Recent advances in sample preparation techniques to overcome di ﬃ culties encountered during quantitative analysis of small molecules from bio ﬂ uids using LC-MS/MS

b Liquid chromatography-mass spectrometry analysis of small molecules from bio ﬂ uids requires sensitive and robust assays. Because of the very complex nature of many biological samples, e ﬃ cient sample preparation protocols to remove unwanted components and to selectively extract the compounds of interest are an essential part of almost every bioanalytical work ﬂ ow. This review describes the most common problems encountered during sample preparation, ways to optimize established sample preparation techniques and important recent developments to reduce or eliminate major interferents from bio ﬂ uids.


Introduction
The primary goal of sample preparation is to isolate one or several target analytes from the other components of the sample mixture (matrix).Depending on their nature and concentration levels, co-components of the sample matrix can inuence the quantitation of target analyte(s) during subsequent liquid chromatography-mass spectrometry (LC-MS) or tandem mass spectrometry (LC-MS/MS) experiments if not removed prior to analysis.The development of new LC-MS/MS methods for small molecules in biological uids is becoming increasingly more challenging, because of the need to continuously achieve higher sensitivity and better assay robustness in complex biouids such as serum, plasma, urine, oral uid or cerebrospinal uid (CSF).In addition, because of the very low concentration levels of pharmaceutical targets, samples oen need to be preconcentrated before analysis.Unfortunately, this does not only increase the concentration of the desired compound in the sample extract but also oen raises the levels of interfering components.As a result, very specic and effective sample clean-up procedures are required for sensitive and selective LC-MS/MS assays today. 1 This short review presents the main difficulties encountered during sample preparation for analysis of small molecules from biouids by LC-MS/MS and summarizes several critical factors that particular attention should be paid to, followed by an overview of the latest developments in sample preparation techniques to overcome common difficulties with complex biouids.

Matrix effects
The general term used today to describe problems encountered during analysis of complex biological samples is "matrix effects".These effects are usually caused by endogenous (e.g.metabolites of the target analyte, proteins or lipids) or exogenous (all substances introduced during sample processing and analysis) compounds.Depending on their chemical properties, it may or may not be necessary to remove all of these interferents from the sample before injection into the LC-MS system.Also, only matrix compounds coeluting with target analytes during the chromatographic separation prior to MS analysis can cause a change in the response of the analyte, either positive (ion enhancement effect) or negative (ion suppression effect). 2 Different methods have been presented to examine matrix effects.A common approach is the post-extraction spike method, [3][4][5] where the peak area of the target analyte that has been spiked into the biological matrix prior to the sample preparation is compared to the area of the same analyte spiked post-extraction into the biological uid extract.The ratio between the two values represents the absolute matrix effect.The relative matrix effect is determined by comparing several lots of the biological matrix. 3Obviously, both absolute and relative matrix effects depend strongly on the target analyte and the ionization technique used for LC-MS/MS.
Another popular method is post-column infusion, [6][7][8] where possible matrix effects are assessed by continuous post-column infusion of the analyte aer injection of a processed blank serum sample onto the chromatography column.Any variation of signal intensity at or near the retention times of the analyte would indicate the presence of substances from the matrix interfering with the analysis.
Matrix effects have been shown to be dependent on the ionization methods used for the LC-MS method, 3 which are usually either electrospray ionization (ESI) or atmospheric pressure chemical ionization (APCI) in most modern LC-MS/MS assays.The chemical structures and the concentration levels of both analyte and co-eluting mixture components determine whether they outcompete each other during the ionization process. 9For example, ESI is particularly sensitive to co-eluting phospholipids because ESI is strongly biased towards surfactants, 10 which enrich at the surface of the droplets during the liquid/gas-phase ion transfer.That is, phospholipids at the surface of droplets can inhibit ejection of analyte ions trapped inside the droplets.On the other hand, APCI is oen less affected by suppression effects, as there is no competition between compounds to enter the gas-phase of the mass spectrometer.Nevertheless, APCI still experiences matrix effects in multicomponent samples.As biouids contain numerous endogenous molecules, oen at high levels, with potentially very high basicities and surface activities, ion suppression effects will almost always be present in any LC-MS/MS assay.
Different strategies are available to eliminate or reduce matrix effects.One approach is to optimize the chromatographic separation to separate the analytes from interfering compounds. 1,11,12This can, however, result in long chromatographic run times.Another approach is to optimize the sample preparation, to obtain clean extracts of the target analytes.With proper sample preparation and the use of isotopically labeled standards, many matrix effects can be eliminated or strongly reduced.][15][16] There are several well-known causes for matrix effects in the analysis of clinically-relevant substances from biological samples.For example, hemolyzed or lipaemic samples have great inuence on the analysis of serum and plasma samples. 17,18Cases also have been reported, where buffers used for solid-phase extraction (SPE) triggered matrix effects in LC-MS/MS. 19The most important interferents, however, are phospholipids, which not only affect MS response of many analytes greatly, but which are also very difficult to remove from the samples.

