 05 00; fax: +32 2 241 61 05.

E-mail address: mariadelmar.ramirezfernandez@just.fgov.beM.d.M.R. Fernandez).

waacbtrti

021-9673/$ – see front matter © 2007 Elsevier B.V. All rights reserved.oi:10.1016/j.chroma.2007.01.137

f the monoamines, notably noradrenaline and serotonin;urrent theories also acknowledge that other factors may benvolved in the pathogenesis of depression.

Pharmacological treatment for depression has advancedreatly since the development of the first therapies in the 1950s,ith the introduction of monoamine oxidase inhibitors (MAOIs)

nd tricyclic antidepressants (TCAs) [1]. Since the late 1980s,whole new generation of chemically and neuropharmacologi-ally unrelated agents have been introduced which appear toe safer and better tolerated [2]. These include: selective sero-

onin reuptake inhibitors (SSRIs), serotonin and noradrenalineeuptake inhibitors (SNaRIs), noradrenergic and specific sero-onergic antidepressants (NaSSAs) and noradrenaline reuptakenhibitors (NaRIs) [3].

4 romat

dctCTitiatctwtelemruplArpctT[

ocpgflmLfthcdchsamiha

teept

hapmbtsootcitw

2

2

iptdmd(mULsUF

((gT(

t

2

vbc

2

A. de Castro et al. / J. Ch

A shared characteristic to the majority of the antidepressantrugs is the high interindividual variability in the plasmaticoncentrations. It can be explained partly due to the exis-ence of polymorphism in the genes that encode some of theYP450 isoenzymes implicated in their metabolization [4,5].hese genetic differences have been well characterized mainly

n the case of CYP2D6 and CYP2C19, and some studies abouthe recommendations on individual dose adjustment regard-ng it have been carried out [6]. These clinical consequencesre more dramatic in the case of TCA since they are charac-erized by having a narrow therapeutic window with risk ofardiac and CNS toxicity [7,8]. Thus, therapeutic drug moni-oring (TDM) for this class of antidepressant agents has becomeidely used because of its efficacy, safety and cost-effectiveness,

he novel antidepressants show a less predictable concentration-fficacy relationship [9–12], in combination with a relativelyow toxicity. Therefore, several authors also support the ben-fit of TDM for these drugs in special situations, such as toonitor compliance in the case of those patients who do not

espond to an apparently adequate dose to identify those individ-als who have particularly slow or rapid drug clearance, duringharmacokinetic studies and when administered to special popu-ations (i.e., elderly or people with organic diseases) [10,13–15].lthough, there are some discrepancies about the therapeutic

ange for antidepressants, mainly the new generations, pro-osed therapeutic concentrations have been determined. Theyan be found in the list published by Winek et al., as well ashe one elaborated by some recognised organizations such asIAFT (The International Association of Forensic Toxicologist)

16,17].Several methods have been published for the determination

f one or more antidepressants in different biological matri-es such as plasma or urine for monitoring or toxicologicalurposes [18–32]. In these reports, the use of gas chromato-raphy (GC) coupled to nitrogen–phosphorus (NPD) [18,19],ame ionization (FID) [20], electron-capture (ECD) [21] andass spectrometry (MS) [21–23] detection have been described.iquid–liquid or solid-phase extraction (SPE) have been used

or sample clean-up followed, in some cases, by a derivatiza-ion step [21–24]. However, in most of these reports, HPLCas been used in conjunction with UV [25–27] or fluores-ence (FL) detection [27,28], thus necessitating an appropriateerivatization step to increase the fluorescence capacities of theompounds of interest. More recently, LC–MS or LC–MS/MSave been applied [29–32]. These techniques provide a highelectivity and sensitivity in combination with a good precisionnd accuracy over a wide dynamic range, allowing the develop-ent of very rapid and efficient analytical methods. Therefore,

n many cases it is now the sample pre-treatment process thatas become the bottleneck in method development and samplenalysis [33,34].

High-throughput analysis is becoming increasingly impor-ant in all areas of science, the forensic sciences being no

xception. Many efforts have been made to develop on-linextraction techniques, allowing automation and a high through-ut of samples [35–37]. These procedures have been applied forhe detection of only a few antidepressants.

scof

ogr. A 1160 (2007) 3–12

The aim of this study was to develop a simple, rugged andigh-throughput on-line SPE–LC–MS/MS method for rapidnd simultaneous bio-analysis of the main antidepressantsrescribed in Belgium and their metabolites in plasma. Theethod involves a fully automated SPE system (Spark Sym-

iosis Pharma), which allows the simultaneous extraction ofhe second sample in one clamp and the elution of the firstample in a second clamp, as such, achieving an optimal usef the extraction time. This system offers the entire processf conditioning, sample application, washing and elutionaking place at constant flow rates, yielding better precision inomparison with off-line driven extraction procedures. Anothermportant advantage is that no manual transfers are made andhat the whole of the eluate is loaded onto the LC columnithout the need for a pre-concentration step.

