Introduction

Sample pretreatment is one of the most critical steps for drug screening by gas chromatography–mass spectrometry (GC–MS) and liquid chromatography–mass spectrometry (LC–MS) to efficiently detect a wide range of drugs and poisons. Protein precipitation (PP), liquid–liquid extraction (LLE), and solid-phase extraction (SPE) have been used for these purposes [1, 2]. Simple PP has become a widely used method, especially for LC–MS–MS screening [3, 4], and more time-consuming preparation methods, such as specific LLE for basic or acidic drugs [3, 5–7] and various types of SPE [8–11], are often necessary to remove interfering impurities prior to GC–MS and LC–MS–MS analyses and to avoid the risk of confounding matrix effects, especially in LC–MS procedures. Therefore, simplified extraction procedures that can simultaneously detect acidic and basic drugs with less interference by contaminating impurities are desirable. Furthermore, if drug confirmation and rough estimation of drug levels can be simultaneously performed by the initial screening procedure, adequate treatments of patients or forensic diagnosis of poisoning can be performed more smoothly.

Recently, we developed a unique GC–MS screening method to estimate levels of toxic substances without the need for reference standards by constructing calibration-locking databases for various types of drugs [12–14] and pesticides [15] using the NAGINATA software (Nishikawa Keisoku, Tokyo, Japan). This module was originally designed for system quality control and data analysis support as an add-on to the Chemistation software package for use with Agilent 6890GC/5973MS and 7890GC/5975MSD instruments. A system performance evaluation using a criterion sample mixture solution can compensate for instrument-to-instrument and day-to-day conditional variations before sample measurements. An automatic compound search based on the calibration-locking database, including retention times, calibration curves, and electron impact ionization (EI) mass spectra of compounds, can be performed after sample recording, followed by quick confirmation and tentative quantification of all compounds to give analytical results without preparing the respective reference standards. This system has led to significant improvements in accuracy and speed of drug identification, and has been adopted by many forensic and clinical laboratories in Japan. However, because the calibration curve of each drug is produced using reference standards, tentative drug concentrations in samples show significant dependence on pretreatment procedures. Therefore, the recovery rate of each drug at every pretreatment step critically influences the reliability of the estimated level of the drug. For example, we used Focus columns (Agilent, Santa Clara, CA, USA) for drug extraction, and found that the recovery rates of acidic drugs were very low [13]. Therefore, to overcome such problems, we present a rapid and cost-effective pretreatment method to simultaneously extract basic and acidic drugs for our NAGINATA drug screening system [13] in this report.

Materials and methods

Chemicals and reagents

All drugs and poisons tested in this study are listed in Table 1. Standard drugs and metabolites were obtained from pharmaceutical or chemical manufacturers. Amphetamine sulfate was a generous gift from the Department of Forensic Medicine, Fukuoka University School of Medicine (Fukuoka, Japan). Tetrahydrocannabinol was kindly provided by Emeritus Prof. I. Yamamoto (Hokuriku University, Kanazawa, Japan). Diazepam-d 5 was purchased from Sigma-Aldrich (St Louis, MO, USA); a QuEChERS prepacked extraction powder (containing 6 g magnesium sulfate and 1.5 g sodium acetate), dispersive SPE kit [containing 25 mg primary secondary amine (PSA) and/or 25 mg end-capped octadecylsilane with 150 mg magnesium sulfate], Focus columns, Captiva ND Lipids cartridges, four different silica-based C18 SPE columns (SPEC C18, SPEC C18AR, Bond Elut C18, and Bond Elut C18 EWP) and alumina SPE columns (Bond Elut Alumina-N) from Agilent; frozen human whole blood from Biopredic (Rennes, France); LC–MS grade acetonitrile, methanol, and trifluoroacetic acid (TFA) from Wako Pure Chemical (Osaka, Japan); pyridine (silylation grade) from Thermo Scientific (Bellefonte, PA, USA). All other chemicals were of analytical reagent grade and commercially available.

Table 1 Tentative concentration values estimated by the NAGINATA screening system and measured recoveries for 65 drugs
Full size table

Standard solutions

Standard stock solutions of each analyte were prepared in 100 or 80 % methanol at a concentration of 1 mg/ml as free bases and stored at −20 °C in darkness. Mixed analyte working solutions were prepared by serial dilution of the stock solutions with methanol. As an internal standard, diazepam-d 5 was prepared in acetonitrile at a concentration of 10 μg/ml.

Preparation of blood samples

Step 1: extraction/partitioning

For extraction, 0.5 ml of whole blood was diluted with 1 ml of distilled water in a 15-ml centrifuge tube, and 1.5 ml of acetonitrile and 50 μl of diazepam-d 5 solution (10 μg/ml in acetonitrile) were added. The mixture was vortexed for 10 s, and two stainless steel beads (diameter 4 mm) and 0.5 g of the QuEChERS prepacked extraction powder were added to the solution, which was then vigorously shaken several times by hand, vortexed for 10 s, and centrifuged at 850 g for 10 min.