Phospholipids
Phospholipids (PPL) are major constituents of cell membranes and are therefore very abundant in serum and plasma. 20They consist of two functional groups: a hydrophilic head group composed of phosphate and choline units, and a hydrophobic tail, made up of fatty acyl chains.The most abundant phospholipids are glycerophosphocholines (GPChos) (70% of total phospholipids) and lysophosphatidylcholines (10% of total phospholipids) (Fig. 1). 11These two groups are known to cause serious ion suppression effects in LC-MS analysis, caused by competition for space on the surface of droplets formed during the ESI process (vide supra). 3,10Phospholipids are present at different concentration levels in serum and plasma samples, depending on the sampling device used. 21A very simple method to monitor possible ion suppression effects from GPCho was described by Little et al. as in-source multiple reaction monitoring (IS-MRM). 22Using the positive ion mode, a common product ion for the most abundant GPCho is trimethylammonium-ethyl phosphate at m/z 184, which was monitored during analysis of an analyte-free sample.This class-specic product ion was generated using in-source dissociation of the eluting GPCho during the chromatographic run. 22Other methods have been described that allow screening for less abundant phospholipids by adding a precursor ion in the negative mode or by using positive ion neutral loss scans. 23tudies have shown that the use of methanol as a mobile phase for chromatographic separation provided signicant advantages over acetonitrile, because elution of all GPCho occurred in a very narrow time window and their retention behavior on reversed-phase columns could be predicted and decreased by increasing the percentage of the organic phase. 24he PPL tended to elute at a high content of the organic mobile phase 25 and were completely removed from the system at the end of a run by ushing the analytical column with isopropanol. 26he behavior of PPL has also been investigated on hydrophobic interaction liquid chromatography (HILIC) columns: 27 the compounds were focused into 2 groups of peaks (phosphatidyl cholines and lyso-phosphatidyl cholines) and eluted completely from the column in a one gradient cycle.In comparison, on a reversed-phase material, a strong carry-over was observed from one gradient cycle to another. 27n some cases, where retention times of target analytes and PPL overlapped, elution of the target substance could be shied aer adding mobile phase modiers. 27

Internal standards
The use of isotope-labeled internal standards can help overcome most of the matrix effects during sample preparation and LC-MS/MS analysis.However, in some cases the internal standard cannot completely fulll its purpose, because of slight differences in the chemical behavior of the target analyte and internal standard.For example, particular attention has to be paid to analytes showing strong protein binding. 28Generally it is necessary to allow enough time for the internal standard to properly equilibrate and bind to the protein before extraction, to ensure identical behavior of the internal standard and target analyte. 29A method has been described to determine the extent of protein binding of corticosteroids. 30In theory, this method could be extended to other substances and be used to compare the protein-affinity of an analyte and its internal standard.It is important that the release of analytes from the protein (e.g. by adding organic solvents for protein precipitation, o-phosphoric acid for breakdown of non-covalent intermolecular interactions 31 or dithiothreitol (DTT) or tris(2-carboxyethyl)phosphine (TCEP) for reduction of disulde bonds) has the same impact on the analyte and isotope-labeled standard.A case was reported, where the higher susceptibility of the internal standard for matrix effects than the target analyte led to an underestimation of up to 50% in the presence of specic buffers used for SPE sample preparation (Fig. 2). 19enerally, 13 C, 15 N or 18 O-labeled internal standards are preferable to deuterium labeled analogs, 32 because slight differences of physicochemical properties between hydrogen and deuterium can result in small shis of retention times of the analyte and internal standard.In some cases, this has led to a different degree of ion suppression for the analyte and the internal standard, resulting in changed analyte/internal standard peak area ratios. 33,34Also, deuterium-hydrogen backexchange can occur, which has led to false positive results. 35nfortunately, in many cases only deuterated compounds are commercially available, which increases the need to carefully investigate the stability of the reference standards and the inuence of matrix effects on the method.