. Materials and methods

.1. Reagents

Individual stock solutions of amitriptyline, nortriptyline,mipramine, desipramine, trazodone, fluoxetine, norfluoxetine,aroxetine, fluvoxamine, sertraline (all certified at a concen-ration of 1 mg/mL in methanol), and the internal stan-ards (I.S.) [2H3]imipramine (imipramine-d3), [2H3]desipra-ine (desipramine-d3), [2H3]clomipramine (clomipramine-

3), [2H6]fluoxetine (fluoxetine-d6) and [2H6]paroxetineparoxetine-d6) (certified concentration of 0.1 mg/mL inethanol) were obtained from Cerilliant (Round Rock, TX,SA). Venlafaxine as a solid was obtained from Lederleabs. (New York, USA). Norclomipramine and citalopram inolid form were supplied by Sigma–Aldrich (St. Louis, MO,SA) and clomipramine as a solid was a gift from Novartisarmaceutica (Barcelona, Spain).

Ammonium hydrogencarbonate (99% purity) and formic acidmass spectroscopy grade) was purchased from Sigma–AldrichSteinheim, Germany). Acetonitrile and methanol (both LC–MSrade) and water were purchased from Biosolve (Valkenswaard,he Nederlands). Isopropyl alcohol and ammonia solution

32%, extra pure) were from Merck (Darmstadt, Germany).Oasis MCX (mixed-mode cation-exchange) Prospekt car-

ridges (30 mg, 1 mL) were from Waters (Milford, MA, USA).

.2. Specimens

Pooled blank plasma samples were used for development andalidation of the procedure and were obtained from a local bloodank. Authentic plasma samples were obtained from hospitalases.

.3. Preparation of standard solutions

Separate working solutions of the drugs, for tuning and

electivity experiments, were prepared in the laboratory at aoncentration of 1 mg/L in methanol. A mixed working solutionf non-deuterated compounds at 10 mg/L in methanol was usedor the preparation of calibrators and QC samples. A mixed I.S.

romat

ws

di

2

2

SNalc(wwia5uottacoipccslwlws

2

cf5taiese5tn7tev

is

2

uilt1

msgimsppcaa

Mu

2

r

2d

tatp2eamnb

tol

pabr

A. de Castro et al. / J. Ch

orking solution of 1 mg/L was prepared in methanol. Workingolutions were stored at −20 ◦C, and were prepared monthly.

To obtain the lower concentrations needed for internal stan-ardization and validation of each experiment, further dilutionsn water were prepared the same day.

.4. SPE–LC–MS/MS

.4.1. Sample preparation: XLC (on-line SPE)Sample extraction was performed using the on-line SPE

ymbiosis Pharma System (Spark Holland, Emmen, Theetherlands). It comprises two integrated units: the Reliance

utosampler with two binary LC pumps integrated and the on-ine SPE unit Prospekt-2 system [consisting of the automatedartridge exchange (ACE) unit and two high-pressure dispensersHPD)]. The entire system was operated by SparkLink soft-are (version 3.10, Spark Holland). The extraction procedureas carried out in total recovery screw top vials of high qual-

ty glassware (Waters). A 950 �L volume of 0.1% formic acidnd 50 �L of the I.S. working solution (0.1 mg/L) were added to0 �L of plasma. The following XLC program was subsequentlysed (Fig. 1): After conditioning with 1 mL of methanol, 1 mLf water and 1 mL of 0.1% formic acid (Fig. 1A), 100 �L ofhe diluted plasma sample was applied onto the SPE MCX car-ridge (cation-exchange mode) using 1 mL of 0.1% of formiccid as transport solvent. Clean-up was accomplished with suc-essive 1 mL washes of 0.1% of formic acid and methanol inrder to wash out salts and endogenous interferences presentn the biological sample (Fig. 1B). The cartridge was thenhysically moved with a robotic arm to the elution (right)lamp in line with the LC pumps, leaving the extraction (left)lamp ready to start with a new sample. Whilst the elutiontep was being performed, a new cartridge was conditioned,oaded and washed in the left clamp. The elution was performedith 300 �L of 5% ammonia in methanol (at 100 �L/min). Fo-

lowing the elution step, several automated clamp and valveashes were carried out to avoid contamination between

amples.

