Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Antidepressants were first introduced into the market in the 1950s with the serendipitous discovery of the antidepressant effect of two drugs initially evaluated for other medical uses: Iproniazide, a monoamine oxidase inhibitor (MAOI), and Imipramine, a tricyclic antidepressant (TCA). Since then, a whole new generation of chemically and pharmacologically unrelated compounds have been introduced, which appear to be safer and better tolerated due to a more specific mechanism of action. These include selective serotonin reuptake inhibitors (SSRIs), serotonin and noradrenalin reuptake inhibitors (SNaRIs), noradrenergic and specific serotoninergic antidepressants (NaSSAs), and noradrenalin reuptake inhibitors (NaRIs).

Figure 1 shows the chemical structures of some of the most common antidepressants used in clinical practice.

Fig. 1
figure 1

Chemical structure of some antidepressants commonly used in clinical practice

Full size image

The main indication for antidepressants is depressive disorders, which have received increased attention owing to the growing recognition of their high prevalence. Other antidepressant uses include treatment of anxiety disorders, attention deficit hyperactivity disorders, nocturnal enuresis or psychosomatic disorders ­developed in several illnesses such as chronic pain, fibromyalgia, irritable bowel syndrome, or chronic fatigue syndrome.

Due to their high prescription, these therapeutic drugs are easy to acquire by depressed patients, who are prone to suicide attempts. This explains why antidepressants are among the most frequent therapeutic drug classes involved in forensic and clinical intoxications, mainly associated to voluntary intoxications [1–4]. Antidepressants with sedative side effects, like TCAs, could also be potentially used in drug facilitated sexual assault (DFSA) crimes [5]. Therefore, analytical techniques for the reliable identification and quantification of antidepressants should be available in clinical and forensic toxicology laboratories in order to perform a competent toxicological report. In addition, some antidepressants impair cognitive and psychomotor functions, and may increase the risk of driving accidents when under their influence [6–8]. For this reason, the Guidelines for Research in Drugged Driving elaborated by international experts in order to harmonize research in this field recommended the inclusion of sedative antidepressants in the panel of drugs to be analyzed in specimens collected from the roadside [9]. In addition to toxicological applications, analytical methodologies are also required to monitor plasmatic concentrations of some antidepressants. Therapeutic drug monitoring (TDM) of TCA is widely accepted [10–13] because of their narrow therapeutic window, and the development of severe cardiotoxicity and CNS toxicity close to the upper therapeutic concentrations; moreover, the enzymes involved in their metabolization show genetic polymorphism, being in part responsible for the interindividual variability in the plasmatic concentrations achieved at a given dose. Although new generations of antidepressants are less toxic and have wider therapeutic ranges due to a more selective mechanism of action, TDM of their plasmatic levels could be justified in special situations (assess compliance in nonresponder patients, hepatic or renal impairment, polymedicated patients, poor metabolizers, etc.) [11–13]. Antidepressant determination also is required to perform pharmacokinetic, bioavailability, and bioequivalence studies.

Different immunoassays are commercialized for the analysis of some antide-pressants [14–17]. Although useful for fast identification of these analytes, these ­techniques have several limitations: TCA are the main targeted analytes, it is not possible to differentiate between structurally related antidepressants, and several substances can cross-react with antidepressant assays [18–20]. Therefore, positive results must be confirmed with more specific techniques, usually using chromatography based procedures coupled to different detectors. Mass spectrometry is one of the most common detectors employed in clinical and toxicological analysis because of its high selectivity and sensitivity. Gas chromatography coupled to mass spectrometry (GC-MS) is a robust and well-established technique, and it is considered the golden standard for general unknown screening. Several GC-MS analytical methodologies have been described for identification and quantification of ­antidepressants in different biological matrices [21–25]; however, in most cases, antidepressants have to be derivatized because of their relatively high polarity. In the 1990s, the development of atmospheric pressure ionization interfaces (API) allowed the successful hyphenation of liquid chromatography to mass spectrometry (LC-MS). LC-MS combines the power of MS detection with the versatility of LC, which allows chromatographic separation of thermally unstable analytes, as well as compounds with a wide range of polarities without performing derivatization ­processes. For this reason, LC-MS and LC-MS/MS applications for the analysis of antidepressants have significantly grown over the last decade.

2 Sample Preparation for Antidepressants LC-MS(MS) Analysis

In the early stages of LC-MS, it was thought that the high selectivity provided by this technique could effectively eliminate intereferences caused by endogenous matrix compounds, and thus, the need for sample cleanup [26]. However, nowadays, it is well known that one limitation associated to LC-MS analysis is the suppression or enhancement on the analyte response when coeluting undetected compounds compete with the analyte in the ionization process [27, 28]. This phenomenon, generically known as matrix effect, may affect several validation parameters, such as the limit of detection, linearity, precision, and accuracy. Therefore, matrix effect may diminish method sensitivity and the reliability of quantitative results [29]. One of the main strategies to overcome matrix effect is to minimize the presence of these coeluting interferences through a more effective sample cleanup; thus, this step should be always taken into consideration when developing an LC-MS method.