Optimization of established sample preparation methods
Even though there has been some recent interest in quantitative analysis of pharmaceutical compounds from biological samples using ambient, direct mass spectrometry techniques such as desorption electrospray ionization (DESI) or direct analysis in real time (DART), with little or no prior sample preparation or chromatography, 36 sample clean-up remains a critical step in most LC-MS analyses of small molecules in biouids.

Protein precipitation
The simplest sample preparation approach for biouids is protein removal.Proteins can be denatured using acids or heat, or removed by using ultraltration cut-off membranes. 37nother possibility is to use organic solvents for protein precipitation (PPT).PPT removes a part of the phospholipid content present in serum and plasma samples, depending on the organic solvent used.Studies have shown that methanol extracts contain 40% more phospholipids compared to acetonitrile, 11 and are also less clean than tetrahydrofuran or ethanol extracts. 38

Solid-phase extraction (SPE)
Silica-based sorbents in SPE cartridges have excellent retention capacity for PPL when eluted with 100% acetonitrile. 39Clean extracts were also obtained by including a washing step with up to 50% methanol, but this strongly affected the recovery of polar analytes. 39Large amounts of methanol eluted signicant amounts of phospholipids from silica-based reversed-phase SPE cartridges.Methanol contents of 60, 70 and 80% for elution of samples on phenyl, C8 and C18 phases resulted in a high concentration of phospholipids in the extracts.Acetonitrile appeared to be a stronger eluent for phospholipids on reversedphase materials when present at levels up to 50%.The same study showed that the recovery of lysophosphatidylcholines decreased with the increasing content of acetonitrile (>50%), reaching its minimum at a 100% organic phase. 40The retention of phospholipids on the sorbent increased by interactions with residual silanol groups, as was shown by comparison of endcapped and non-endcapped materials.Silica-based sorbents were compared to polymeric phases regarding extraction of phospholipids, and the tested materials showed comparable efficiency. 40tudies comparing different sample preparation methods in terms of matrix effects and analyte recovery demonstrated that mixed-mode strong anion exchange SPE was more effective than PPT and LLE for polar and non-polar analytes in plasma (Fig. 3). 11,41ILIC-SPE was evaluated as an effective method to remove phospholipids from serum and plasma samples. 26The retention of phospholipids was shown to increase when samples were diluted with acetone.For some applications to urine samples, HILIC materials were more effective than reversedphase materials. 42The polar metabolites in urine had to be separated from the salts and other polar components present in urine.Orthogonal separation using both HILIC and reversedphase materials for sample preparation and chromatography improved the effectiveness of sample clean-up. 42[45][46][47] Liquid-liquid extraction (LLE) Liquid-liquid extraction has found numerous applications for analysis of pharmaceuticals and their metabolites.The concentration of residual phospholipids in the extract is usually lower compared to other techniques such as mixed-mode SPE; on the other hand, the extraction efficiency for highly polar analytes is also lower. 29The choice of extraction solvent is very important to reduce unspecic extraction of matrix components. 41Halogenated solvents such as chloroform or dichloromethane [48][49][50] are commonly used in combination with hydrophilic solvents (e.g.alcohols) for extraction of polar compounds; they also have high affinity for lipids. 38s non-ionized analytes are more efficiently extracted by organic solvents than charged species, particular attention has to be paid to the pH of the sample prior to LLE.As a general rule, the pH should be between pK a and (pK a À 2) for acidic analytes and between pK a and (pK a + 2) for basic analytes, 51 to increase the extraction recovery.This obviously applies only if  the stability of the main analyte and its potentially labile metabolites is given in this pH range. 29xtraction using methyl-tert-butylether (MTBE) has shown good results, 52 but signicantly lower analyte recoveries were seen compared to mixed mode SPE and PPT, especially for polar analytes. 11Only traces of phospholipids were found in MTBE and n-butylchloride extracts of serum and plasma samples. 53owever, particular attention has to be paid to the process, when several sample preparation steps are combined.The clean extracts obtained with MTBE for untreated serum or plasma can show a high recovery for phospholipids if the samples contain a high percentage of acetonitrile, e.g.aer protein precipitation (Fig. 4). 53xtraction time also plays an important role for the specic extraction of target analytes compared to matrix components.A study showed that a 5 min extraction time yielded a cleaner extract and better recovery for the target compound than 20 min, indicating that matrix compounds diffuse slower into the extraction solvent. 52o improve low recovery rates of LLE for strongly hydrophilic compounds, extraction procedures using water miscible solvents have been considered.Complex methods were reported in the past that use temperatures below 0 C to achieve phase separation of serum samples and extraction solvent. 54A more convenient way to achieve phase separation between an aqueous sample and a water-miscible solvent is salt-assisted liquid-liquid extraction (SALLE), where the polarity of the aqueous phase is increased by adding high concentration of salt, leading to phase separation. 55[59]