.4.2. Chromatographic conditionsFocusing of the eluate was simultaneously performed as the

ompounds were eluted from the SPE cartridge by the use of aocusing column, Gemini C18 guard column (4 mm × 2.0 mm,�m) (Phenomenex, Torrance, CA, USA), and a gradient elu-

ion with 10 mM ammonium hydrogencarbonate (pH 10) (A)nd acetonitrile (B) (Fig. 1C). A gradient was carried out start-ng from 0% B and a flow rate of 1 mL/min for 3 min, as theluate was diverted to the waste using the MS/MS Rheodynewitching valve. At 3.01 min, a switch of the valve delivered theluent to the analytical Gemini C18 column (150 mm × 2 mm,�m) (Phenomenex) (Fig. 1D) to start with the separation of

he compounds at a flow rate of 0.3 mL/min and 50% B over theext minute. From 4 to 5 min, B was subsequently increased to

0%, and then kept for 6.5 min. At 11.5 min, B was increasedo 95% in 1.5 min before returning to 50% within 0.5 min andquilibrating for 4.5 min. At 18 min a switch of the MS/MSalve diverted the eluent again to the waste, returning to the

cep

ogr. A 1160 (2007) 3–12 5

nitial conditions to be ready for the analysis of the followingample.

.4.3. Tandem mass spectrometryA Quattro Premier tandem mass spectrometer (Waters) was

sed for all analyses. Ionization was achieved using electrosprayn positive ionization mode (ESI+). Nitrogen was used as nebu-isation and desolvation gas at a flow rate of 800 L/h and heatedo 350 ◦C. Capillary voltage and source block temperature werekV and 120 ◦C, respectively.

In order to establish the appropriate multiple reactiononitoring (MRM) conditions for the individual compounds,

olutions of standards [200 �g/L, in 10 mM ammonium hydro-encarbonate (pH 10)–acetonitrile (50:50, v/v)] were infusednto the mass spectrometer and the cone voltage (CV) opti-

ised to maximise the intensity of the protonated molecularpecies [M + H]+. Collision-induced dissociation (CID) of eachrotonated molecule was performed. The collision gas (argon)ressure was maintained at 0.35 Pa (3.5 × 10−3 mBar) and theollision energy (eV) adjusted to optimise the signal for the mostbundant product ions, which were subsequently used for MRMnalysis.

All aspects of data acquisition were controlled usingassLynx NT 4.0 software with automated data processing

sing the TargetLynx software (Waters).

.5. On-line SPE–LC–MS/MS assay validation

The analytical validation was performed according to theecommendations of Peters and Maurer [38] and Shah et al. [39].

.5.1. Linearity, limit of quantification (LOQ), limit ofetection (LOD), precision and accuracy

Quantification was performed by integration of the area underhe specific MRM chromatograms in reference to the integratedrea of the deuterated analogues. Freshly prepared working solu-ions of 0.01, 0.05, 0.25, and 1 mg/L in water were used torepare plasma calibrators at a concentration of 10, 25, 50, 125,50, 500 and 1000 �g/L. Standard curves, freshly prepared withach batch of QC and authentic samples, were generated usingleast-squares linear regression, with a 1/x-weighting factor forost of the compounds, except for trazodone, nortrityline and

orclomipramine, for which a quadratic response was found toe more suitable.

The limit of quantification (LOQ) was defined as the concen-ration of the lowest calibrator that was calculated within ±20%f the nominal value and with a relative standard deviation (RSD)ess than 20%.

The limit of detection (LOD) was estimated from blanclasma samples, spiked with decreasing concentrations of thenalytes, where the response of the qualitative ion could relia-ly differentiate from background noise and with signal to noiseatio (S/N) of the qualifier equal to or greater than three.

QCs were prepared for every run in blank plasma at a con-entration of 40, 200 and 800 �g/L. Intra-assay precision wasvaluated by replicate (n = 5) analysis of the three QC sam-les in one run. Inter-assay precision was evaluated by replicate

6 A. de Castro et al. / J. Chromatogr. A 1160 (2007) 3–12

Fig. 1. Scheme of on-line SPE system-LC (Symbiosis Pharma) coupled to the MS/MS. (A) The sample is loaded into the sample loop while the first SPE cartridge isconditioned in the left clamp. (B) The sample is loaded onto the cartridge followed by a washing step. (C) The cartridge is moved to the right clamp, where the sampleis eluted; the analytes are subsequently retained on the focussing column. Simultaneously, a second sample is subjected to the initial procedure of extraction in theleft clamp. (D) After the elution step the MS valve is switched and the analytes are now directed to the analytical column in connection to the MS/MS. Followingthe analysis of the first sample, the second SPE cartridge is transferred to the right clamp in order to be eluted and analysed.

romat

aotnm

2

fop(

pasa

tl

2

pitwwiof

Fifl

A. de Castro et al. / J. Ch

nalysis of the QC samples in several experiments performedn six different days. Comparing the calculated concentra-ions of all calibrators and QC samples to their respectiveominal values, provided data on the bias (accuracy) of theethod.

.5.2. Selectivity and specificityThe selectivity of the method against endogenous inter-

erences was verified by examination of the chromatogramsbtained after the extraction of six different blank plasma sam-les conserved in two different anticoagulants (sodium citraten = 3) and sodium fluoride (n = 3).

