Advertisement

Novel Selectivity-Based Forensic Toxicological Validation of a Paper Spray Mass Spectrometry Method for the Quantitative Determination of Eight Amphetamines in Whole Blood

  • Sebastiaan F. Teunissen
  • Patrick W. Fedick
  • Bjorn J. A. Berendsen
  • Michel W. F. Nielen
  • Marcos N. Eberlin
  • R. Graham Cooks
  • Arian C. van Asten
Research Article

Abstract

Paper spray tandem mass spectrometry is used to identify and quantify eight individual amphetamines in whole blood in 1.3 min. The method has been optimized and fully validated according to forensic toxicology guidelines, for the quantification of amphetamine, methamphetamine, 3,4-methylenedioxyamphetamine (MDA), 3,4-methylenedioxy-N-methylamphetamine (MDMA), 3,4-methylenedioxy-N-ethylamphetamine (MDEA), para-methoxyamphetamine (PMA), para-methoxymethamphetamine (PMMA), and 4-fluoroamphetamine (4-FA). Additionally, a new concept of intrinsic and application-based selectivity is discussed, featuring increased confidence in the power to discriminate the amphetamines from other chemically similar compounds when applying an ambient mass spectrometric method without chromatographic separation. Accuracy was within ±15% and average precision was better than 15%, and better than 20% at the LLOQ. Detection limits between 15 and 50 ng/mL were obtained using only 12 μL of whole blood.

Graphical abstract

Keywords

Forensic toxicology Amphetamines Whole blood Paper spray mass spectrometry Validation Selectivity 

Introduction

Paper spray (PS) was introduced as a new ambient ionization technique for mass spectrometry (MS) by Wang et al. in 2010 [1]. The application of a small triangular shaped paper substrate to create a stable electrospray as a way to ionize and introduce analytes of interest into the mass spectrometer provides unique features. Advantages of paper spray MS include minimal to no sample preparation, low cost and high speed of analysis, and the very low sample volumes/amounts required. This also includes accurate and precise quantitative analysis in the low ppb range (by using isotope labeled internal standards), the minimal risks of carryover (due to single use of the paper substrate), and instrument contamination (because of the retention of matrix compounds on the paper substrate) [2]. These attractive features have resulted in a broad range of applications, including the analysis of food [3], beverages [4, 5], natural products [6, 7], environmental samples [8, 9], algae [10], cell cultures [11], forensic analysis of paper evidence, including ink (i.e., questioned documents) [12, 13], drugs of abuse [14, 15], the characterization of catalysts [16, 17], the monitoring of chemical reactions [18], and the classification of bacteria [19, 20]. PS-MS has also shown educational value in academic chemical analysis and organic synthesis courses [21]. For clinical and biomedical applications, PS-MS by is ideally suited for the quantitative trace analysis of drugs, medicines, and associated metabolites in complex biological matrices such as urine, saliva, serum, whole blood, and biological tissues. Achievements, current development, and future challenges of the PS mass spectrometric analysis of biofluids have recently been discussed [2].

The analysis of drugs of abuse and their corresponding metabolites in human matrices with the simplicity and high speed features of PS-MS are of course also of high relevance to forensic science [22]. In a previous study, we have demonstrated the use of PS-MS for the quantitative analysis of eight drugs of abuse in whole blood [23]. The ability to screen multiple compounds in a single analysis is essential for successful implementation in a forensic toxicology laboratory or on-site using e.g., a miniaturized mass spectrometer [24]. As we reported earlier, developing a suitable assay for “instant” and selective drug quantification would have several important applications not only in road-side testing for driving under the influence but also in forensics and toxicology, employment and workplace screening, and therapeutic drug monitoring [23]. Amphetamine-type stimulants (ATS) are the world’s second most widely used type of drug after cannabis; hence, they are frequently encountered in forensic toxicology (https://www.unodc.org/documents/ATS/ATS_Global_Assessment_2011.pdf). According to the UNODC (United Nation’s Office on Drugs and Crime) more than 170 tons of ATS were seized globally in 2014 [25]. While methamphetamine is the predominantly encountered drug in recent years, the amount of confiscated MDMA doubled that year compared to previous annual averages. In addition to the increase in the world wide trade of ATS there is also a strong increase in the amount and number of new psychoactive substances (NPS). The UNODC has registered over 75 new compounds in the drug market in 2015 and 66 new substances in 2014. Although NPS represent a very broad range of chemical compounds, the UNODC recognizes a specific class of phenethylamines that can also be considered as amphetamine types on the basis of chemical structure and psychoactive nature [25].

