Introduction

Cyanide and azide, which are toxic inorganic anions, are important target compounds to be analyzed in forensic toxicology, because they are used for not only suicide attempts but also criminal purposes [1,2,3,4,5,6]. Cyanide and its metabolite, thiocyanate, in biological fluids are commonly analyzed by gas chromatography–mass spectrometry (GC–MS) after derivatization with pentafluorobenzyl bromide (PFBBr) using extractive alkylation [7,8,9]. A new LC–MS-MS method for analyzing cyanide and thiocyanate in swine plasma after derivatization with naphthalene-2,3-dicarboxaldehyde and taurine for cyanide and with monobromobimane for thiocyanate has also been reported [10]. In contrast, azide in human blood and urine samples is analyzed by GC–MS after derivatization with PFBBr either as a one-phase reaction in acetone and water solution [6, 11] or as a two-phase reaction in ethyl acetate and water solution using phase transfer catalyst [12]. Since plasma samples are not always available in forensic practice, a simple method that can simultaneously determine cyanide, thiocyanate, and azide in whole blood is desirable. However, the reported extractive alkylation methods have some drawbacks, such as the narrow range of linearity of the calibration curve and rapid damage to the GC–MS column. In this study, we examined each step of the reported methods [7, 12] and attempted to establish a more reliable method to simultaneously determine the levels of the above compounds in human whole blood.

Cyanide and azide are extremely dangerous poisons, the handling and storage of which should be performed with extreme caution. It is thus sometimes difficult to keep these authentic samples in small analysis rooms, which do not have good severe regimes for hazardous substances. Recently, we developed a unique GC–MS screening method to estimate the level of intoxicants without the need for reference standards by constructing calibration-locking databases [13, 14] using a NAGINATA software (Nishikawa Keisoku, Tokyo, Japan). We, therefore, examined the applicability of the established method to NAGINATA–GC–MS screening, which can quickly identify the anions and give approximate concentrations without using toxic anions as reference standards.

Materials and methods

Chemicals and reagents

Cyanide ion reference standard solution (1000 mg/L) was purchased from Kanto Chemical (Tokyo, Japan); reference standard solutions of potassium thiocyanate (100 µmol/mL) and sodium azide from Wako Pure Chemical Industries (Osaka, Japan); isotopically labeled potassium cyanide, K13C15N (13C 99%, 15N 98% +), from Cambridge Isotope Laboratories (Abdower, MA, USA); 1,3,5-tribromobenzene (TBB), ascorbic acid, sodium tetraborate, and trichloroacetic acid (TCA) from Wako Pure Chemical Industries; an alkylating agent, PFBBr, from Aldrich (Milwaukee, WI, USA); the phase transfer catalyst, tetradecyldimethylbenzylammonium chloride (TDMBA), from Tokyo Kasei Kogyo (Tokyo, Japan); diazepam-d 5 from Sigma-Aldrich (St. Louis, MO, USA); an ethyl benzoic sulfonic silica gel column (Bond Elut-SCX, 50 mg) from Agilent Technologies (Santa Clara, CA, USA); and frozen human whole blood obtained from healthy individuals from Biopredic International (Rennes, France). All other chemicals were of analytical reagent grade, commercially available.

Preparations of standard solutions and reagents

Oxygen-free water was used throughout this study and was prepared by bubbling nitrogen into distilled water for 15 min. Reference standard solutions of cyanide (1000 mg/L = 38.43 µmol/mL) and thiocyanate (100 µmol/mL) were diluted with 0.1 M NaOH and oxygen-free water, respectively, to give a 20 µmol/mL (= 0.2 µmol/10 µL) solution. A reference standard solution of azide [20 µmol/mL (= 0.2 µmol/10 µL)] was prepared by dissolving sodium azide in a 0.1 M NaOH solution. These solutions were further diluted with 0.1 M NaOH or oxygen-free water to prepare the samples for making calibration curves. Another series of standard solutions were prepared for making quality control (QC) samples in the same way. Standard solution of internal standard (IS)1 [K13C15N, 4 µmol/mL (= 0.04 µmol/10 µL)] was prepared by dissolving K13C15N in a 0.1 M NaOH solution. The solution of phase transfer catalyst (TDMBA, 5 mM) was prepared in a saturated solution of sodium tetraborate (pH 9.2). Derivatizing agent (PFBBr) and IS2 (TBB) were prepared in ethyl acetate at concentrations of 20 mM and 10 µM, respectively. Ten % TCA solution and 0.5 M ascorbic acid solution were prepared in oxygen-free water. These solutions were stored in a refrigerator at 4 °C. All of the solutions were stable for at least 6 months, except for the solutions of PFBBr and ascorbic acid, which were prepared every 2 months.