Most of the published analytical methods for the determination of antidepressants were developed in plasma, serum, whole blood, and urine, which are the most useful matrices for clinical and toxicological analysis of these therapeutic compounds. However, albeit few, some LC-MS methodologies have also been described for the analysis of several antidepressants in hair [30–33], oral fluid [34, 35], breast milk [36], or typical forensic matrices such as gastric content, bile, vitreous humor, brain, liver, lung, and/or muscle [37–40].

Some biological matrices, such as hair and internal organs, require a special pretreatment prior to extraction. Hair samples must be washed to avoid external contamination, including an initial organic solvent, followed by aqueous washes [41]. After cutting or powdering, the hair matrix is usually disintegrated to extract the analytes from its inner structure. Ultrasonication in methanol (MeOH) [31, 32] or incubation in HCl [33] has been employed to extract antidepressants from the hair structure. Internal organs should be initially homogenized, and this process was usually performed in water or basic buffers by means of a blender or an ultrasonicator [37–40]. In addition, urine hydrolysis is sometimes performed to break glucuronide conjugates or some metabolites. β-Glucuronidase was employed for urine hydrolysis of mirtazapine [42] and bupropion metabolites conjugates [43, 44], and HCl was used for hydrolysis of 4-hydroxy-3-metoxy paroxetine metabolite [45], prior to sample extraction.

Several sample preparation or extraction procedures have been described for the analysis of antidepressants, from the most simple protein precipitation to online solid phase extraction (SPE). The optimum extraction technique depends on the analytical requirements, biological matrix, chromatographic separation, or the need for high throughput.

Several authors described a minimum sample pretreatment by plasmatic protein precipitation using acetonitrile (ACN), MeOH, acid solutions, or a mixture of them, for the analysis of one antidepressant and its metabolite(s) [46–52]. Single analyte procedures are useful for TDM or to perform pharmacokinetic, bioequivalence and pharmacogenomic studies, where the targeted antidepressant is known; however, this is an unrealistic situation in clinical and forensic toxicology, and for those applications, multianalyte procedures are preferable. Only Kirchherr et al. [53] used protein precipitation with ACN/MeOH as the only sample treatment procedure for the determination of 48 psychopharmaceuticals in human serum, including the main antidepressant drugs. No matrix effect was reported for any of the analytes, except for olanzapine, for which a 185% signal enhancement was observed.

However, protein precipitation usually produces severe matrix effect [54–56]. Therefore, in most LC-MS published methods, more extensive sample extraction procedures were used, being liquid–liquid (LLE) and SPE the most frequent techniques.

2.1 Liquid–Liquid Extraction (LLE)

In LLE, analytes are isolated from the biological matrix by means of an organic solvent immiscible with aqueous solution. Due to the basic character of antidepressants, samples are initially alkalinized with NaOH, NH4OH or a basic buffer to achieve a pH >8.5. The polarity range of the analytes included in the analytical method, as well as the biological matrix, determine the organic solvent of choice. Hexane, dichloromethane, butyl chloride, butyl acetate, isoamylic acid, or a mixture of them, have been used for the extraction of several antidepressants from the biological samples [39, 57–61]. Apart from those previously mentioned, other solvents employed for the extraction of only one antidepressant and/or metabolite(s) were diethylether, ethyl acetate or methyl tert-butylether [62–64]. After centrifugation, the organic layer is evaporated to dryness and reconstituted in a small volume for LC-MS analysis. In some cases, the organic layer was reextracted before evaporation by addition of aqueous acid solution to obtain cleaner extracts. De Santana et al. [42] and Halvorsen et al. [65] determined one or several antidepressants, respectively, in whole blood and/or plasma, using liquid-phase microextraction (LPME). LPME is a minituarized LLE procedure, where analytes are extracted from the biological matrix through an organic solvent impregnated in the pores of a hollow fiber and into a micro-liquid phase (acceptor solution) inside the fiber, which is subsequently injected into the LC-MS system. The main advantage of this technique is that it enables simultaneous analyte preconcentration and sample cleanup, using small sample and solvent volumes.

2.2 Solid Phase Extraction (SPE)

Different SPE procedures have been described using sorbents based on reversed-phase mechanism (C8, C18, hydrophilic–lipophilic balance), where the sample is loaded in basic conditions to guarantee analyte retention. After washing the ­cartridge, elution is performed with different organic solvents, sometimes in acid conditions [33, 66–71]. Due to the basic properties of antidepressants, mixed-mode sorbents simultaneously acting by reversed-phase and cation-exchange mechanisms, allow for a more selective extraction by removing nonbasic endogenous material [32, 34, 35, 43, 72, 73]. To allow analyte retention by cation-exchange mechanism, the sample should be initially conditioned with acidic solutions; after the washing step(s), analytes are eluted using alkalinized organic solvents. Common elution solvents are MeOH, ACN, dichloromethane, 2-propanol, or mixtures of them. The extraction process can be facilitated using semiautomated SPE robots [32, 66]. Moreover, online SPE-LC-MS instruments have also been employed for antidepressant analysis [72]. In these instruments, SPE extraction of the sample is performed in parallel to the chromatographic separation of the previous one, allowing complete automation of the whole analytical method, and the consequent sample high throughput.