Step 2: removal of matrix interference

The acetonitrile upper layer was passed through the Captiva ND Lipids cartridge which was prewashed with 1 ml acetonitrile under vacuum. To the eluate, 0.1 ml of 0.1 % TFA in acetonitrile was added, and the solvent was evaporated to dryness under a nitrogen stream at 60 °C. The residue was dissolved in 50 μl of pyridine and 50 μl of acetic anhydride, which was then heated at 60 °C for 30 min for acetylation. The organic liquid was evaporated to dryness under a stream of nitrogen at room temperature. The derivatized residue was dissolved in 100 μl of 1-chlorobutane, and a 2-μl aliquot of the solution was injected into the GC–MS apparatus.

GC–MS conditions

The apparatus used for analysis was a combined Agilent 7890A gas chromatograph and 5975C mass spectrometer. Separation was achieved on an Agilent HP-5 ms fused-silica capillary GC column (30 m × 0.25 mm i.d., 0.25 μm film thickness) coated with 5 % phenyl methyl silicone as stationary phase. The splitless injection mode was selected with a valve-off time of 2 min. The GC–MS conditions were: oven temperature, initially 60 °C (2-min hold) increased up to 300 °C at 20 °C/min (10-min hold); total run time, 24 min; injection port and transfer line temperatures; 250 and 280 °C, respectively; carrier gas, helium with constant pressure mode. Decafluorotriphenylphosphine tuning was performed to obtain a uniform mass spectrum. The retention times were fixed using the retention-time locking technique with diazepam-d 5 as the locking compound. The retention time of diazepam-d 5 was set at 13.923 min and the full-scan mode (scanning range 50–550 Da) was used.

Analysis of spiked samples by GC–MS using NAGINATA software

We prepared blood samples (n = 6) spiked with 65 drugs, of which 55 drugs were prepared at concentrations of 0.1 μg/ml (low) and 1 μg/ml (high), and 10 drugs were prepared at 1 μg/ml (low) and 10 μg/ml (high) depending on their reported therapeutic and toxic blood levels. These spiked whole blood samples were then analyzed using our newly developed method. NAGINATA screening was performed using our drug database with acetylating reagents as reported previously [12]. The ability to detect drugs expressed as confirmation class from no mark to +++++ mark and the reliability of tentative concentrations were evaluated. The recoveries of the drugs were carefully measured by comparing the average peak areas of whole blood extracts with those of the reference standards added to the blank extracts [6].

Application to a poisoning case

Whole blood samples from a forensic case, in which multiple drug poisoning was suspected, were analyzed via three different procedures: NAGINATA screening using a Focus column [13], NAGINATA screening with the present newly established extraction method, and conventional quantitative analysis by making calibration curves using standard compounds [6]. The three concentration values were then compared.

Results and discussion

Extraction step

To extract a wide range of drugs and poisons, we first considered a salting-out extraction step using acetonitrile via the QuEChERS method, which was originally developed by Anastassiades et al. [16] for the analysis of pesticide residues in fruits and vegetables. Since then, this technique has been widely applied in the field of pesticides analysis [17–19]. However, its applications in the forensic analysis of blood samples remain limited. In 2006, Plössl et al. [20] evaluated the QuEChERS method by GC-ion trap-MS in a validation study of 8 drugs spiked into 1-ml aliquots of whole porcine blood, and found that the method was useful. Usui et al. [21, 22] developed a rapid extraction method for drugs and pesticides in 0.5 ml of whole human blood using the modified QuEChERS method prior to LC–MS–MS analysis. More recently, a simple one-pot extraction method for 13 drugs and metabolites in a small volume of whole blood (0.1 ml) was reported by Matsuta et al. [23] using a modified QuEChERS method .

Because we used the scan mode for GC–MS screening, a relatively large volume of whole blood (0.5 ml) was required as a sample, and the extraction method reported by Usui et al. [21] was modified. In the present study, deproteinization with acetonitrile followed by salt addition was found to give a constant volume of the upper layer, and recoveries of drugs and metabolites were generally satisfactory. Dilution of the whole blood samples with distilled water was essential to protect the blood from clotting, which is known to decrease recoveries. As shown in Fig. 1, a clear and colorless upper layer could be obtained.