Preparation of whole blood samples

Samples to be tested were prepared by adding 10 µL of each of the reference standard solutions of cyanide, thiocyanate, or azide to a 0.2-mL blank of whole blood.

Examination of each step of the reported papers [7, 12]

Deproteinization method

We examined various kinds of acids (TCA, perchloric acid, and trifluoroacetic acid) and solvents (methanol, ethanol, acetone and acetonitrile) to find the most suitable deproteinization agent.

Internal standard

Benzyl derivatives of cyanide, thiocyanate and azide and isotopes of potassium cyanide (K13C15N) were examined as candidates of IS compounds, and the ranges of linearity of the calibration curves were compared with those of the method using TBB as an IS [7, 12].

Agent(s) causing the column damage

We tested whether the derivatization reagent and the phase transfer catalyst TDMBA caused the column damage.

Effect of ascorbic acid

Seto et al. [15] reported that human oxyhemoglobin (Hb-O2) converted thiocyanate to cyanide under acidic conditions, which might increase the concentration of cyanide in our analysis. They also suggested that this conversion could be suppressed by the addition of ascorbic acid. We analyzed whole blood samples containing 1 mM thiocyanate under acidic conditions, with and without the addition of ascorbic acid, and examined the cyanide levels.

Extraction procedure

Step 1

Based on the detailed examinations of each step, the following procedure was established: a mixture of 0.5 mL of 20 mM PFBBr solution in ethyl acetate, 1.0 mL of 10 µM TBB (IS2) in ethyl acetate and 0.8 mL of 5 mM TDMBA solution in oxygen-free water saturated with sodium tetraborate was prepared in a 15 mL centrifuge tube.

Step 2-1

For the analysis of cyanide, 0.2 mL of whole blood sample was mixed with 10 µL of IS1 (K13C15N, 0.04 µmol) solution, 200 µL of 10% ice-cold TCA solution and 50 µL of 0.5 M ascorbic acid solution, and the mixture was vortex-mixed for 10 s and centrifuged at 6700 g for 5 min. The supernatant was added to the above step 1 mixture, and the preparation was vortex-mixed for 30 s.

Step 2-2

For the analysis of thiocyanate and azide, another 0.2-mL whole blood sample containing thiocyanate and/or azide was directly added to the above step 1 mixture without the deproteinization step, and vortex-mixed for 30 s, followed by step 3.

Step 3

The mixture was heated at 60 °C for 30 min in a heat block, cooled down and then centrifuged at 1500 g for 10 min. Next, 0.5 mL of the upper layer was put in a Bond Elut-SCX column (50 mg) set on an empty 15-mL centrifuge tube, and passed through it by centrifugation at 400 g for 5 min. The eluent was dried with sodium sulfate (about 0.2 g), and a 1-µL aliquot of the solution was injected into the GC–MS apparatus. The flowchart of the procedure is shown in Fig. 1.