The combination of benzodiazepines with antidepressants

otentiates the treatment of depression, thus, benzodiazepinesre frequently encountered in these samples. For this rea-on, the specificity of the method was also assessed by thenalysis of plasma samples spiked at 2 mg/L of a solution con-

oite

ig. 2. MRM chromatograms obtained following the analysis of a spiked plasma smipramine, (5) norclomipramine, (6) nortriptyliline, (7) desipramine, (8) fluoxetinuvoxamine and (14) trazodone. Peak intensity is shown on the right-hand corner of

ogr. A 1160 (2007) 3–12 7

aining 27 benzodiazepines used for routine analysis in ouraboratory.

.5.3. Stability of samplesThe stability of the drugs in plasma was monitored in diluted

lasma samples as follows; 50 �L of blank plasma spiked at thenitial concentrations of 40, 200 and 800 (n = 9, at each concen-ration) were diluted with 950 �L of 0.1% formic acid. The I.S.as added to the control samples (n = 3) and the concentrationsere determined immediately. Another pool of samples was kept

n the autosampler at 6 ± 2 ◦C and analyzed prior to the additionf the I.S., after 24 h (n = 3) and 48 h (n = 3). For an evaluation ofreeze/thaw stability, the control samples at the concentrations

f 40 and 200 �g/L (n = 3) were spiked with the I.S. and analysedmmediately. The stability samples, spiked at the same concen-rations (n = 3), were subjected to three freeze/thaw cycles. Forach freeze/thaw cycle, the samples were frozen at −20 ◦C for

ample with 10 �g/L of (1) clomipramine, (2) amitriptyline, (3) sertraline, (4)e, (9) venlafaxine, (10) citalopram, (11) paroxetine, (12) norfluoxetine, (13)each trace.

8 A. de Castro et al. / J. Chromatogr. A 1160 (2007) 3–12

Table 1MRM transitions and conditions for all the compounds and their deuterated analogues

Compound Precursor ion (m/z) Product ion (m/z) Cone voltage (V) Collision energy (eV)

Trazodone 372.1 147.9 40 35176.0 25

Fluvoxamine 319.1 70.8 25 1586.8 15

Norfluoxetine 296.1 30.0 15 8133.9 6

Paroxetine 330.1 69.8 40 28192.1 20

Citalopram 325.1 108.8 30 25115.8 25

Venlafaxine 278.1 57.8 20 18260.1 12

Fluoxetine 310.1 43.9 20 12147.9 8

Desipramine 267.2 44.0 25 3071.8 15

Nortriptyline 264.1 90.8 25 20233.2 15

Norclomipramine 301.1 43.9 25 3571.8 20

Imipramine 281.2 57.7 25 3585.7 10

Sertraline 306.1 158.8 15 30275.0 12

Amitriptyline 278.1 90.8 30 25104.8 20

Clomipramine 315.1 57.9 30 3085.8 18

Paroxetine-d6 336.1 75.8 35 32Fluoxetine-d6 316.1 43.9 20 12Desipramine-d3 270.2 74.8 25 15I 25 18C 30 20

U

21Ias

2

icmoarrps

Table 2Reported therapeutic range [16] and linearity data for 14 antidepressants andtheir metabolites

Compound Therapeuticrange (�g/L)

r2 LOQ(�g/L)

LOD(�g/L)

Trazodone 500–2500 0.99995 10 0.5Fluvoxamine 50–250 0.99848 10 1Norfluoxetine 100–500 0.99890 10 1Paroxetine 10–75 0.99939 10 0.5Citalopram 20–200 0.99793 10 0.5Venlafaxine 250–750 0.99727 10 0.5Fluoxetine 150–500 0.99720 10 1Desipramine 75–250 0.99945 10 1Nortriptyline 50–250 0.99922 10 1Norclomipramine 150–550 0.99702 10 1

mipramine-d3 284.2 88.8lomipramine-d3 318.1 88.8

nderlined transitions were used for quantification.

4 h, thawed, and then maintained at ambient temperature forh. After the three cycles, the samples were spiked with the

.S. and analyzed. Stability was tested against a lower percent-ge limit corresponding to 90% of the mean value of controlamples by on-sided t-test (P < 0.05).

