Symposium: field impairment testing (fit): assessing methodology to check for impairment

LC-MS/MS method for the determination of 13 antidepressants and metabolites.
Preliminary data about the correlation between plasma and oral fluid levels.
A. de Castro*, M. Concheiro, O. Quintela, A. Cruz, M. López-Rivadulla.
Forensic Toxicology Service. Institute of Legal Medicine. University of Santiago de Compostela. Spain.

Abstract
A selective and fast LC-MS/MS method has been developed for the determination of the main
marketed antidepressants in plasma and oral fluid. The antidepressants included in the
methodology were amitriptyline, imipramine, clomipramine, fluoxetine, paroxetine,
fluvoxamine, sertraline, citalopram and venlafaxine, as well as some of their main metabolites.
For sample clean up, 200 µL of sample previously conditioned with 1mL of acetate buffer pH
3.6 was extracted using an automated solid phase extraction system (ASPEC XL) and OASIS
MCX mixed mode cartridges. The chromatographic separation was performed in a reverse-phase
Sunfire C18 I.S. analytical column (20 x 2.1 mm, 3.5 µm), using a gradient of ACN and 2 mM
ammonium formate as mobile phase. With the applied conditions, all the compounds eluted in
less than 5 minutes, with a total run time of 8 minutes. A Quattro Micro API tandem mass
spectrometer was employed for the detection of the compounds. Both methodologies were fully
validated, including linearity, within-day and between-day precision and accuracy, limit of
detection and limit of quantitation, recovery, relative ions intensity, matrix effect and stability
after 3 freeze/thaw cycles. The method was applied to a preliminary study to asses the degree of
correlation between plasma and oral fluid samples.