Novel sample preparation methods
Many common interferents can be removed with conventional sample preparation methods (e.g. protein precipitation, SPE, and LLE), but optimization of these techniques for specic applications is oen complex, time-consuming and frequently involves multiple steps.Many common interferents can be removed with conventional sample preparation methods (e.g. protein precipitation, SPE, and LLE), but optimization of these techniques for specic applications is oen complex, timeconsuming and frequently involves multiple steps.Moreover, some challenges involving very small sample volumes and low abundant analytes remain.If repeated analyses are required from the same sample and if no further sampling is possible, sample preparation sometimes has to be performed using a sample volume as low as a few microliters.Similar difficulties apply to assays for metabolites or biomarkers that are present at very low concentration levels in human samples.Here, the method must be able to pre-concentrate the target substance(s), additionally to removing all other components of the matrix.New developments for sample preparation methods are therefore oen directed towards simplication and possible automation, miniaturization and specicity enhancements of the clean-up process.New developments for sample preparation methods are therefore oen directed towards simplication and possible automation, miniaturization and specicity enhancements of the clean-up process.In the following, the most promising recent developments are briey summarized.

Supported liquid extraction (SLE)
Even though LLE is mostly a very effective sample preparation method, it has limitations, in particular low sample throughput.Several extractions are oen required to improve analyte recovery, sample handling is labor-intensive and timeconsuming, and emulsions can form at the interface between liquid layers.These limitations can be overcome by using supported liquid extraction (SLE), where aqueous samples are adsorbed on a porous solid support material, e.g.diatomaceous earth.Some studies have shown analyte recovery from SLE that was comparable or higher than LLE.SLE has been shown to effectively remove the majority of phospholipids when the extraction conditions were carefully optimized. 61The efficiency of several extraction solvents was also compared for SLE: 7 ethyl acetate removed about 85%, MTBE removed more than 99% of total phospholipids.Dichloromethane removed 99.5% of the phospholipids when used alone; its removal efficiency decreased to 95% when isopropanol was added.However, addition of water-soluble solvents to the samples (e.g., acetonitrile or methanol) prior to SLE extraction led to higher matrix effects for some analytes. 7sopropanol combined with dichloromethane also yielded low concentrations of phospholipids in the extract. 73][64][65][66][67] It was particularly powerful for normal phase separation systems, since the high percentage of organic solvent in the eluate did not need to be evaporated prior to injection into the LC-MS/MS system. 68

Phospholipid removal plates
The use of hybrid precipitation/SPE plates for selective removal of phospholipids and precipitated proteins has been increasing over the past few years. 4,38,69,70Several types of these plates are now commercially available, e.g.Hybrid SPE™ (Sigma Aldrich), Ostro™ (Waters), Captiva™ ND (Agilent) and Phree™ (Phenomenex).These plates have shown very effective extractions of phospholipids compared to PPT. 71 For example, the Hybrid SPE plate specically retains phospholipids by Lewis acid-base interactions between zirconia ionswhich are bonded to the stationary phaseand the phosphate group of the phospholipids.Acetonitrile with 1% formic acid is used as the precipitation agent; formic acid has important inuence on the recovery of the analytes. 72Hybrid SPE extracts have shown to contain signicantly lower phospholipid concentrations as compared to PPT. 4 Ostro uses a combination of protein precipitation and extraction on a C18 sorbent.][75] Other approaches are also possible for removal of phospholipids.A study showed that addition of a colloidal silica suspension together with lanthanum chloride to plasma samples resulted in a reproducible sample clean-up without loss of the analyte of interest. 76