Fig. 1
figure 1

Improved extraction and alkylation procedure (underlines indicate improvements). PFBBr pentafluorobenzyl bromide, TBB 1,3,5-tribromobenzene, TDMBA tetradecyldimethylbenzylammonium chloride, IS internal standard, TCA trichloroacetic acid, GC–MS gas chromatography–mass spectrometry, GC–MS gas chromatography–mass spectrometry

Full size image

Conditions of GC–MS

The conditions of GC–MS that can separate the azide derivative from the derivatization reagent PFBBr were determined as follows. The apparatus used for the analysis was a combination of an Agilent 7890B gas chromatograph and an Agilent 5977B mass spectrometer with an Agilent HP-5MS UI fused-silica capillary GC column (30 m × 0.25 mm i.d., 0.25 µm film thickness), coated with 5% phenyl methyl silicone as a stationary phase (Agilent Technologies). The splitless injection mode was selected with a valve-off time of 2 min. The GC–MS conditions were as follows: oven temperature, initially 40 °C (2-min hold) increased up to 100 °C at 10 °C/min, and then increased up to 300 °C at 20 °C/min (5-min hold); total run time, 23 min; injection port and transfer line temperatures, 250 and 280 °C, respectively; and carrier gas, helium with the constant pressure mode. 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 17.919 min, and the full-scan mode (scanning range, m/z 50–400) was used. The quantifier and qualifier ions for each compound were as follows: cyanide (m/z 157, 188), IS1 (13C15N, m/z 159, 190), thiocyanate (m/z 239, 211), azide (m/z 194, 223), and IS2 (TBB, m/z 314, 235).

Calibration, precision and accuracy

The calibration samples containing cyanide (0.005, 0.01, 0.02, 0.05, 0.1, 0.5, and 1 µmol/mL) or thiocyanate and azide (0.005, 0.01, 0.02, 0.05, 0.1, 0.5, and 1 µmol/mL) were prepared by adding the reference standard solution of each anion to 0.2 mL of blank whole blood. These samples were extracted and derivatized in the same manner as described above. A calibration graph for cyanide was obtained by plotting the peak area ratio of the cyanide derivative to the IS1 (13C15N−) derivative versus the cyanide concentration. A calibration graph for thiocyanate or azide was obtained by plotting the peak area ratio of the thiocyanate derivative or the azide derivative to IS2 (TBB) versus the concentration of thiocyanate or azide. The limit of detection (LOD) for each anion was estimated at a signal-to-noise ratio equal to 3 in spiked whole blood samples. QC samples (n = 5) were prepared at concentrations of 0.05, 0.2, and 0.8 µmol/mL for cyanide and thiocyanate, and 0.05, 0.2, and 0.4 µmol/mL for azide, and analyzed as described above. Intraday precision as relative standard deviation (%) and accuracy as relative error (%) were calculated on the basis of the prepared calibration curves within 1 day analysis. Interday precision and accuracy of the method were calculated by using the data of QC standards obtained on three separate days.

Recovery and stability

The three sets of QC standard samples (n = 3 at each concentration) were prepared in human whole blood for cyanide and thiocyanate (0.05, 0.2 and 0.8 µmol/mL) and azide (0.05, 0.2 and 0.4 µmol/mL), and the samples were analyzed as described above.

The recovery of each anion was calculated by comparing the peak area of the derivative of each anion in the above QC samples with that in a blank blood extract spiked with each anion just before the derivatization step.

The preliminary experiment was carried out using QC samples (n = 2 for each condition) at the three concentrations described above, in order to examine the stability of the cyanide, thiocyanate, and azide in whole blood samples. The samples were analyzed immediately, after storage at room temperature, 4 and − 20 °C for 7 days, and after three sessions of freezing and thawing (FT) within 7 days. The peak area ratio of the derivative of each anion to IS was used for comparison. The stability of the derivatives was also examined during the 24 h of storage at room temperature and the 7 days of storage in a refrigerator at 4 °C.

Registration of each set of data to the ‘‘inorganic anion database’’

Decafluorotriphenylphosphine tuning was carried out to obtain a uniform mass spectrum. IS2 (TBB) was used to make a calibration curve for all anions including cyanide, because TBB is readily available at laboratories, and NAGINATA software can register not only a straight line but also a curved one. The same target (quantifier) and qualifier ions described above were used for each compound, and calibration curves were obtained using MSD ChemStation F.01.03.2357 (Agilent Technologies) by plotting the peak area ratio of the compound to IS2 versus the amount of compound. The retention time and mass spectrum of each anion were obtained at the level of 5 ng injection to the column. The data obtained for each anion, such as retention time, qualifier ion/target ion ratio, mass spectrum, and calibration curve (values of slope and intercept), were registered using NAGINATA software (Nishikawa Keisoku) as the “inorganic anion database.”