.5.4. Assessment of matrix effectsTo assess any potential suppression or enhancement of ion-

zation due to the sample matrix, two different analyses werearried out. The first one involved a post-column infusion experi-ent. The study was based on a continuous post-column infusion

f a mixture of the drugs and their internal standards (10 �g/Lt a flow rate of 10 �L/min) to produce a constant elevated

esponse in the MRM channels. The interference of this constantesponse was monitored following the injection of plasma sam-les (in two different anticoagulants: sodium citrate (n = 3) andodium fluoride (n = 3) and compared to the response following

Imipramine 45–150 0.99951 10 0.5Sertraline 50–250 0.99900 10 1Amitriptyline 50–300 0.99845 10 0.5Clomipramine 20–250 0.99947 10 0.5

romat

tcowl

2

I

atsfc

TI

C

T

F

N

P

C

V

F

D

N

N

I

S

A

C

A. de Castro et al. / J. Ch

he injection of mobile phase only. A second type of experimentonsisted of a comparison of the peak responses of the analysisf a blank plasma sample spiked at 1000 �g/L calibrator (n = 3)ith those obtained from water spiked at the same concentration

evel.

.5.5. RecoveryRecoveries were estimated by performing two experiments.

n order to optimize the elution conditions i.e. elution volume

lttp

able 3ntra-assay and inter-assay precision and bias of the QC samples prepared in plasma

ompound Concentrationof QC (�g/L)

Intra-assay precision (n = 5)

Mean concentrationfound (�g/L)

RSD (%

razadone 40 43.6 6.1200 233.2 1.5800 692.6 4.7

luvoxamine 40 41.9 7.2200 216.5 3.3800 843.8 3.8

orfluoxetine 40 45.6 7.5200 236.1 3.8800 880.9 7.5

aroxetine 40 40.8 6.4200 212.8 2.2800 807.3 4.3

italopram 40 44.1 4.4200 222.0 2.0800 796.7 2.8

enlafaxine 40 44.0 5.6200 227.4 1.7800 780.8 5.5

luoxetine 40 43.8 7.3200 220.0 3.0800 817.0 8.3

esipramine 40 43.3 5.4200 213.7 3.1800 830.6 3.2

ortriptyline 40 41.7 6.4200 211.0 3.6800 813.4 5.1

orclomipramine 40 40.0 6.5200 226.9 15.5800 798.6 3.4

mipramine 40 43.4 2.3200 226.1 1.3800 827.1 3.8

ertraline 40 42.2 5.2200 221.9 3.5800 850.2 5.5

mitriptyline 40 38.3 5.9200 203.1 3.1800 805.6 4.2

lomipramine 40 43.9 6.5200 219.1 1.7800 764.4 5.1

ogr. A 1160 (2007) 3–12 9

nd flow rate, a 1000 �g/L calibrator (n = 3) was injected inhe absence of both the analytical and focusing column. Asuch the compounds achieve directly the MS/MS as they eluterom the SPE cartridge. In a second experiment, performed toalculate de total recovery, a 1000 �g/L calibrator (n = 3) was

oaded and washed in a first SPE cartridge while a second car-ridge was placed in series to determine the breakthrough ofhe first one. Both cartridges where subsequently eluted inde-endently. Recovery was considered to be the ratio between

at a concentration of 40, 200 and 800 �g/L

Inter-assay precision (n = 6)

) Bias (%) Mean concentrationfound (�g/L)

RSD (%) Bias (%)

9.0 41.0 11.5 2.516.6 199.6 7.0 −0.2

−13.4 673.4 13.2 −15.8

4.8 36.5 10.9 −8.88.3 189.4 7.6 −5.35.5 807.0 8.7 0.9

14.0 41.1 14.3 2.818.1 193.2 15.0 −3.410.1 801.5 8.6 0.2

2.0 38.5 6.5 −3.86.4 193.6 7.2 −3.20.9 805.9 8.8 0.7

10.3 43.1 7.8 7.811.0 201.3 7.8 0.7−0.4 780.7 10.1 −2.4

10.0 43.1 12.5 7.813.7 198.3 10.4 −0.8−2.4 785.4 13.0 −1.8

9.5 40.0 10.7 0.010.0 196.3 9.8 −1.82.1 814.3 9.9 1.8

8.2 40.5 5.4 1.36.9 194.4 6.2 −2.83.8 791.0 11.1 −1.1

4.3 38.1 8.4 −4.85.5 192.1 8.3 −4.01.7 800.0 13.6 0.0

0.0 36.6 15.4 −8.513.5 197.0 7.0 −1.5−0.2 808.9 5.5 1.1

8.5 40.1 6.4 0.313.1 198.1 7.4 −1.03.4 806.5 11.4 0.8

5.5 37.2 14.1 −7.011.0 193.9 9.2 −3.16.3 824.6 7.6 3.1

−4.3 35.2 10.1 −12.01.6 181.8 9.2 −9.10.7 780.1 9.7 −2.5

9.8 41.5 9.9 3.89.6 207.1 7.5 3.6

-4.5 808.2 10.3 1.0

1 romat

ttc

3

acfqbottslsa

eotd

r

fm

saa

rats

cta

m6fao

F(

0 A. de Castro et al. / J. Ch

he response obtained after elution of the first cartridge andhe total response (sum of both, the first and the second SPEartridge).