2.3 Online Extraction

Online extraction techniques by means of supports coupled to the chromatographic system have also been applied to the determination of antidepressants [74–78]. These systems allow for direct sample injection, since large molecules from the biological matrix can readily pass through the column, while the analytes of interest can be retained under aqueous conditions, and subsequently eluted using high organic solvents composition. There are different supports, such as the so-called restricted access media (RAM), large particle size (LPZ) and monolithic supports, which can be used in the single column configuration or, more frequently, in the column-switching approach.

2.4 Solid-Phase Microextraction (SPME)

Another alternative technique, solid-phase microextraction (SPME), was used for the determination of fluoxetine [79] and several TCAs [80]. SPME is a miniaturized and solvent-free technique, where analytes are extracted from the sample by adsorption on a thin polymer coating fixed to the solid surface of a fiber, located inside an injection needle or a capillary. Its main disadvantage is that special strategies are needed to couple SPME to the LC-MS analysis.

3 Chromatographic Separation of Antidepressants by LC-MS(/MS)

LC is a very versatile technique that allows separation of nonvolatile, thermolabile and high polar analytes, without the need to perform derivatization processes. LC coupled to traditional detectors such as UV requires separation of interferences from the analyte of interest, and between them when several analytes are included in the analytical method; this is usually traduced in long chromatographic run times. The hyphenation of the LC system to the MS significantly reduced the chromatographic run, as there is no need for complete chromatographic resolution of the analytes included in the method. However, as previously mentioned, although endogenous interferences are not observed when analyte specific masses (m/z) or transitions are monitored, they can cause ion suppression or enhancement if they coelute with the analyte(s) of interest. Therefore, efficient chromatographic separation can minimize matrix effects and increase precision and accuracy of the assay.

In LC, several parameters such as the mobile phase, stationary phase, and column temperature can be optimized for a specific analytical application.

3.1 Mobile Phase

Mobile phase options are quite restricted, as only volatile buffers are suitable for LC-MS. In addition, ion-pairing agents traditionally used in LC to improve peak shape and retention time such as trifluoroacetic acid (TFA), have shown to produce ion suppression, and they are not recommended for LC-MS analysis [27]. Therefore, although few applications for specific antidepressants employ TFA or its ammonium salt due to sensitivity enhancement [48, 81–85], aqueous phases in most LC-MS analytical methods are composed by formic or acetic acid in water, or its ammonium buffers. Although acid mobile phases are by far the most common, basic aqueous mobile phases (pH ranging from 8 to 10) have been used for specific applications in order to increase antidepressant retention time or to couple the online SPE elution to chromatographic analysis [57, 72, 86, 87]. Organic phase composition was typically ACN and/or MeOH.

3.2 Stationary Phase

Antidepressant separation was usually ­performed by reversed-phase chromatography with typical C8 or C18 alkyl chain columns, although phenyl [30, 59] or cyano [48, 64, 84] stationary phases were also employed. As an exception, hydrophilic interaction liquid chromatography (HILIC), a variation of normal phase chromatography, was employed in two analytical ­methods for duloxetine [38] and paroxetine [85] determination, respectively. HILIC ­columns allow adequate ­retention of polar analytes poorly retained by reversed-phase chromatography, while still retaining less polar analytes; moreover, the high organic composition of the mobile phase increases electrospray efficiency, thus ­providing higher ­sensitivity. Chiral stationary phases were employed for enantioselective analysis of antidepressants marketed as a racemic mixture such as fluoxetine, citalopram, venlafaxine, mirtazapine, or bupropion [42, 43, 82, 88, 89].

Analytical methods for the determination of one antidepressant and/or its metabolite(s) were usually performed in isocratic mode, with total run times from seconds to a few minutes. However, as previously mentioned, multianalyte procedures are preferable, particularly if the method is intended for clinical or forensic analysis. Gradient separation was usually applied when the most common antidepressants were included in the methodology; however, total chromatographic run times varied widely, from 5 to 40 min [57, 76], depending on column length, extraction technique (offline vs. online techniques), biological matrix or the specific application of the method.

4 Mass Spectrometry Characteristics for Antidepressants LC-MS(MS) Analysis

4.1 Atmospheric Pressure Ionization Interfaces

Several LC-MS interfaces have been developed since 1974, when Arpino et al. [90] described the first attempt to couple the LC system to the mass spectrometer. Some of them were commercialized [91], but it was the development of atmospheric ­pressure ionization interfaces (API) which actually lead to the great expansion of LC-MS applications.