An ambient MS-based quantitative method for the analysis of amphetamines in biological matrices would allow for the rapid screening of these drugs in forensic settings. We therefore developed a PS-MS method for the simultaneous quantitative analysis of eight amphetamines in human whole blood. Five of these were taken from the set of illicit drugs that were selected in another screening method [23]. para-Methoxyamphetamine (PMA) and para-methoxymethamphetamine (PMMA) were added since these are increasingly encountered as contamination in XTC tablets, resulting in injuries and several fatalities. MS analyses of these amphetamines, either by LC-MS/MS [26] or GC-MS [27], have already been briefly described in literature. The compound 4-fluoroamphetamine (4-FA) was also added because the use of this new psychoactive substance (NPS) as a recreational drug is becoming more common in The Netherlands. Since 4-FA was absent from the list of illicit drugs until recently, few analysis methods for this drug have been described [28, 29, 30].

Prior to application in forensic casework, analytical methods require thorough validation and should preferably be ISO 17025 accredited. A validation protocol, according to international guidelines, was therefore applied to test the robustness and forensic suitability of the method [31, 32, 33, 34]. Selectivity is an important aspect of the validation, and a thorough validation of this feature is needed to prevent false positive findings, especially when no separation methods are applied prior to MS analysis. We therefore discuss a new and extended approach to study and determine compound selectivity in the absence of a chromatographic separation.

Experimental

Chemicals and Reagents

All organic solvents (HPLC grade) were purchased from VWR International (Radnor, PA, USA). Amphetamine, methamphetamine, 3,4-methylenedioxyamphetamine (MDA), 3,4-methylenedioxy-N-methylamphetamine (MDMA), 3,4-methylenedioxy-N-ethylamphetamine (MDEA), para-methoxyamphetamine (PMA), para-methoxymethamphetamine (PMMA), and 4-fluoroamphetamine (4-FA), including their deuterated internal standards, were purchased, in solution, from Cerilliant (Reston, VA, USA) and stored at –20 °C. Whatman grade 31ET-Chr paper was obtained as a sample from Whatman (Piscataway, NJ, USA). Human whole blood with sodium heparin anticoagulant was purchased from Bioreclamation IVT (Hicksville, NY, USA) and stored at 4 °C.

Sample Preparation

Working drug solutions were prepared same-day for experiments by serial dilution with methanol. Internal standards (IS) were first diluted with methanol/water (1:1, v/v) and then spiked together with the working drug solutions into the liquid blood and vortexed at 3000 RPM for 15 s. Both the working drug solutions and internal standard solutions were spiked by 20-fold dilution: IS/drug/blood (10:10:180, v/v/v).

Paper Spray Ionization

Paper spray was performed using a Velox automated ion source with disposable cartridges manufactured by Prosolia, Inc. (Indianapolis, IN, USA). The cartridges consisted of a teardrop-shaped paper substrate (Whatman chromatography paper, grade 31ET-Chr), a plastic casing, and a steel ball for electrical contact.

Twelve μL of spiked blood was spotted onto the paper and dried for 2 h at room temperature. Solvent was applied to the cartridge by eight injections of 3 μL on top of the blood spot, and seven injections of 10 μL behind the blood spot. The solvent used was methanol/dichloromethane/hydroxylamine (80:20:0.2, v/v/v). Hydroxylamine originated from a 50 wt. % in H2O solution.

Mass Spectrometry, Data Collection, and Data Processing

Analyses were performed on a TSQ Quantum Access MAX (Thermo Scientific, San Jose, CA, USA). The inlet capillary temperature was 300 °C and the spray voltage was 4 kV. The amphetamines were analyzed by collision induced dissociation in the selected reaction monitoring (SRM) mode, utilizing isolation widths of 0.30 m/z and scan times of 70 ms. For six of the eight analyzed compounds, deuterated internal standards were available. The signal fluctuations of PMA and PMMA were corrected using MDA-D5 and MDMA-D5, respectively, as internal standard. The amphetamine fragmentation appeared to be strongly dependent on small variations in the collision pressure. In order to analyze all eight compounds from a single blood spot, a mass spectrometric method was therefore developed consisting of two segments, each with a different collision pressure. The mass spectrometer required 10 s to stabilize after being switched between 1.3 and 1.7 mTorr. Data was processed using the Xcalibur Quan Browser. Figure 1 and Table 1 show an overview of the mass spectrometry analysis conditions for each sample. As can be seen in Figure 1, the analysis of each analyte or internal standard starts and ends with a short period in which zero voltage is applied, thereby creating a baseline. This yields the characteristic shape of a chronogram and also facilitates integration (as illustrated in Figure 2).
Figure 1