Results and discussion

Extractive alkylation procedure

We first attempted to analyze all three anions simultaneously. However, the analysis of cyanide in whole blood was possible only when the blood was deproteinized with TCA, whereas thiocyanate and azide were not detected after deproteinization with TCA. We, therefore, decided to analyze cyanide separately from thiocyanate and azide. Because the recovery rates of thiocyanate and azide were significantly decreased with all kinds of acids and solvents, it was determined that these compounds should be separately analyzed without a deproteinization step. The calibration curves of cyanide tended to lose their straightness when benzylcyanide and TBB were used as ISs, which caused a narrow range of linearity of these curves. This problem was overcome by using K13C15N as the IS, which enabled good linearity of the calibration curve to be obtained. The calibration curves of azide and thiocyanate were linear using the respective benzyl derivative as well as TBB as the IS. Because lesser background peaks were observed with the ions of TBB as compared with those of benzyl derivatives, TBB was selected as an IS for thiocyanate and azide determination.

In the previous papers by Kage et al. [7, 12], quantitative determination of cyanide, thiocyanate and azide was carried out by gas chromatography with electron capture detection (ECD). Since ECD is highly sensitive, the upper layer obtained in an extractive alkylation step was submitted to GC after 20-fold dilution with hexane. When we attempted to determine these anions by GC–MS by injecting the upper layer without dilution, significant damage to the GC–MS column was observed even after only 20 injections of the samples, and the peak shape of the derivative of each anion became broad. We searched for the agent causing this column damage, including a derivatization reagent and the phase transfer catalyst, TDMBA. The damage to the column was found to be caused by the phase transfer catalyst moving into the organic layer and remaining in the column. After several attempts to remove TDMBA before injection into the column, we finally succeeded in eliminating it by passage through an ethyl benzoic sulfonic silica gel column (Agilent Bond Elut-SCX, 50 mg). By this treatment, the column was well protected, and sharp and symmetrical peaks of target compounds could be obtained even after 200 injections of the samples.

About 2% thiocyanate was found to convert to cyanide during deproteinization with TCA followed by an extractive derivatization step. This conversion was successfully suppressed by adding ascorbic acid, as described by Seto [15]. In the analysis of human saliva, Paul and Smith [9] reported that the formation of cyanide from thiocyanate was suppressed by using tetrabutylammonium sulfate (TBAS) as the phase transfer catalyst instead of TDMBA. Bhandari et al. [8] also used TBAS for the analysis of cyanide and thiocyanate in plasma samples. When we used TBAS for cyanide analysis in whole blood samples following a deproteinization step with TCA, many interfering peaks appeared on the chromatograms (unpublished observation). Therefore, the use of TDMBA with ascorbic acid was considered superior for the analysis of cyanide in whole blood samples. For the azide and thiocyanate analyses, ascorbic acid was not used, because azide was unstable in the presence of ascorbic acid. The thiocyanate in whole blood was stable even if ascorbic acid was not added during the extractive alkylation procedure. A sharper cyanide peak was obtained by drying the organic layer with sodium sulfate at the final step. The established method is summarized in Fig. 1. Because this method is simple and does not require a condensation step, we could easily use it to treat more than 40 samples in one day.