Within API interfaces, electrospray ionization (ESI) is the preferable ionization method for polar analytes, and it was used in the great majority of LC-MS methods for antidepressants analysis. However, ESI interfaces are more susceptible to matrix effects than atmospheric pressure chemical ionization (APCI) [54, 92, 93] due to ­differences in the ionization process. ESI is based on liquid phase reactions, where ion suppression is more likely than in gas phase due to high concentrations of nonvolatile materials present in the spray with the analyte [92]. Because of the lower susceptibility for matrix effects, although few, there have been authors who selected APCI as ionization mode in spite of the lower sensitivity for these analytes [59, 62, 74, 75]. Although ESI and APCI are by far the most popular ionization methods, other alternative interfaces were used for antidepressant analysis. Shinozuka et al. [69] used sonic spray ionization (SSI), a variant of ESI, where ionization is produced by high sonic gas velocity during pneumatic nebulization, instead of the electric field and capillary high temperatures applied in ESI. Atmospheric pressure photoionization (APPI) is a novel interface originally developed to widen the group of analytes to be determined by LC-MS towards less polar compounds that are not efficiently ionized by ESI or APCI. Theron et al. [61] compared ESI and APPI ionization for the analysis of venlafaxine and its metabolite O-desmethylvenlafaxine in plasma and water solutions. They concluded that the linearity of response in water, which is lost at high concentrations with ESI due to droplet surface saturation or limited amount of excess charge on the droplet, could be increased with APPI ionization. Moreover, APPI was less susceptible to ion suppression than ESI, with which signal response and calibration range was dramatically reduced in plasma compared to water, while similar results were obtained with APPI in both experiments. Regardless of the ­ionization method, positive ionization mode was employed in all cases, except for sulfate conjugates, which are more efficiently ionized in negative mode [45, 73].

4.2 Mass Analyzer

With regard to the mass analyzer, quadrupoles (Q) were usually employed due to its suitability for quantitative analysis and its relatively low prices, affordable for most ­laboratories. Ion trap mass spectrometers (IT) were employed in some but few LC-MS methodologies [65, 69, 74, 75]. Although these analyzers are less robust for quantitative analysis than Q, MSn spectrums can be obtained by successive selection and ­fragmentation of specific product ions. Franceski et al. [94] used Q-IT tandem mass spectrometer, combining the advantages of both analyzers, for the determination of fluoxetine and its metabolite norfluoxetine. A time-of-flight (TOF) mass analyzer was used for fragmentation pathway elucidation of some TCAs [58] and sertraline [62], structural elucidation of bupropion metabolites in urine [45], and fluoxetine quantitation [95]. TOF analyzers allow accurate mass measurement with an assignation of four ­decimal digits, which dramatically reduces the possible ­elemental formula of the detected analyte; however, TOF is more expensive and has a narrower linear dynamic range than that achieved with Q and, therefore, the later is preferable for quantitative analysis of target analytes.

5 LC-MS Applications for Antidepressant Quantitative Analysis

Quantitative LC-MS analysis is usually performed by selection of the pseudomolecular ion of the analyte of interest in the selected ion monitoring (SIM) mode. In single quadrupole instruments, selection of fragment ions (m/z) is also possible by promoting collision induced dissociation (CID) reactions in the entrance to the mass spectrometer. Tandem LC-MS/MS instruments enhance selectivity and signal-­to-noise ratio compared to single quadrupole instruments by operating on multiple reaction monitoring (MRM) mode. In MRM mode, specific precursor-to-product ion transitions can be monitored by fragmentation of the analyte in the collision cell located between the two analyzers, and subsequently monitoring a selected fragment. According to the European Commission Decision 2002/657/EC [96], quantitative LC-MS(MS) methods for confirmation purposes should include a ­minimum number of fragments (m/z) or precursor-to-product ion transitions, which depends on the resolution of the mass analyzer. For low resolution analyzers, at least the parent ion (m/z) and two fragments (m/z) in SIM mode, or two precursor-to-product ions transitions in MRM mode, should be monitored.

Quantitative analytical methods should be validated in order to guarantee the quality, accuracy, and precision of the results. This is especially important in forensic and clinical toxicology, where the correct interpretation of the toxicological findings depends on the reliability of the analytical method. Minimum parameters that should be validated in quantitative procedures include selectivity, calibration model, lower limit of quantification (LLOQ), precision (repeatability and intermediate precision), accuracy (bias), and stability. If the method is applied to specimens where the analyte(s) concentration(s) is above the validated range, dilution integrity evaluation should also be included in the validation process. In addition to these parameters, in LC-MS methods it is essential to assess suppression or enhancement of the analyte signal due to matrix effects. Additional parameters that might be evaluated are recovery, limit of detection (LOD), reproducibility, and robustness. A review on validation experiments and acceptance criteria for LC-MS analysis has been published by Peters et al. [29, 97].