. Results and discussion

The method was validated for linearity, LOQ, precision andccuracy by the analysis of spiked plasma samples. In eachase, a weighted (1/x) linear regression line was applied, exceptor trazodone, nortriptyline and norclomipramine, for which auadratic response was found to be more suitable to obtain theest fit across the calibration range [38]. Correlation coefficientf r2 > 0.99 was achieved in the range investigated: from 10 upo 1000 �g/L. The range investigated was considered accordinghe therapeutic range elaborated by TIAFT [16] (Table 2) Fig. 2hows the MRM chromatograms obtained following the ana-ysis of the lowest calibrator (10 �g/L). At this concentration aignal-to-noise ratio (S/N) > 10:1 was observed for the qualifiernd the criteria for LOQ were satisfied.

The applied gradient ensured the elution of all the drugsxamined within 14 min and produced chromatographic peaksf acceptable symmetry. Since a focusing step is a crucial factor

o obtain good peaks and sensitivity, many efforts were made toetermine the optimal conditions.

Selectivity of the method was achieved by a combination ofetention time, precursor and product ions.

tt

l

ig. 3. Post-column infusion experiments: matrix effect. (1) trazodone, (2) paroxetineA) and a blank sample following the extraction of plasma (B). The shaded area indic

ogr. A 1160 (2007) 3–12

The most prominent precursor-product transitions were usedor quantification of the non-deuterated compounds and the nextost abundant, used as qualifiers.For the corresponding deuterated analogues, only one tran-

ition was monitored. Table 1 summarises the MRM transitionsnd conditions of all quantifiers and qualifiers for all analytesnd I.S.

Injection of single analyte solutions did not produce interfe-ence in the other MRM channels. Linearity data, LOQ and LODre shown in Table 2. No interference peaks were observed inhe antidepressants MRM channels when blank plasma samplespiked with 27 benzodiazepines were analysed.

The intra-assay precision (repeatability) and inter-assay pre-ision were satisfactory, with all relative standard deviations lesshan 20% (Table 3). Results indicated that the accuracy of thessay was >81%.

The stability of spiked samples (40, 200, 800 �g/L) wasonitored at 24 and 48 h while kept in the autosampler at± 2 ◦C. No statistical significant difference could be observed

or the three different concentrations, except for clomipraminend norclomipramine where a slightly negative trend wasbserved (P > 0.05), but it did not compromise de quantifica-

ion. All compounds were stable (P < 0.05) after the three freezehaw/cycles.

Insufficient sample clean-up can result in matrix effects,eading to either suppression or enhancement of the analyte

, (3) nortriptyline and (4) amitriptyline, of an injection of a mobile phase controlates the elution position of the respective antidepressants.

romatogr. A 1160 (2007) 3–12 11

rpLseaoafFAtrpricttts

mtrdac

avd

Table 4Extraction recovery and matrix effect

Recovery (%) Estimated matrixeffect after SPE (%)

(n = 3) (n = 3)

Trazodone 99.9 −10.9Venlafaxine 99.8 −10.6Citalopram 99.9 −5.9Desipramine 99.8 −9.1Imipramine 99.9 −7.1Nortriptyline 99.6 −11.8Amitriptyline 99.9 −12.8Paroxetine 99.8 −6.7Fluvoxamine 99.7 −16.8Norfluoxetine 99.8 0.5Fluoxetine 99.9 −12.1Sertraline 99.9 −15.2Clomipramine 99.7 −17.7N

D

ostsh

pH

Ffc

A. de Castro et al. / J. Ch

esponse. This can lead to variable sensitivities and decreasedrecision and accuracy. Consequently, in the development of anyC–MS(/MS) method, the potential for any such ion suppres-ion or enhancement should be assessed. Post-column infusionxperiments (based on the method described by Bonfiglio etl. [40]) were performed to provide information of the effectf the matrix throughout the course of the elution time for thenalytes. An example of the effect on drug response, obtainedollowing the injection of a mobile phase control, is shown inig. 3A. As expected, no changes in response were observed.nother example of the effects obtained following the injec-

ion of a sample subjected to XLC are given in Fig. 3B. Theesults confirm the utility/benefits of the extraction as a sam-le clean-up before chromatography to obtain reproducible andeliable quantitative results for all the compounds without majornterferences of matrix compounds. A second experiment wasarried out and we compared peak responses obtained whenhe antidepressants were spiked to a blank plasma sample withhe responses obtained when the antidepressants were addedo a sample where the plasma was substituted with water. Notatistically significant differences in peak areas were observed.

The Oasis MCX Prospekt cartidges (30 mg, 1 mL) utiliseixed-mode (cation-exchange) sorbents, which provide effec-

ive sample clean-up for basic drugs. The results of the extractionecovery study are presented in Table 4. Very high and repro-ucible recoveries were obtained with this SPE procedure forll analytes and all compounds were totally eluted from the SPEartridge at the elution step conditions.