GC–MS analysis

Figure 2 shows mass spectra of the pentafluorobenzyl (PFB) derivatives of cyanide, IS1, thiocyanate, azide, and IS2. The molecular ions of the derivatives of cyanide, IS1, thiocyanate, azide, and IS2 were observed at m/z 207, 209, 239, 223, and 312, respectively. These molecular ions and fragment ions were carefully examined, and target (quantifier) and qualifier ions for each anion were selected. Figure 3a shows extracted ion chromatograms of the alkylated extracts from whole blood containing 0.5 µmol/mL cyanide. Sharp and symmetrical peaks of the derivatives of cyanide and IS1 were observed, at retention times of 8.346 and 8.347 min, respectively. Extracted ion chromatograms of the alkylated extracts from whole blood containing 0.5 µmol/mL thiocyanate and azide are shown in Fig. 3b. Sharp and symmetrical peaks of the derivatives of thiocyanate, azide, and IS2 were observed, at retention times of 7.635, 10.562, and 11.932 min, respectively.

Fig. 2
figure 2

Mass spectra of pentafluorobenzyl (PFB) derivatives of cyanide, thiocyanate and azide, together with ISs

Full size image
Fig. 3
figure 3

Extracted ion chromatograms of the PFB extracts from a quality control whole blood sample containing 0.5 µmol/mL cyanide (a) or 0.5 µmol/mL azide plus 0.5 µmol/mL thiocyanate (b), together with ISs

Full size image

Validation and stability

The calibration curve of each compound was linear in the concentration range from 0.01 to 1.0 µmol/mL for cyanide (0.26–26.0 µg/mL), 0.01–1.0 µmol/mL for thiocyanate (0.58–58.1 µg/mL), and 0.005–0.5 µmol/mL for azide (0.21–21.0 µg/mL), with a correlation coefficient of 0.999. These ranges can cover the toxic to fatal levels of these anions [2, 4]. The limits of detection were 0.005 (0.13 µg/mL) and 0.002 µmol/mL (0.08 µg/mL) for cyanide and azide, respectively. In our study, scan mode was used for quantification and confirmation of the presence of the poisons at the same time. The sensitivity could be increased by using the selected ion monitoring mode. There was a small peak of the PFB derivative of thiocyanate in blank whole blood at the concentration range of 0.005–0.006 µmol/mL, probably derived from food. Therefore, a standard addition method is recommended when a small amount of thiocyanate in whole blood of a healthy person needs to be determined.

The results of the intraday and interday precision and accuracy are summarized in Table 1. The precision, as measured by percent relative standard deviation, was not larger than 4.3% in the intraday assay analysis, and not larger than 9.1% in the interday assay analysis. The accuracy bias was within ± 5.9% of the nominal concentration in both assays. Thus, reliable quantification of each anion in whole blood became feasible. The recoveries of cyanide and thiocyanate at three concentrations, 0.05, 0.2, and 0.8 µmol/mL, were 72, 55, and 54% for cyanide, and 93, 78, and 91% for thiocyanate, respectively. The recoveries of azide at three concentrations, 0.05, 0.2, and 0.4 µmol/mL, were 93, 87, and 96%, respectively.

Table 1 Precision and accuracy of the method
Full size table

Our preliminary experiments showed that cyanide in whole blood was stable at room temperature, 4, and − 20 °C for 7 days, and after three sessions of FT within 7 days. Many studies on the stability of cyanide in biological samples have been performed [8, 10, 16,17,18]. Bhandari et al. [10] reported that cyanide in spiked plasma was stable only for 2 days at − 80 and − 20 °C and was quickly removed at 4 °C. Another study showed that cyanide in spiked whole blood was stable over a 2-week period at 4 and − 20 °C [16]. These findings may indicate that cyanide in whole blood is more stable than that in plasma. Because these findings were obtained with spiked samples, a stability experiment using real samples is considered to be necessary. Thiocyanate in whole blood was stable at all conditions examined, which is in agreement with the previously published data [8, 10]. Azide in whole blood was stable at 4 and − 20 °C for 7 days, and after three sessions of FT within 7 days, but was unstable at room temperature; a 27–46% decrease of azide was observed in our experiment. Ohmori et al. [11] reported that azide in spiked whole blood was stable for up to 3 days when stored at − 30 °C, but was decreased to 48% when stored at 4 °C, probably due to oxyhemoglobin. Further experiments using real samples are also needed to determine the stability of azide in whole blood samples.