Another requirement for qualitative or quantitative analysis is the use of internal standards (IS) to compensate for sample preparation or chromatographic variability. This is of particular importance in LC-MS analysis, as an adequate IS can also compensate for the negative influence of matrix effects on method precision and accuracy. Stable-isotope-labeled ISs are the most appropriate for this ­purpose. If a specific ­deuterated analogue is not commercially available, it could be substituted for deuterated substances with similar physicochemical properties to the analyte of interest. However, the use of other marketed pharmaceuticals for this purpose should be avoided, as it cannot be excluded that the patient to be monitored has taken that drug.

LC-MS(MS) quantitative methods for antidepressant determination should fulfill special requirements regarding linearity range, LLOQ or the need for high throughput, depending on the specific application for which they were developed. Conditions and studied validation parameters for selected LC-MS(MS) methodologies for ­antidepressant determination are shown in Table 1.

Table 1 Selected LC-MS(MS) methods for antidepressants analysis in serum, plasma, and whole blood
Full size table

Several LC-MS and LC-MS/MS methods were developed in plasma for only one antidepressant and, sometimes, its major metabolite(s) to perform pharmacokinetic, bioavailability, or bioequivalence studies. Analytical methods developed for these purposes require very low LLOQ values and, usually, narrow linear ranges covering the low range of the therapeutic concentrations are validated. In this context, several methodologies were described for the determination of fluoxetine [94, 95, 98–100], paroxetine [44, 71, 85, 101, 102], venlafaxine [48, 61, 64, 86, 103, 104], sertraline [62, 68, 83], citalopram [46, 89] and escitalopram [105], mianserine [106, 107], mirtazapine [42], trazodone [84], nefazodone [51, 81], duloxetine [47, 50, 73], and bupropion [43]. Deuterated analogues of the analyte of interest or of other drugs were employed by few authors as IS [43, 61, 73, 81, 85, 99]; however, in most of these methods, another antidepressant or other therapeutic drug was used for this purpose. Although it is not recommended, the use of commercially available pharmaceuticals as IS could be only justified if the method is applied to a controlled administration study in which the presence of that IS in the biological matrix can be ruled out.

Analytical methods for TDM should cover therapeutic ranges, and LLOQ requirements are not as low as those for pharmacokinetic applications. LC-MS methodologies for the determination of one specific antidepressant could also be applied in these cases. However, as previously stated in this chapter, multianalyte procedures are preferable because they are simpler, faster, cheaper, and allow for the determination of several analytes using the same sample aliquot. Moreover, ­multianalyte procedures are not only useful, but also required in clinical or forensic applications, where the target antidepressant is initially unknown, and several of them could be involved in the intoxication.

LC-MS(MS) methods including several antidepressants will be described in more detail. Some of these methods include analytes belonging to the same ­antidepressant group. Alves et al. [80] developed an LC-MS method for the determination of some TCAs in plasma with a total run time of 18 min. Kollroser et al. described two LC-IT-MS procedures in plasma for the determination of seven TCAs [74] and three SSRIs [75], respectively, using an online preparative column. Total analysis time per sample was 12 and 6 min, respectively. The main disadvantage of these methods is that a therapeutic drug was used as IS, hindering its applicability to TDM or clinical and forensic toxicology analysis. Also Juan et al. [66] used fluvoxamine as IS for the determination of several SSRIs; however, the authors argue that this drug was not commercially available in their region, and that patients’ history was studied in order to exclude fluvoxamine intake. Breaud et al. [78] developed a LC-MS/MS method for two TCAs and their main metabolites by turbulent flow LC-MS/MS, which allowed analyte determination from only 10 μL of plasma, with a total analysis time of 3.5 min from the second injection. Zhang et al. [58] used a LC-TOF-MS instrument for accurate mass measurement of TCAs in 18 s, using reference standards and plasma samples fortified with these analytes; however, validated linearity ranges were below the upper therapeutic concentrations and, therefore, evaluation of dilution integrity would be needed in order to analyze specimens from patients on antidepressant treatment or from intoxication cases. More details on validation experiments performed in these methods are shown in Table 1.