Key words: antidepressant, LC-MS/MS, plasma, oral fluid
1. Introduction
Antidepressants are widely prescribed in depression and other psychiatric illnesses. Therapeutic monitoring (TDM) of antidepressants is a widely accepted practice for tricyclic antidepressants because of their narrow therapeutic window with high risk of cardiotoxicity and CNS toxicity, as well as the high intra and interindividual variability at the concentrations reached at a given dose. For the new generations of antidepressants, TDM is not justified routinely because of their wide therapeutic window and its relative security, but in special situations (elderly, slow or rapid metabolizers, polimedicated patients, when changing treatment or suspecting patients non-compliance) [1]. Plasma is the preferable specimen to be used in TDM, as it represents the concentration of the analyte responsible for the pharmacological effect, as well as the side and toxic effects. Oral fluid as an alternative to plasma samples has been studied for TDM of different compounds [2] because of the advantages of this matrix, mainly related to is easy and non-invasive collection. However, this specimen has also some disadvantages, like the small volume of sample usually available, as well as the fact that several factors can affect the diffusion of the analytes from plasma to oral fluid (pH, oral contamination, collection by stimulation vs non-stimulation). For these reason, correlation between plasma and oral fluid concentrations should be studied before using this alternative specimen for TDM purposes. 2. Materials and methods
2.1. Chemicals, reagents and biosamples
Individual stock solutions of amitriptyline, imipramine, desipramine, clomipramine, fluoxetine,
norfluoxetine, paroxetine, citalopram, sertraline (1 mg/mL in methanol), and the deuterated
internal standards (I.S.) nortriptiline-d3, imipramine-d3, clomipramine-d3, paroxetine-d6,
fluoxetine-d6 and norfluoxetine-d6 (0.1 mg/mL in methanol) were purchase from Cerillant (Round Rock, TX, USA). Nortriptyline, norclomipramine and venlafaxine were obtained in solid form from Fluka-Sigma-Aldrich Chemie (Steinheim, Switzerland). All other chemicals were of the highest available grade. 2.2. Extraction procedure For sample extraction, an automated solid phase extraction (SPE) system ASPEC XL (Gilson, Middletown, USA) and mixed mode OASIS MCX cartridges 3 cc 60 mg (Waters Corporation, Mildford, USA) were employed. Before extraction, the ASPEC XL was used to add 1 mL of sodium acetate buffer pH 3.6 and 50 µL of the I.S. mixture (at 0.4 mg/L for plasma samples and 0.2 mg/L for oral fluid samples) to 0.2 mL of specimen. After conditioning the SPE cartridges with 2 ml methanol and 2 mL of water, the samples were applied onto the cartridges. Clean-up was accomplished with successive 2 ml washes of formic acid 2% in water and methanol. Cartridges were then dried by positive pressure with a flow of nitrogen for 5 min before elution with 2 ml of dicloromethane/2-propanol/ammonium (75:24.5:0.5). The elution solution was evaporated to dryness at 35ºC under a stream of nitrogen. The dried extract was re-dissolved in 200 μl of a mixture of ammonium formate buffer pH 3.0 and acetonitrile (85:15, v/v) for plasma samples, and 100 µL for oral fluid samples. The sample was subsequently transferred to autosampler vials, and 20 μl were injected onto the LC-MS. 2.3. Liquid chromatography tandem mass spectrometry The HPLC system was a Waters Alliance 2795 Separation Module with a Waters Alliance series column heater/cooler (Waters, Mildford, MA, USA). A Sunfire C18 (20 x 2.1 mm, 3.5 µm) Intelligent Speed™ column was employed for the chromatographic separation of the antidepressants, using 2 mM ammonium formate buffer pH 3.0 and acetonitrile as mobile phase at a flow rate of 0.4 mL/min. The column temperature was kept at 26ºC and the following gradient was applied: 15% acetonitrile until minute 0.5; then, acetonitrile percentage was gradually increased to 50% until minute 4 and kept for 0.3 minutes, to increase again to 70% at minute 5. From minute 5.5 to minute 6 it was returned to initial conditions and kept for 2 minutes. With these conditions, all the compounds eluted within 5 minutes, with a total run time of 8 minutes. For the detection, a tandem mass spectrometer Quattro Micro™ API ESCI (Waters, Mildford, USA) with a triple quadrupole was employed. The instrument was operated in electrospray in the positive ionization mode (ESI +). Nitrogen was used as nebulization and desolvation gas at a flow rate of 800 L/h, heated to 400ºC, and as cone gas at a flow of 50 L/h. Capillary voltage and source block temperature were 0.5 kV and 130ºC, respectively. Data acquisition was controlled using MassLynx 4.0 software and processed with QuanLynx 4.0 software (Waters, Mildford, USA). 2.4. Validation assay
Validation of the analytical methods in plasma and oral fluid were performed following the
recommendations of international organizations like the FDA [3] and ICH [4], and Peters and
Maurer [5]. The studied parameters included selectivity, calibration model (linearity), LOD and
LLOQ, within and between precision and accuracy, recovery, matrix effect, relative ions
intensity and stability after three freeze/thaw cycles.
Table 1. MRM method, retention time (tR) and internal standards (I.S.) employed for the detection of the studied compounds.
Function Compound
transition
CV: cone voltage (V); CE: collision energy (eV);
2.5. Study of correlation between plasma and oral fluid antidepressant levels The method was applied to the analysis of plasma and oral fluid samples from a preliminary study to asses a possible correlation between plasma and oral fluid concentrations of the studied compounds. For this purpose, in a first stage, plasma and oral fluid samples from patients on treatment with any of the studied antidepressant were simultaneously collected on two consecutive weeks. In a second stage, only samples from patients on venlafaxine treatment were collected, but this time in four different days.
3. Results and discussion
3.1. Analytical methodology
The described analytical methodology allows the determination of the main antidepressants used
in the clinical practice. The use of an automated solid-phase extraction system allows, besides
improving the high-throughput, minimizing systematic errors because of sample manipulation.
OASIS MCX mixed mode cartridges were selected. These supports are specially designed for the
extraction of basic analytes as they retain the compounds by both, reverse-phase and cation
exchange mechanism, allowing obtaining cleaner extracts.
The selectivity of the method was verified as no interferences were found at the retention time of
any of the compounds in their MRM channels when blank samples or real cases positive to drugs
of abuse and other medicines like the benzodiazepines were analysed. The LOD was 0.5 ng/mL
for both, plasma and oral fluid samples. The LLOQ in plasma was 2, 4 or 10 ng/mL, depending
on the compound, and 2 ng/mL in oral fluid for all the compounds. The calibration model that
was best adjusted to our data was weighted (1/x) linear regression for all the compounds, except
for fluvoxamine, for which a quadratic response was observed. Although lineal models are
always preferable, the use of quadratic models is accepted, and even recommended, when it is
more adjusted to the behaviour followed by the compound. In that case, more concentration
levels are needed to define the calibration range [5]. Coefficient of determination (r2) was > 0.99
for all the compounds from 2 to 500 ng/mL (in oral fluid) or from 2, 4 or 10 to 1000 ng/mL (in
plasma). Within-day and between-day precision and accuracy were satisfactory for all the tested
concentrations. The calculated recoveries for all the compounds were included between 49 to
72%. Values of C.V. for within-day and between-day precision for the relative ions intensities
were included between 0.23 to 17% for all the compounds, except for norfluoxetine, for which
the values were around 1.08-26%. These results fulfilled the criterion for permitted variations of
relative intensities of MRM transitions [6]. No significant matrix effect was found for any of the
studied compounds as the signal intensities of the post-column infused analytes when injecting
mobile phase were similar to those obtained when injecting blank plasma and oral fluid samples
previously extracted with the described procedure. Stability studies of the analytes after three
freeze/thaw cycles of plasma and oral fluid samples indicate that all the compounds are stable
when subjected to these conditions, except in the case of sertraline in oral fluid, for which a
slight decrease in the signal at 250 ng/mL was found.