Mass spectrometry analysis conditions for each sample

Table 1

SRM Parameters and Probabilities for Analytes and Internal Standards

Compound

Precursor ion m/z

Fragment ion m/z

Collision (arbitrary units)

Tube Lens (V)

CP (mTorr)

Ion ratio

Amphetamine

136.1

91.2

12

55

1.3

100:59

 

119.1

7

55

  

Methamphetamine

150.2

91.3

19

77

1.3

100:10

 

119.0

6

77

  

MDA

180.1

133.2

13

87

1.7

100:86

 

105.2

17

87

  

MDMA

194.1

163.1

10

95

1.7

100:22

 

105.2

19

95

  

MDEA

208.1

133.2

16

91

1.7

100:37

 

77.3

41

91

  

4-Fluoroamphetamine

154.1

109.2

14

88

1.7

100:10

 

83.3

36

88

  

PMA

166.1

121.2

14

78

1.7

100:7

 

78.3

45

78

  

PMMA

180.1

121.2

16

82

1.7

100:83

 

149.2

9

82

  

Amphetamine-D8

144.2

97.3

14

55

1.3

100:38

 

127.2

6

55

  

Methamphetamine-D8

158.2

93.3

21

77

1.3

100:71

 

124.3

8

77

  

MDA-D5

185.1

138.2

14

85

1.7

100:80

 

110.2

17

85

  

MDMA-D5

199.1

165.2

10

98

1.7

100:21

 

107.3

19

98

  

MDEA-D5

213.1

135.2

19

86

1.7

100:28

 

77.3

45

86

  

4-Fluoroamphetamine-D5

159.0

111.2

16

70

1.7

100:11

 

84.2

38

70

  
Figure 2

Example of a chronogram consisting of methamphetamine (left) and MDA (right) as obtained using the method as presented in Figure 1

Validation

Validation of Forensic Toxicological Methods

The reliability, robustness, and accuracy of analytical data in forensic toxicological casework is of great importance because of the impact that the findings can have on the outcome of a criminal investigation and subsequent court ruling. Therefore, methods are preferably ISO 17025 accredited to ensure the highest quality standard. As part of the accreditation process, validation according to accepted standards and protocols is mandatory. For forensic toxicological analysis, ISO 17025-based standard practices and guidelines are usually provided by international and national committees and working groups [31]. The validation of bioanalytical methods in forensic science, in line with existing quality frame-works, has been discussed in detail by Peters et al. [32, 33, 34] In forensic toxicological laboratories currently high performance liquid chromatography with tandem mass spectrometry (LC-MS/MS) is the preferred instrumentation for the forensic analysis of biofluids. Validation of such LC-MS/MS methods (see for example [35, 36]) is generally conducted according to the guidelines as described in these papers and hence this framework also formed the basis for the validation of the current paper spray method which lacks an LC separation prior to mass spectrometric analysis.

Regression Model

Seven non-zero calibration standards, a single blank (containing internal standard), and a double blank (not containing internal standard nor analyte) were prepared in duplicate for each run and analyzed in three independent runs. Pooled whole blood from seven batches was used for these experiments. Calibration curves were constructed using the area ratio of the transition of interest referenced to the internal standard and plotted against nominal concentrations and fitted by least-squares linear regression using the reciprocal of the concentration (1/x) as a weighting factor. According to the guidelines, as described above, to assess linearity, deviations from the mean calculated concentrations over three runs should be within 85%–115% of nominal concentrations. At the lower limit of quantification level (LLOQ) a deviation of 20% is permitted. The LLOQ was defined as the concentration at which the signal to noise ratio was equal to or greater than 5.