Other authors described LC-MS or LC-MS/MS methods in serum, plasma or whole blood for the determination of several analytes belonging to different antidepressant groups. LC-MS methods were developed by Halvorsen et al. [65] and Gutteck et al. [108] using an ion trap and a single quadrupole mass analyzer, respectively, with chromatographic elution of the analytes included in each method within 10 min. Analytes included in Gutteck’s et al. method were divided in four groups depending on their therapeutic ranges, and slightly different extraction and chromatographic procedures were applied to each group. Shinozuka et al. [69] extended the number of antidepressants using a LC-IT-MS instrument, and chromatographic separation was performed in 30 min. Tandem LC-MS/MS instruments were also employed in several occasions [53, 59, 60, 72, 77, 109]. Kirchherr et al. [53] employed protein precipitation of 0.1 mL of serum for the determination of 48 psycotherapeutics, including the most common antidepressants, with a total chromatographic run of 8 min. Also Remane et al. [59] developed a multianalyte procedure for the determination of 136 analytes from different drug classes using a generic LLE procedure; however, only matrix effect, recovery, and process efficiency were evaluated for method validation. Santos-Neto et al. [77] developed a LC-MS/MS method for the analysis of some antidepressants in plasma and urine using a monolithic column for simultaneous online sample preparation and chromatographic separation, with a total analysis time of 8 min. Sauvage et al. [76] extended the number of antidepressants determined in 0.1 mL of serum using turbulent flow LC-MS/MS, and decreased the total analysis time to 6 min. De Castro et al. [72] developed an online SPE-LC-MSMS procedure using 0.05 mL of plasma, where the complete analytical process for the determination of nine antidepressants was performed in 20 min. Table 1 shows the parameters evaluated for method validation, as well as analytes employed as IS, in each of these methods.

Although most LC-MS(MS) methods for antidepressants were applied to plasma, serum, or whole blood specimens, some analytical methodologies were also described in other biological matrices. Petsalo et al. [45] identified 20 bupropion metabolites in urine by accurate mass measurement using a LC-TOF-MS instrument. De Santana [42] applied a LC-MS/MS method with SPME extraction to urine samples to assess cumulative urinary excretion of mirtazapine and two metabolites. LC-MS/MS methods were also developed to compare enantiomers disposition of reboxetine [110], and bupropion and its metabolite [43] in plasma and urine specimens. Bupropion and some metabolites were also determined in urine by LC-MS/MS in a fatal overdose case where this antidepressant was involved [70]. LC-MS/MS was also applied to the analysis of other typical postmortem specimens, including gastric content, bile, vitreous humor, cerebrospinal fluid, brain, liver, lungs, kidney, or muscle, usually for the determination of one antidepressant and its main metabolite [38–40]. Goeringer et al. [57] extended the number of analytes included in the method to seven antidepressants, some metabolites and one antipsychotic in different forensic specimens. Unfortunately, validation performed in most of these methods was poor and, sometimes, only described for some of the analyzed specimens.

Few authors described antidepressant analysis in alternative specimens, such as hair or oral fluid. LC-CID-MS and MS/MS mass spectra libraries for identification of several drugs were employed by Müller et al. [32] for the detection of maprotiline, citalopram, and their desmethyl metabolites in authentic hair specimens; extracted ions chromatograms were employed for subsequent antidepressant quantification. Also Klys et al. [33] applied LC-MS/MS to the analysis of blood, urine, and hair specimens in a fatal case due to clomipramine overdose in combination with alcohol. Blood clomipramine and norclomipramine concentrations explained the fatal outcome, and hair analysis confirmed that the deceased was on clomipramine treatment for, at least, 12 months prior to his death. With regard to oral fluid analysis, de Castro et al. [34] developed and validated a LC-MS/MS procedure for the determination of the main marketed antidepressants in plasma and oral fluid to assess the possible correlation between the concentrations found in these two biological matrices. Also Coulter et al. [35] validated a LC-MS/MS methodology for several antidepressants and two other psychopharmaceuticals in oral fluid for its application to driving under the influence of drugs (DUID) cases.

6 LC-MS/MS Method for the Determination of Nine Antidepressants and Some of their Main Metabolites in Oral Fluid and Plasma

An LC-MS/MS method [34] will be described in detail in order to serve as an ­example of method development and validation. This method allows the determination of some of the most common antidepressants used in clinical practice and some of their main metabolites in oral fluid and plasma samples.

Plasma is the main biological sample used in clinical and toxicological analysis, as concentrations found in this matrix are correlated to the pharmacological effect, as well as to the side and toxic effects. However, oral fluid has also been employed in some specific applications because of the advantages associated to this alternative specimen: easy, painless, and noninvasive collection, which does not require qualified personnel, it represents the free analyte fraction, and it has a window of detection similar to that in plasma. Within the possible applications of oral fluid analysis, two are of special relevance:

  1. 1.

    Detection of subjects driving under the influence of sedative antidepressants effects.

  2. 2.

    TDM: As previously stated in this chapter, TDM is recommended for TCAs and, in special situations, for the new generations of antidepressants. However, ­several factors can affect diffusion of the analytes from plasma to oral fluid (pH, oral contamination, collection with or without stimulation). Therefore, the ­correlation between antidepressant plasma and oral fluid concentrations should be studied before using this alternative specimen for TDM purposes.

An example of method application to a study to assess the possible correlations between antidepressant concentrations in plasma and oral fluid will also be ­discussed in this section.