Magnetic beads
Magnetic particles and nanoparticles (MNPs) are becoming increasingly interesting for sample preparation.8][79] They consist of a magnetic core (e.g.Fe 3 O 4 ) coated with a polymer material, to which specic functionalities can be added (Fig. 5). 80Sample preparation steps are similar to SPE (loading, washing and elution).The magnetic particles suspended in solution can be handled as a liquid.Obviously, the big advantage of magnetic beads is that aer sample extraction, the beads are pulled to the tube wall, the supernatant is removed and the wall-bound beads washed with an appropriate solvent.The loaded beads are then re-suspended.
The entire procedure is fast and simple, and complete automation is readily possible.
Several applications have been reported, where either analytes are selectively extracted from a complex matrix [81][82][83] or where the matrix components were removed from the sample, leaving a clean extract behind that can be directly injected into the LC-MS system. 84Using matrix-assisted laser desorption/ ionization (MALDI), the analytes can also be analyzed without having to be eluted from the magnetic beads rst. 85The possible modications on the surface of the magnetic beads are similar to conventional SPE and involve hydrophobic coatings, ion exchange functionalities, molecular imprinted polymers (MIP), 86 restricted access 87 or affinity materials. 88Magnetic particles have also been coated with carbon nanotubes and used to extract aromatic compounds. 89rboow Turboow extraction is usually carried out online before chromatographic separation and uses columns with large particle sizes in conjunction with high ow rates. 90Samples can be directly injected aer dilution; sometimes a protein precipitation step is required before injection.The target analytes are retained in the pores of the column, whereas matrix components are ushed through and discarded directly to waste.The analytes are then eluted from the trapping column using organic solvents.This method has the advantage of fast and generic method development but unfortunately it can also show high carry-over effects. 91A study reported that this technique had no signicant impact on phospholipid removal from serum and plasma samples, and still needed extensive chromatographic separation aer clean-up to avoid matrix effects. 92Other groups reported successful applications for quantication of various substances (drugs, steroids, phenolic compounds, etc.) [93][94][95][96][97][98][99][100][101][102] in human serum, urine and dried blood spots using reversed-phase, ion exchange or mixed-mode materials.

Monolithic spin column extraction
Monolithic spin column extraction is a fast sample preparation method that uses a spin column packed with octadecyl silanebonded monolithic silica as the extraction device. 103The sample is loaded onto the sorbent by centrifugation; the same procedure is performed for washing and elution steps. 104This technique is fast and easy, requires only small amounts of solvents and allows high sample throughput.Unfortunately, the method can only be applied over a limited pH range because of possible degradation of the monolithic silica phase. 104[112]

Microextraction by packed sorbent (MEPS)
This recent sample preparation technique is based on the miniaturization of conventional SPE, using a gas-tight syringe as extraction device.The method is designed for sample volumes from 10 to 1000 mL and can be connected online to LC-MS or GC-MS.Compared to conventional SPE, MEPS is easy to use, faster and needs signicantly lower amounts of organic solvents.Additionally, MEPS sorbents can be used for up to 100 extractions. 113acking materials for MEPS are similar to sorbents used for SPE.Essentially, any sorbent material and functionalization can be applied.For example, silica-based materials (C2, C8, C18), [114][115][116][117][118] with additional ion exchange functionality 119 or even as mixed-mode materials, 120 restricted access materials (RAM), HILIC, carbon, polystyrene-divinylbenzene copolymers (PS-DVB) or molecular imprinted polymers 113,121 have been utilized for MEPS.
The method has been implemented in several recent reports for quantication of pharmaceutical compounds from human biological samples (urine, plasma, oral uid and whole blood), including antipsychotic drugs, 119 cardiac drugs, 114 local anesthetics, 115,121 phenolic acid, 116 immunosuppressants, 117 opioids 120 and antidepressants. 118Recent studies have also reported the successful extraction of trazodone from plasma with polymer nano-bers as the extraction sorbent. 122