Accuracy and Precision

Intra- and inter-assay accuracies and precisions of the method were determined by assaying quality control samples at the LLOQ, a low, a mid, and a high concentration level in three separate runs with five replicates per concentration level. The concentration of each quality control sample was calculated using the calibration curve that was constructed from data obtained during the same run. The differences between the nominal and the measured concentrations were calculated for each replicate. This was then used to calculate accuracies. According to the guidelines, as described above, the average intra- and inter-assay accuracy should be within 85%–115% except at the LLOQ, where a range of 80%–120% is allowed. The average intra- and inter-assay precision, expressed as the coefficient of variation, should not exceed 15%, except for the LLOQ where it should not exceed 20%.

Cross-Analyte Interference

Cross-analyte interferences were tested for each of the amphetamines to determine if the presence of a high concentration of one amphetamine would produce a false-positive result for one of the other amphetamines. Whole blood samples without any amphetamines were therefore spiked, each containing only one amphetamine at its upper limit of quantification. These samples were then analyzed for the presence of each of the other amphetamines that were not spiked into the sample. The response of these amphetamines should not exceed 20% of their corresponding response in an LLOQ whole blood standard.

Selectivity

Although PS-MS usually offers simplified and faster analysis in comparison to LC-MS [2], this benefit also comes at a cost of reduced selectivity. With the direct introduction of analytes into the mass spectrometer, there is a risk of false positive results when isomers and isobars are present in the sample since they have the same molecular mass as the analyte of interest. Tandem mass spectrometry provides additional selectivity as a result of the fragmentation step but there is still a chance that isomers and isobars are present showing equal mass transitions, especially if the analyte exists of C, H, N, and O only and its functional groups are common. The use of chromatography prior to the MS analysis might then provide the additional required selectivity for discrimination but this step is avoided for the sake of speed and ease of analysis when using paper spray. Hence, in the validation of PS-MS methods, additional attention is required with respect to correct confirmation of the identity of the compound in order to prevent false positive findings. In addition to performing conventional selectivity analysis as described in international guidelines, we have therefore defined and investigated two novel features to establish selectivity in PS-MS:
  1. (1)

    Intrinsic selectivity, expressed as a random chemical match probability (RCMP), i.e., the probability that a random compound would have the same precursor ion mass and generate fragment ions with equal nominal masses, thus leading to a false positive outcome.

     
  2. (2)

    Application-based selectivity, a process used to establish if, for the analyte of interest in the matrix considered and for the PS-MS parameters applied, any other compound could a priori be present that could yield a false positive result when applying the selected PS-MS ionization and fragmentation parameters.

     

The intrinsic selectivity as determined by the RCMP provides an indication of the general discrimination power of the PS-MS analysis for that given compound and ionization and analysis parameters. This discrimination power has to be very high when analyzing complex and unknown samples to minimize the chance of an unexpected false positive outcome. In addition, the application-based selectivity ensures that the method is checked against known associated compounds that could be present in the matrix and on the basis of chemical similarity could possibly result in a false positive result. This requires the input of forensic and chemical expert knowledge and a proactive approach to challenge the method by exhaustively searching for potential false positive candidates.

Conventional Selectivity Analysis

Seven individual batches of control human whole blood were used to assess selectivity of the method. To determine whether endogenous constituents interfere with the assay, a blank and a sample spiked at the LLOQ were processed from these batches. The samples were subsequently analyzed according to the above described procedure. The area ratio observed in the blank should be below 20% of the area ratio in the LLOQ sample in each of the seven batches of control human blood.

Intrinsic Selectivity Analysis

To estimate intrinsic selectivity in PS-MS based on the RCMP, the methodology as introduced by Berendsen et al. has been adopted [37]. In this approach, the individual probabilities of the occurrence of a selected precursor ion mass (Mpc) and two product ion masses (Mpd1 and Mpd2) are estimated from comprehensive LC-MS databases containing LC-MS data of an extensive range of chemical compounds that have been detected and analyzed by the RIKILT institute. The selectivity of the precursor ion(s) is given by the fraction of compounds in the data collection showing the same precursor ion mass: P(Mpc). The selectivity of the product ions is not only related to the product ion mass (Mpd) itself but also the resulting neutral loss (Mpc – Mpd) is considered. For instance, a product ion of m/z 200 has a low occurrence, but if it originates from a precursor ion of m/z 218 and, thus, if a neutral loss of 18 Da is involved, this transition should be considered nonselective given the frequent occurrence of the loss of water with CID. In the proposed worst case approach, the selectivity of the product ion is determined by selecting the highest value of either P(Mpd) or P(Mnl), the probability of the corresponding neutral loss. The probability of the occurrence of a compound having the same MS characteristics as the compound of interest is now estimated by multiplying the probabilities of the precursor ion and the two worst case values of the two product ion masses, i.e.:
$$ \mathrm{RCMP}=\mathrm{P}{\left({\mathrm{M}}_{\mathrm{pc}}\right)}^{\ast }\ \max\ {\left[\mathrm{P}\left({\mathrm{M}}_{\mathrm{pd}1}\right),\mathrm{P}\left({\mathrm{M}}_{\mathrm{nl}1}\right)\right]}^{\ast }\ \max\ \left[\mathrm{P}\left({\mathrm{M}}_{\mathrm{pd}2}\right),\mathrm{P}\left({\mathrm{M}}_{\mathrm{nl}2}\right)\right]. $$