6.1 LC-MS/MS Method

A Waters Alliance 2795 Separation Module with a Waters Alliance series column/heater (Waters Corp., Milford, MA) was employed, using a Sunfire C18 (20 mm  ×  2.1 mm, 3.5 μm) Intelligent Speed™ column (Waters Corp., Milford, MA) for chromatographic separation at 26 °C. Mobile phase was ammonium formate buffer (pH 3.0; 2 mM) and ACN at a flow rate of 0.4 mL/min, applying the following ­gradient: 15% ACN for 0.5 min, increased to 50% over 3.5 min, and increased again to 70% at minute 5. With these conditions, all compounds eluted within 5 min, with a total run time of 8 min. Figure 2 shows the chromatograms of the 13 antidepressants in oral fluid at the LLOQ (2 ng/mL).

Fig. 2
figure 2

Chromatograms of the quantitative transitions for the 13 antidepressants included in the LC-MS/MS method in oral fluid at the LLOQ (2 ng/mL) (Source: From de Castro A et al. (2008) J Pharm Biomed Anal. 48:183. With permission)

Full size image

For the detection, a tandem mass spectrometer Quattro Micro™ API ESCI (Waters Corp., Milford, MA) with a triple quadrupole was employed. The instrument was operated in electrospray in the positive ionization mode (ESI+) with the following optimized parameters: capillary voltage, 0.5 kV; source block temperature, 130 °C; nebulization and desolvation gas (nitrogen) heated at 400 °C and delivered at 800 L/h, and as cone gas at 50 L/h; collision cell pressure, 3  ×  10–6 bar (argon). Data was recorded in the multiple reaction monitoring (MRM) mode by selection of the two most intense precursor-to-product ion transitions for each analyte, except for the ISs, for which only one transition was monitored. The most intense transition for each analyte was used for quantitative purposes. Table 2 shows MRM transitions, cone voltages and collision energies used for the analysis of the antidepressants included in the LC-MS/MS method.

Table 2 MRM transitions, cone voltages (CV), and collision energies (CE) used in the analysis of the antidepressants and the deuterated internal standards included in the LC-MS/MS method
Full size table

6.2 Oral Fluid and Plasma Extraction Procedure

Solid phase extraction (SPE) was performed with an ASPEC XL automated ­system (Gilson, Middletown, USA) and mixed mode OASIS MCX cartridges 60 mg 3 cm3 (Waters Corp., Milford, USA). Before SPE, 1 mL of sodium acetate buffer pH 3.6 and 50 μL of IS mixture (nortriptyline-d3, clomipramine-d3, paroxetine-d6, norfluoxetine-d6, and fluoxetine-d6) at 0.2 mg/L in oral fluid and 0.4 mg/L in plasma were added to 0.2 mL of sample. The applied SPE procedure is summarized in Fig. 3.

Fig. 3
figure 3

Solid-phase extraction (SPE) procedure applied to plasma and oral fluid samples for LC-MS/MS determination of nine antidepressants and some metabolites

Full size image

6.3 Method Validation

LC-MS/MS method validation was performed as follows:

6.3.1 Selectivity

Selectivity of the method was evaluated by analysis of blank oral fluid and plasma specimens from ten healthy subjects. In addition, potential exogenous interferences were assessed by analysis of authentic plasma specimens containing other common therapeutic drugs like benzodiazepines and/or drugs of abuse. No endogenous or exogenous interferences were found in the monitored MRM channels in any of the analyzed specimens.

6.3.2 Calibration Model

Calibration curves (n  =  6, six different days) were generated at eight concentration levels in the range from 2 to 500 ng/mL and 2–1,000 ng/mL for oral fluid and plasma, respectively. Coefficient of determination (r 2) was >0.99 for all analytes using a 1/x weighted linear regression, except for fluvoxamine, for which a ­quadratic response was observed.

6.3.3 Limit of Detection (LOD) and Limit of Quantification (LLOQ)

LOD was defined as the lowest concentration for which the two monitored transitions could be detected, and the peak area of the quantifier transition was, at least, three times the background noise. LOD was 0.5 ng/mL for all analytes. LLOQ was defined as the lowest concentration that could be quantified with appropriate ­imprecision (coefficient of variation (%CV) <20%) and inaccuracy (mean relative error (MRE)  ±  20%). LLOQ was 2 ng/mL in oral fluid, and 2–4 ng/mL in plasma for most of the analytes. Carryover after automated SPE of clomipramine at 1,000 ng/mL was half of the signal obtained at 2 ng/mL; therefore, LLOQ in plasma was increased to 10 ng/mL for this analyte.

6.3.4 Intra- and Interday Imprecision and Inaccuracy

Intraday imprecision and inaccuracy were evaluated at low, medium and high ­concentrations within the validated concentration ranges for each matrix by analysis of five replicates on the same day. Interday imprecision and inaccuracy were assessed by analysis of five replicates at the same concentrations on five different days. MRE was ±15% and %CV <17% in all cases.