Carbon nanotubes
Carbon surfaces have the ability to retain substances by strong hydrophobic interactions.These materials are therefore interesting for reversed-phase extractions of hydrophilic substances.Carbon nanotubes (CNTs) are hollow cylinders that consist of one (single-wall carbon nanotubes, SWCNTs) or several (multiwalled carbon nanotubes, MWCNTs) graphene layers. 123ecause of their large surface areas, CNTs have a high adsorption capacity.They show high affinity towards aromatic compounds that can be adsorbed via p-p interactions. 1246][127] Common target analytes are small, hydrophobic molecules extracted from water samples.Very few applications to biouids have been reported so far.A method for quantitation of diuretics from urine 128 has been published as well as plasma peptide analysis. 129The specicity of the extraction can be enhanced by derivatizing the surface of CNTs with functional groups.A method was recently shown for the determination of anti-inammatory drugs from urine using carboxylated CNTs for sample clean-up. 130To further improve both specicity and handling of the sample clean-up, magnetic CNTs coated with molecular imprinted polymers have been synthetized and used for extraction of BSA from serum samples. 131stricted access materials (RAM) Restricted access materials allow extractive clean-up of biouids by utilizing physical and chemical diffusion barriers.RAM consist of a porous material with a restrictive and hydrophilic outer surface that prevents retention of large interfering molecules such as proteins and phospholipids, combined with smaller inner pores with hydrophobic surfaces that only molecules with low molecular weight can reach. 132This technique is commonly used for online sample clean-up, with the advantage that samples dissolved in almost any solvent can be loaded, even MS incompatible solvents, before elution with the mobile phase used for chromatographic separation.There are two types of RAM phases: 133 internal surface phase (ISP) materials use size exclusion to prevent the matrix components from reaching the inner layer; semi-permeable surface (SPS) materials chemically exclude matrix components by polymeric-or protein coating of the outer layer.In both cases, the inner layer can be functionalized to enhance the specicity of the method. 134Molecular imprinted polymers are a special form of restricted access materials; they are discussed below.
Application of sample clean-up using RAM includes quanti-cation of antimicrobial agents, immunosuppressants etc. from human biological samples prior to LC-MS/MS analysis. 135,1368][139][140] An application was published that reported the synthesis of chiral RAM materials for extraction of enantiomeric drugs from plasma samples. 141munosorbents Immunosorbents use the principle of antigen-antibody affinity for highly specic retention of target substances.The desired antibody is bound to a solid support or gel, which can be used as SPE or micro-SPE sorbent, MEPS or in columns. 123The target analytes can be specically extracted from complex matrices, which allows thorough sample clean-up prior to instrumental analysis.A study has shown that the capacity of monoclonal antibodies was signicantly higher than that of polyclonal antibodies. 142This technique been used as in-tube SPME to quantify interferon a from plasma samples 143 as well as SPE extraction of ProGRP 144 and ochratoxin 145 from serum.Sample preparation techniques with high specicity towards the target analyte are required if the target analyte is present at very low concentration levels or in cases where structurally similar interferents (e.g.isobars) inuence the analysis. 146The immunosorbent extraction usually involves high costs, however, and also requires host animals to grow the required antibodies.Sometimes, the antibodies can be replaced by synthetic alternatives of comparable specicity, such as molecular imprinted polymers or aptamers (see below).