3.2. Study of correlation between plasma and oral fluid antidepressant levels
In a first stage, out of the 28 patients included in the study, 4 of them had to be excluded because
oral fluid or both plasma and oral fluid samples from the second week were not available. The
included patients were on treatment with venlafaxine (n=6), citalopram (n=6), paroxetine (n=4),
sertraline (n=3), fluoxetine (n=3), amitriptyline (n=2) and clomipramine (n=1). Interindividually,
results were analized by linear regression for the compounds for which more than two patients
were included, linking plasmatic concentrations (CPL) vs oral fluid concentrations (COF). The
obtained coefficient of regression for each compound indicates that there is not a good correlation for none of them when the data were analyzed interindividually. Intraindividually, results could not be analized by linear regression as only data from two occasions were available. Venlafaxine was the compound with the best results as the ratio between CPL and COF on both weeks was similar in 5 out of the 6 studied patients. For this reason, in a second stage, plasma and oral fluid samples from 5 patients on venlafaxine treatment were collected on 4 different days. Plasma samples were analyzed before and after filtrating plasmatic proteines. Interindividually, a high variability in venlafaxine COF was reached in different patients when the same dose was administered, but also in CPL and CPL free fraction. Coeffitient of variation (C.V.) > 41% in COF/CPL and COF/CPL free fraction indicate that a good correlation between plasma and oral fluid was not found. Intraindividually, CPL were similar in the 4 analyzed samples, but high differences in COF were found. The high variability in CPL free fraction could be due, among other factors, to the retention of the compounds in the filter. For each patinet, COF/CPL and COF/CPL free fraction the four different days was analyzed by linear regression. Only in one patient a coeffitient of determination (r2) > 0.8 was found, and C.V. was between 24.2-69.61 and 12.92-58.81, respectively, indicating that a good correlation is not likely to be found. Table 2. Values for CPL, CPL free fraction and COF, as well as the ratio between COF/CPL and COF/CPLfree fraction is
shown for each patient on venlafaxina treatment when samples were collected on four different days. Conclusions
A fast, sensitive and selective method for the determination of the main antidepressants used in
the clinical practice was developed in both plasma and oral fluid samples. This method was
applied to a preliminary study to asses the correlation between plasma and oral fluid levels of the
antidepressants. Results indicate that a good correlation between both specimens is not likely to
be found for none of the compounds, so oral fluid would not be a good predictor of the plasmatic
concentrations of antidepressants. However, this specimen could be useful in special situations
like when suspecting non-patient compliance or to evaluate antidepressant levels in cases of
intoxication or death.
References
[1] M.J. Burke, S.H. Preskorn. Clin. Pharmacokinet., 1999; 37 (2): 147-165.
[2] H. Liu, M.R. Delgado. Clin. Pharmacokinet, 1999; 36 (6): 453-470.
[3] U.S. Department of Health and Human Services. Food and Drug Administration. Guidance
for industry, bioanalytical method validation 2001.
, March 2007.
[4] International Conference on Harmonization (ICH). Validation of analytical methods:
Definitions and terminilogy ICH Q2 (R1). ,
March 2007.
[5] F.T. Peters, H.H. Maurer. Accred. Qual. Assur. 7: 441-449 (2002).
[6] L. Rivier. Anal. Chim. Acta 492: 69-82 (2003).

Source: http://www.icadts2007.org/print/167antidepressants.pdf