Fig. 1
figure 1

Photograph showing good separation of clear and colorless upper acetonitrile layer after the salting-out step via the QuEChERS method

Full size image

Removal of interfering impurities from the matrix

Detection of drugs at high concentrations (>0.5 μg/ml) in whole blood was possible with only salting-out extraction with acetonitrile described above, without the need for further purification steps. However, when we tried to detect drugs at low concentrations (≤0.1 μg/ml), impurities in the matrix sometimes interfered with the detection; therefore, a further purification step was needed. The dispersive SPE using PSA and/or C18 in the QuEChERS method was first examined for the removal of matrix interferences. Recoveries of acidic drugs were significantly decreased when PSA was used for purification. When we used C18, better recoveries of acidic drugs were observed. However, a small amount of salts dissolved in the acetonitrile precipitated when the solvent was evaporated to dryness. Salts were considered to interfere with the consistency of drug detection by GC–MS, especially for acetylated compounds. Therefore, we examined the use of various SPE cartridges for purification. Among the six different SPE cartridges tested (four silica-based C18 columns: SPEC C18, SPEC C18AR, Bond Elut C18, and Bond Elut C18 EWP; Bond Elut Alumina-N column and the Captiva ND Lipids cartridge), the number of detectable drugs at low concentration (0.1 μg/ml) was greatly increased when the Captiva ND Lipids cartridge was used for purification. The Captiva ND Lipids cartridge is a filter for removal of protein and phospholipids, and has been recently introduced into the market for LC–MS–MS analysis. Similar products are currently available from several manufacturers, including the Ostro Sample Preparation Products (Waters, Ann Arbor, MI, USA), the Phree Phospholipid Removal System (Phenomenex, Torrance, CA, USA), and the Supelco Hybrid SPE Phospholipid Removal System (Sigma-Aldrich). According to these manufacturers, such filters can efficiently remove phospholipids from serum and plasma samples, thereby significantly reducing matrix effects prior to LC–MS–MS analysis. Although no applications using whole blood samples for toxicological screening have been reported so far, three anticancer drugs (imatinib, nilotinib, and dasatinib) in dried blood samples were successfully analyzed by LC–MS–MS following a preparation step with the Captiva ND Lipids cartridges [24]. Figure 2 shows the total ion current chromatograms of acetylated extracts from the blank whole blood sample before and after purification by the Captiva ND Lipids cartridge; significant decreases in interfering peaks, including cholesterol, were observed. Therefore, this purification step was found useful for analysis of whole blood samples after a salting-out extraction step using acetonitrile. TFA was added to the filtrate to protect the drugs with high volatilities, such as methamphetamine and nicotine, from evaporation. Because the established method can extract a wide range of drugs, including more polar compounds as compared with our previously reported Focus column extraction method [13], 1-chlorobutane, which is known to dissolve various types of drugs and metabolites [25], was selected as the final dissolution solvent instead of ethyl acetate.

Fig. 2
figure 2

Total ion current chromatograms of the acetylated extract from blank human whole blood before (a) and after (b) purification with a Captiva ND Lipids cartridge

Full size image

Qualification and tentative quantification of 65 drugs using the improved NAGINATA screening system

The tentative concentrations obtained by the NAGINATA screening and the actual measured recovery rates of 65 drugs at two different concentrations are shown in Table 1. At high concentrations, variation coefficients were not larger than 18 % for all drugs including acidic and basic drugs, and the recovery rates were not less than 56 % except those of diclofenac and ibuprofen. The tentative concentrations obtained from the drug database with acetylating reagents were in the range of 52.5–168 % of the expected values. Similar results were obtained at low concentrations, although the range of tentative concentrations was somewhat wider than at high concentrations. Therefore, more reliable estimation of drug and poison levels has become feasible upon screening without the need for reference standards by the present study.

Application to an actual case

Whole blood from a forensic autopsy case of multiple drug poisoning was analyzed via the NAGINATA screening following two kinds of extraction procedures (Focus column method and present method) to give tentative concentrations of 13 drugs. Figure 3 shows a comparison of these results with the concentrations obtained by the conventional quantitative method [6] using reference standards. We observed a large difference between the tentative concentrations obtained by the Focus column method and the actual concentrations for acetaminophen, bromvalerylurea, and caffeine because of the low recovery rates of acidic drugs for the Focus column. On the other hand, the tentative concentrations obtained by the present method (QuEChERS-Captiva) were similar to those obtained by the conventional quantification for all drugs. In this case, the tentative sertraline concentration reached the fatal level; thus, rapid diagnosis of drug intoxication was possible even at the screening.

Fig. 3
figure 3

Tentative concentrations of 13 drugs in whole blood obtained by the NAGINATA screening system following two different extraction procedures and actual quantification values using the conventional analytical method in a forensic autopsy case

Full size image

Conclusions

In this study, we developed a simple preparation procedure to extract a wide range of acidic and basic drugs. By combining this extraction process with the NAGINATA drug database system, it was possible to acquire tentative but reliable concentrations at the screening step. Because this method is very simple and has excellent reproducibility, it is likely to prove useful and applicable to a wide range of drugs for GC–MS and LC–MS–MS analyses.