Molecular imprinted polymers (MIPs)
MIPs use the principle of affinity chromatography to maximize the specicity for the analyte(s) of interest.The target analyte or a structurally-related compound is used as a template for the synthesis of the MIP by copolymerization of a complex formed by the template and a functional monomer.The template molecule is then removed, leaving a rigid three-dimensional cavity that is complementary to the target analyte. 147he synthesis of these adsorbents is oen inexpensive and has shown to be fast and reproducible; the materials also have high capacity and can be regenerated and used several times. 148he MIP principle enables highly specic extraction of the target and structurally similar compounds (e.g. a drug and its metabolites) from complex matrices, and pre-concentration of the sample.The specicity of this technique has been shown in several applications.For example, a MIP sorbent developed for tylosin was able to differentiate between tylosin and the closely related narbomycin as well as the remotely similar tylactone.(Fig. 6).Both the target analyte and structurally similar compound were quantitatively extracted, whereas the interfering substance did not show any affinity for the sorbent. 149IP can be used in various forms, for online or off-line processes such as molecular imprinted solid phase extraction (MISPE), 150 magnetic MIP, 151,152 solid-phase micro-extraction (SPME), needle/micropipette tip, dynamic liquid-liquid-solid micro-extraction (DLLSME) or molecular imprinted stir-bar sorptive extraction (MI-SBSE).153,154 This concept has been applied to samples with complex matrices, for example, for benzodiazepines in plasma, 155 nucleoside reverse transcriptase inhibitors in serum, 156 cocaine 157 or ketamine 158 from hair extract, testosterone 159 and tobacco-specic cancer biomarkers 160 from urine.MIP-coated bers for solid phase microextraction (SPME) have also been used for extraction of linezolid from human biouids.161 This technique has shown to provide much cleaner extracts than other sample preparation methods such as LLE.150 Aptamers Another possibility to increase specicity for the target analyte is the use of aptamers immobilized on a solid sorbent for sample preparation. Aptamer are synthetic single-stranded oligonucleotides capable of binding specic analytes with a high affinity through hydrogen bonding, van der Waals forces and dipole interactions.123,162 They are specically prepared for each target molecule; that is, several nucleic acids have to be tested in vitro for each target.Selected nucleic acids with high affinity for the analytes are isolated and amplied using a process called systematic evolution of ligands by exponential enrichment (SELEX).163 The major advantage compared to antibodies is that aptamers can be synthetized directly, without the need for laboratory animals.They can be regenerated within minutes and reused several times. The tecnique has been used for the selective extraction of cocaine from plasma 164,165 and for extraction of tetracyclines from biological uids in combination with ion mobility spectrometry.166 The high affinity of a target substance to an extraction sorbent is clearly shown in these applications as well as the importance of the sequence of the oligonucleotides.The sequence is specic for a particular compound and will become inactive if the oligonucleotides are graed in a randomized order.163 Recoveries of up to 90% conrm the high specicity of this technique, even in complex samples such as plasma.163 Aptamers have been immobilized on polymeric nano-bers and extraction of thrombin from serum was shown.167

Conclusion
Common problems encountered during development of an LC-MS/MS assay for the quantication of small molecules from biological samples include loss of sensitivity and specicity due to matrix effects.Sample preparation is therefore an indispensable part of the analytical workow.The possible inuence of matrix effects on LC-MS/MS assays has been extensively studied several methods have been published to identify and avoid these effects.Considerable progress has been made in the improvement of sample preparation routines in the last few years.New trends are directed towards either increasing the specicity of the extraction for the target analyte or removing as much of the matrix components as possible.Miniaturization and automation of these techniques are on-going efforts, leading to cheaper, more robust and fully automated LC-MS/MS assays that will signicantly impact pharmaceutical analyses of biouids in the future.

Fig. 1
Fig. 1 Chemical structures of the two most important groups of phospholipids.

Fig. 2
Fig. 2 Injection of extracted blank human plasma (+0.2 mL triethylamine, blue and red traces) with an overlay of the control sample (20 ng mL À1 , grey and green traces) containing piperaquine (PQ) and internal standard (d 6 -PQ) during post-column infusion at 10 mL min À1 of PQ and d 6 -PQ (1.2 ng mL À1 ).Electrospray ionization of the analytes was performed in positive ion mode; the MRM transitions were m/z 535 / 288 and m/z 541 / 294 for PQ and D6-PQ, respectively (reprinted with permission from ref. 19).

Fig. 4
Fig. 4 Extraction of C16:0 lysophosphatidylcholine (C16:0 lyso-PC) from human plasma using liquid-liquid extraction with three different solvents at different pH values.Comparison to solid-phase extraction and two commercial phospholipid removal sorbents (PR-plate 1 and PRplate 2).Lyso-PC was monitored using the following MRM transition: m/z 496 / 184 (reprinted with permission from ref. 53).

Fig. 5
Fig. 5 Assembly of polymers onto the surface of magnetic nanoparticle cores (reprinted with permission from ref. 168).