Of course additional selectivity can be obtained by considering additional product ions but using two product ions, preferably both of high signal intensity, is common practice. It should be noted that in this calculation it is assumed that the probabilities are independent and thus that the product ions and the precursor ion are completely independent and uncorrelated. From a chemical perspective it is clear that this assumption will not always be valid. As an example, if a product ion of m/z 105 is observed, which is usually the benzoyl cation, in many cases a product ion of m/z 91, the tropylium cation, will also be observed. If correlated product ions are selected for confirmation of the identity of a particular compound, the selectivity can be compromised. In other words, the use of multiple fragment ions is of limited added value and will lead to an underestimation of the RCMP. Furthermore, a relationship can also exists between the precursor ion and the product ion, as reported previously [37]. This also introduces an additional uncertainty in the RCMP calculations. However, this reduction in RCMP will be partly counteracted by the worst-case approach in the estimation of the probabilities of the product ions. Overall, a RCMP, calculated as described above, is not a quantitative measure but should be considered as an indicator of the discriminative power of the PS-MS method for a given compound. Nevertheless, calculation of the RCMP is a valuable, easy, and fast tool to investigate the intrinsic selectivity for the confirmation of the identity of many compounds under selected instrumental conditions. Additionally, the RCMP can be a very useful parameter for selecting appropriate product ions when developing and optimizing an MRM method as is further discussed in the next section.

Application Based Selectivity Analysis

A robust PS-MS method requires a high intrinsic selectivity and the RCMP protocol provides an elegant means to demonstrate that this requirement is met. However, a low RCMP value alone does not warrant sufficient selectivity for a given compound in a given matrix. RCMP values are estimated from databases constructed from the LC-MS analysis of a broad range of compounds that are mostly unrelated to the compounds of interest. Therefore, the intrinsic selectivity only provides generic insight into selectivity and does not consider the existence of specific isomers and isobars that are likely to yield a false positive outcome and that could a priori be present in the considered matrix, e.g., human whole blood. Therefore, after establishing sufficient intrinsic validity, an additional step is required in the selectivity validation process to ensure that the PS-MS method is truly fit for purpose and with a negligible risk for false positive results. This step, defined as the application-based selectivity protocol, is based on the following iterative process:
  1. (1)

    A search for all compounds that have the same molecular formula as the compound of interest,

     
  2. (2)

    selection of all compounds that comply with step (1) and have functional groups that could result in the same selected product ions as the compound of interest,

     
  3. (3)

    assessment of the probability of the occurrence of the candidate compounds resulting from step (2), in the matrix in which the analysis is performed.

     

This approach results in a short list of compounds that could result in false positive outcomes in actual casework as these compounds have an identical precursor ion mass, could yield the same fragments during CID, and could be present in the samples analyzed. If such compounds exist and further investigation confirms the false positive potential, the PS-MS method should be adapted, e.g., by selecting other product ions, extended, e.g., by employing reactive paper spray or using additional information such as the ion ratio or combined with an additional analytical method for confirmatory analysis.

Even though this approach yields a good approximation of the probability of a false positive result, it should be noted that the application-based selectivity assessment can be an extensive and time-consuming process that should be carried out for every individual compound under the selected instrumental conditions. In addition such an assessment is not a ‘one-time effort’ and should be repeated over time as new compounds constantly emerge. In case of illicit drug analysis in human whole blood, the appearance of a new NPS could for instance suddenly yield a potential false positive result in a validated method. In this study the application-based selectivity has been assessed for the target amphetamines.