The PFB derivatives of three anions were very stable, and no decrease of each anion was observed during the storage at 24 h at room temperature and during 7 days of storage in a refrigerator at 4 °C. Therefore, the samples did not deteriorate during the analysis.

Qualification and tentative quantification of cyanide, thiocyanate and azide using the NAGINATA screening system

The three anions were rapidly identified using NAGINATA software, and the approximate concentration of each compound in whole blood was obtained at the same time. Figure 4 shows the operating screen of the “NAGINATA™ browser” for a whole blood sample containing azide. The left windows include extracted ion chromatograms (Fig. 4a, b) and the total ion chromatogram (Fig. 4c); the upper right windows show actual (Fig. 4d), reconstructed (Fig. 4e) and expected mass spectra (Fig. 4f) of the peak, and the lower right window (Fig. 4g) presents a combined extracted ion chromatogram of the four ions with the highest abundance. Via this screen, the existence of azide in whole blood could be easily confirmed. Because a large volume of an alkylating agent, PFBBr, was completely separated from the derivative of azide, another step for removing an excess amount of PFBBr carried out in a previous study [6] was not necessary. Table 2 shows the tentative concentration values of cyanide, thiocyanate and azide in QC samples obtained by NAGINATA screening. Although QC samples were analyzed on two different days, the values were similar to the spiked amounts. Thus, tentative concentration values obtained based on the calibration curve in the database were considered to be helpful to estimate the levels of the compounds for toxicity at the stage of screening.

Fig. 4
figure 4

Operating screen of the “NAGINATA™ browser” for a whole blood sample containing azide. The left windows include extracted ion chromatograms (a, b) and the total ion chromatogram (c); the upper right windows show actual (d), reconstructed (e) and expected mass spectra (f) of the peak; and the right bottom window presents combined extracted ion chromatogram (g) of the four ions with the highest abundance

Full size image
Table 2 Tentative concentration values of cyanide, thiocyanate and azide obtained by NAGINATA screening in spiked samples
Full size table

Practical application

A man in his 70s was found dead in a car at a parking lot. He had been troubled about his job. An autopsy was performed by the judicial request approximately 30 h after the estimated time of death. On external examination, the cadaver was 146 cm tall and weighed 51 kg. There were no apparent injuries that could be related to the death of the decedent. Internal examination of the body showed only pulmonary congestion and edema. The stomach contents (120 mL) had an irritating smell, and pH was about 9. Alcohol in the blood and the immunochemical drug screening with Triage® DOA in the urine sample (Alere, Waltham, MA, USA) were negative. Drug and pesticide screening by GC–MS also showed negative results. Thus, NAGINATA–GC–MS screening was carried out using the established extraction method and the “inorganic anion database”. Cyanide and thiocyanate were quickly identified in the blood samples. Tentative concentrations of cyanide and thiocyanate were 0.18 and 0.03 µmol/mL in the femoral vein blood, 0.51 and 0.06 µmol/mL in the right heart blood, and 0.90 and 0.05 µmol/mL in the left heart blood, respectively. Further quantification using the established method was carried out. The concentrations of cyanide and thiocyanate were 0.14 and 0.02 µmol/mL in the femoral vein blood, 0.51 and 0.07 µmol/mL in the right heart blood, and 1.12 and 0.06 µmol/mL in the left heart blood, respectively. The cyanide concentrations in this case (0.14–1.12 µmol/mL = 3.6–29.1 µg/mL) reached to fatal levels (1.1–53 µg/mL) in the reported cases [19]. Thus, the cause of death was determined to be cyanide poisoning.

Conclusions

A reliable method for determining the levels of cyanide, thiocyanate and azide in human whole blood was developed by GC–MS using an improved extractive alkylation technique. This established method was efficiently applied to the NAGINATA–GC–MS screening system. Because NAGINATA–GC–MS screening can rapidly identify these poisons without using toxic compounds as reference standards, it should be useful in forensic departments as well as laboratories of emergency hospitals.