The validated SPE–LC–MS–MS method was applied to thenalysis of 11 authentic samples from clinical cases and pre-iously analysed by liquid chromatography with diode arrayetection (DAD) using a routine screening method. The results

fSfl

ig. 4. Typical MRM chromatograms obtained following the analysis of one authentior trazodone (B). The figure shows the response for the two transitions of both comorner of each trace.

orclomipramine 99.9 −10.4

ata represent the mean of three experiments with a 1000 �g/L calibrator.

f both methods were compared qualitatively (Table 5). Fig. 4hows the chromatogram obtained after the analysis of one ofhese samples, positive for trazodone and venlafaxine. The mea-ured concentrations varied in a wide range and several samplesad to be re-analysed after 1:10 dilution with blank plasma.

Several analytical methods for the determination of antide-ressants have been published using LC or LC–MS/MS.owever, most of them only allow the determination of one or

ew compounds with longer off-line SPE procedures [30,41–43].auvage et al. [44] have also developed an innovative methodor the determination of 13 antidepressants and some metabo-ites using turbulent-flow liquid chromatography (TFC). The

c plasma sample. Concentrations were 127 �g/L for venlafaxine (A) 1036 �g/Lpounds (quantifier and qualifier). Peak intensity is shown in the top right-hand

12 A. de Castro et al. / J. Chromat

Table 5Comparative results when analysing the real samples with the on-lineSPE–LC–MS/MS method and LC–DAD

Sample ID Concentration (�g/L) SPE-LC-MS/MS LC-DAD

1 Citalopram (14) −Fluoxetine (114) +Norfluoxetine (47) +

2 Trazodone (1036) +Venlafaxine (127) +

3 Fluoxetine (214) +Norfluoxetine (170) +

4 Trazodone (1963) +

5 Fluoxetine (381) +Norfluoxetine (95) +

6 Amitriptyline (132) +Nortriptyline (90) +

7 Trazodone (1284) +

8 Fluoxetine (92) +Norfluoxetine (47) +

9 Amitriptyline (1542) +

10 Fluoxetine (114) +Norfluoxetine (160) +

1

actsp

4

wtptwmttc

A

tn

R

[[[[[[

[[

[

[

[

[

[

[[

[

[[

[

[[

[[[

[[[

[[[

[

[

[

1 Fluoxetine (108) +Norfluoxetine (183) +

ntidepressants were divided into two groups depending on theirhromatographic properties, thus two injections were necessaryo screen all the compounds. On the other hand, it shows lowerensitivity and longer duration of entire total process (samplereparation and analysis) when comparing to the present paper.

. Conclusions

To our knowledge, this is so far the first on-line SPE methodith single use cartridges, coupled to LC–MS/MS applied for

he direct analysis of 14 antidepressants and metabolites inlasma. The combination of on-line SPE with MS/MS allowedhe development of a high-throughput, fast and sensitive methodith a 20 min total analysis time without compromising theethod validation criteria. The method was successfully applied

o 11 authentic plasma samples, proven to be appropriate forhe quantification of low both dose as high dose of theseompounds in plasma collected from forensic toxicology cases.

cknowledgement

This work was done in part thanks to financial support fromhe Ministerio de Educacion y Ciencia of Spain (F.P.U. Grantumber AP-2002-2878).

eferences

[1] S.E. To, R.A. Zepf, A.G. Woods, J. Neurosci. Nurs. 37 (2005) 102.[2] S. Caccia, Clin. Pharmacokinet. 34 (1998) 281.

[

[

ogr. A 1160 (2007) 3–12

[3] J.M. Kent, Lancet 355 (2000) 911.[4] L.J. Cohen, C.L. De Vane, Ann. Pharmacother. 30 (1996) 1471.[5] S. Caccia, Clin. Pharmakinet. 34 (1998) 281.[6] J. Kirchheiner, K. Brosen, M.L. Dahl, L.F. Gram, S. Kasper, I. Roots,

F. Sjoqvist, E. Spina, J. Brockmoller, Acta Psychiatr. Scand. 104 (2001)173.

[7] L.F. Gram, P.K. Sorensen, C.B. Kristensen, M. Moller, O.L. Pedersen, P.Thayssen, Adv. Biochem. Psychopharmacol. 39 (1984) 399.