6.3.5 Relative Ions Intensity

Intraday imprecision of relative ions intensity for each analyte was evaluated by analysis of five replicates at four concentration levels analyzed on the same day, and interday imprecision on this parameter was calculated at the same concentration levels (n  =  5 each) analyzed on five different days. %CV for the relative ions intensities was <17% in all cases, except for norfluoxetine, for which %CV was <26%.

6.3.6 Extraction Recovery

Extraction recovery was assessed at low and high concentrations by comparing analyte-to-IS peak area ratio when standards were added before extraction (n  =  5) to that obtained when added after extraction (n  =  5). IS mixture was added after extraction in all cases. Recoveries ranged from 49 to 72 % for all analytes.

6.3.7 Matrix Effect

Matrix effect was initially evaluated by postcolumn infusion of a mixture containing the analytes and the ISs (1 μg/mL, 10 μL/min) in “T” with the effluent of the chromatographic system. Chromatograms after the injection of extracted blank plasma (n  =  6) and oral fluid samples (n  =  6) from different sources were compared with the chromatograms after the injection of mobile phase (no matrix effect). Quantification of matrix effect was performed comparing average peak areas of blank plasma (n  =  6) and oral fluid samples (n  =  6) fortified with the analytes (100 ng/mL) and ISs after extraction to those obtained when the same amount of analytes and ISs were added to a clean tube, evaporated and reconstituted in mobile phase. Matrix effect was <15% for all analytes, except for norfluoxetine (signal enhancement <45%) and paroxetine (signal enhancement <30%). Nevertheless, inclusion of the deuterated analogues for these two analytes compensate for the possible errors in imprecision, inaccuracy and recovery.

6.3.8 Stability Study

Antidepressant stability after three freeze–thaw cycles was evaluated in triplicate at low and high concentrations. Calculated concentrations in the samples subjected to these conditions (stability samples) were compared to those obtained in freshly prepared samples (control samples). Stability and control samples were quantified with a calibration curve prepared on the day of the analysis. All analytes were stable under these conditions, except sertraline, for which a slight signal decrease was observed at 250 ng/mL in oral fluid (MRE  =  −33.4%; %CV  =  6.0%).

6.4 Oral Fluid-Plasma Preliminary Correlation Study

Initially, plasma and oral fluid specimens from patients (n  =  21) on different antidepressant treatment were collected twice to assess if any of the studied analytes was likely to show a good correlation. The best results were obtained for venlafaxine (%CV for plasma/oral fluid concentrations ratio (R OF/PL) <27%). Therefore, the study was extended for this antidepressant by analysis of oral fluid and plasma specimens from five patients on venlafaxine treatment collected on four occasions. Daily doses of venlafaxine retard formulations were 75 mg for two patients, and 150 mg for the remaining participants. Collection of oral fluid (direct spitting into polypropylene tubes) and plasma (heparinized tubes) specimens was performed, when possible, before the next dose to ensure the drug was in the elimination phase. The dose and the time of collection was the same on the four different occasions for each patient. For the analysis, oral fluid and plasma specimens were centrifuged at 14  ×  103 rpm, and 0.2 mL of the supernatant were extracted. In addition, correlation between the concentrations in the plasmatic free fraction and in oral fluid was also evaluated. Plasmatic proteins were eliminated by filtering 0.5 mL of plasma samples using Microcon filter devices Ultracel YM-3 (Millipore Corp., Billerica, MA, USA).

Table 3 shows plasmatic and oral fluid concentrations, ratios between those concentrations and calculated coefficients of determination for the five participants on venlafaxine treatment. Results were not analyzed interindividually due to the low number of patients and differences in the daily dosage. As expected, oral fluid concentrations were higher than those found in plasma in all cases. This is explained by venlafaxine’s weak basic properties, which favors oral fluid accumulation due to the slightly lower pH of this matrix compared to that in plasma. Intraindividually, plasmatic concentrations (C PL) were similar in the four analyzed specimens, but higher variability was found in oral fluid (C OF). Also, a great variability was observed in plasmatic free fraction concentrations (C PL-FF), which could be favored by some retention of the analytes in the filter used to eliminate the plasmatic proteins. Correlation between C OF vs. C PL or C PL-FF on the four different days was analyzed by linear regression. Only r 2 values for one patient were >0.6; therefore, our data indicates that a correlation between venlafaxine concentrations in oral fluid and plasma or the plasmatic free fraction is unlikely. Nevertheless, a strong conclusion would require the extension of the study, including a higher number of patients on the same daily dose, collecting more samples from each patient, and standardizing the interval time between drug administration and sample collection. Although our data do not support oral fluid analysis for antidepressant TDM, this alternative matrix could be employed in special situations to assess patients’ noncompliance or to detect impaired drivers under antidepressant influence.

Table 3 Results from venlafaxine study
Full size table