[8] P.J. Goodnick, Clin. Pharmacokin. 27 (1994) 307.[9] S.H. Preskron, G.A. Fast, J. Clin. Psychiatry 52 (1991) 23.10] B.B. Rasmussen, K. Brosen, Ther. Drug Monit. 22 (2000) 143.11] P. Baumann, Clin. Pharmacokinet. 31 (1996) 444.12] C. Hiemke, S. Hartter, Pharmacol. Ther. 85 (2000) 11.13] P.B. Mitchel, Clin. Chem. Lab. Med. 42 (2004) 1212.14] M.J. Burke, S.H. Preskorn, Clin. Pharmacokinet. 37 (1999) 147.15] J. Lundmark, F. Bengtsson, C. Nordin, M. Reis, J. Walinder, Acta Psychiatr.

Scan. 110 (2000) 354.16] Terapeutic-toxic drug concentrations list. www.tiaft.org.17] C.L. Winek, W.W. Wahba, C.L. Winek Jr., T. Winek Balzer, Forensic Sci.

Int. 122 (2001).18] M.A. Martınez, C. Sanchez de la Torre, E. Almarza, J. Anal. Toxicol. 28

(2004) 174.19] R. De la Torre, J. Ortuno, J.A. Pascual, S. Gonzalez, J. Ballesta, Ther. Drug

Monit. 20 (1998) 340.20] J.J. Berzas, M.J. Llerena, A.M. Contento, E. Aguas, J. Chromatogr. Sci. 38

(2000) 200.21] Z.Q. Liu, Z.R. Tan, D. Wang, S.L. Huang, L.S. Wang, H.H. Zhou, J.

Chromatogr. B 769 (2002) 305.22] C. Salgado-Petinal, P. Lamas, C. Garcıa-Jares, M. Llompart, R. Cela, Anal.

Bioanal. Chem. 382 (2005) 1351.23] B.A. Way, D. Stickle, M.E. Mitchell, et al., J. Anal. Toxicol. 22 (1998) 374.24] C.B. Eap, G. Bouchoux, M. Amey, N. Cochard, L. Savary, P. Baumann, J.

Chromatogr. Sci. 36 (1998) 365.25] K. Titier, N. Castaing, E. Scotto-Gomez, F. Pehourcq, N. Moore, M. Moli-

mard, Ther. Drug Monit. 25 (2003) 581.26] J. Karpinska, B. Starczewska, J. Pharm. Biomed. Anal. 29 (2002) 519.27] L. Kristoffersen, A. Bugge, E. Lundanes, L. Slødal, J. Chromatogr. B 734

(1999) 229.28] S. Erturk, S. Muge, S. Atmaca, L. Ersoy, G. Baktir, Ther. Drug Monit. 27

(2005) 38.29] H. Zhang, K. Heinig, J. Henion, J. Mass Spectrom. 35 (2000) 423.30] K.E. Goeringer, I.M. McIntyre, O.L. Drummer, J. Anal. Toxicol. 27 (2003)

30.31] U. Guttek, K. Rentsch, Clin. Chem. Lab. Med. 41 (2003) 1571.32] M. Kollroser, C. Schobert, Ther. Drug Monit. 24 (2002) 537.33] R.S. Plumb, G.J. Dear, D.N. Mallet, D.M. Higton, S. Pleasance, R.A.

Biddlecombe, Xenobiotica 31 (2001) 599.34] E. Brewer, J. Henion, J. Pharm. Sci. 87 (1998) 395.35] J.L. Veuthey, S. Souverain, S. Rudaz, Ther. Drug Monit. 26 (2004) 161.36] S. Souverain, M. Mottaz, S. Cherkaoui, J.L. Veuthey, Anal. Bioanal. Chem.

377 (2003) 880.37] M. Kollroser, C. Schober, Chromatographia 57 (2003) 133.38] F.T. Peters, H. Maurer, Accred. Qual. Assur. 7 (2002) 441.39] V.P. Shah, K.K. Midha, J.W.A. Findlay, H.M. Hill, J.D. Hulse, I.J.

McGilveray, G. McKay, K.J. Miller, R.N. Patnaik, M.L. Powell, A. Ton-nelli, C.T. Viswanathan, A. Jacobi, Pharm. Res. 17 (2002) 1551.

40] R. Bonfiglio, R.C. King, T.V. Olah, K. Merkle, Rapid Commun. MassSpectrom. 13 (1999) 1175.

41] A. Eerkes, N. Weng, M. King, A. Du, W.Z. Shou, J. Liq. Chromatogr. Relat.Technol. 25 (2002) 1215.

42] P. Massaroti, N.M. Cassiano, L.F. Duarte, D.R. Campos, M.A.M. Mar-chioretto, G. Bernasconi, S. Calaffati, F.A.P. Barros, E.C. Meurer, J.

Pedrazzoli, J. Pharm. Sci. 8 (2005) 340.

43] S.J. Jain, M. Sanyal, G. Subbaiah, U.C. Pande, P. Shrivastav, J. Chromatogr.B 829 (2005) 69.

44] F.L. Sauvage, J.M. Gaulier, G. Lachatre, P. Marquet, Ther. Drug Monit. 1(2006) 123.