Results and Discussion

Method Development

The goal of this research was to develop and validate a quantitative method for the analysis of eight amphetamines in dried blood spots at therapeutically relevant concentrations using a separation-free, simplified, and fast direct MS method. Appropriate sensitivity was considered to be the ability to measure below 50 ng/mL, which is the impaired driving limit for all amphetamines listed in the Dutch Road Traffic act. Amphetamine ions show multiple ion fragments upon CID. Of the observed fragments, two were chosen: one for quantification and one for confirmation of the identity. RCMPs were estimated for precursor masses and all observed product masses. An LLOQ was subsequently determined for each mass transition. If multiple transitions were available at the appropriate sensitivity, the fragment ions with the lowest RCMP were selected. MDA and PMMA were exceptions to this strategy since they have identical molecular masses and share various fragment ions with identical masses. It was pivotal in this case to search for selective fragment ions of these amphetamines in the absence of chromatography, and selection of the identical fragment ions would result in cross-analyte interference and therefore possibly false-positive results. Both drugs indeed had a distinctive fragment, which could be selected without comprising the LLOQ. Table 1 displays an overview of the MS parameters of the final method.
Table 2

Assay Performance Data

Analyte (linear range)

Nominal concentration (ng/mL)

Inter-assay accuracy (% dev)

Inter-assay precision (% CV)

Amphetamine

(50–4000 ng/mL)

50

12.2

19.9

150

4.72

11.9

750

5.38

2.93

3000

6.83

3.84

Methamphetamine

(15–1200 ng/mL)

15

-5.30

4.94

45

-6.16

4.56

225

-2.93

2.98

900

-0.102

3.45

MDA

(30–2400 ng/mL)

30

-4.83

14.9

90

-8.78

9.72

450

2.07

13.1

1800

1.30

4.40

MDMA

(15–1200 ng/mL)

15

-3.09

4.88

45

-8.44

3.91

225

-4.53

12.4

900

-1.03

4.20

MDEA

(30–2400 ng/mL)

30

1.31

5.32

90

-6.15

8.43

450

-3.08

5.32

1800

-1.45

4.55

4-Fluoroamphetamine

(15–1200 ng/mL)

15

-0.258

18.4

45

-1.17

10.6

225

-2.18

13.3

900

0.193

7.97

PMA

(50–4000 ng/mL)

50

5.58

8.30

150

-2.49

6.22

750

6.47

5.16

3000

6.03

5.89

PMMA

(15–1200 ng/mL)

15

-3.44

4.57

45

-5.20

7.31

225

5.53

10.3

900

9.03

4.79

Inter-assay data (n = 15) is based on three runs with n = 5 per run. Nominal concentrations are from quality control samples at the LLOQ, a low, a mid, and a high concentration level

We selected a solvent system based on dichloromethane and hydroxylamine. As stated before, hydroxylamine is a very mild source of protons without requiring adjustment of the pH [23]. Since the ionization of amphetamines appeared to be unaffected using hydroxylamine instead of the more common additive acetic acid, we continued using the hydroxylamine-based solvent system. This would allow the possible addition of drugs from the previously published method to the current one since hydroxylamine is necessary for the analysis of morphine [23]. Dichloromethane is also beneficial since it results in better wetting of the blood spot than a water/methanol mixture.

Validation

All the calibration curves, constructed using a weighting factor of 1/x, showed coefficients of variation (CV) less than 15% and accuracies were in all cases within 85%–115%. Assay performance data (inter-assay accuracy and precision) of all analyzed compounds is summarized in Table 2. Details of the calibration can be found in section S1 of the Supplementary Information. Both Inter-assay and intra-assay accuracies and precisions were within the defined limits for all drugs. Intra-assay data are not shown because of space limitation. No substantial cross-analyte interference was observed indicating that the eight amphetamines can be accurately co-analyzed in a single PS-MS method.

Selectivity

Conventional Selectivity Analysis

PS-MS MRM chronograms of five out of seven batches of control human blood contained no interference larger than 20% of the area at the LLOQ level of all amphetamines. In two batches, an interference was observed up to 48% of the signal as observed in the LLOQ samples of amphetamine, 4-FA, and MDA, respectively. Visual inspection showed these batches of whole blood to be extremely viscous and full of clots. It seems likely that these aspects have caused interference in the analysis but further research is warranted to analyze how this influences the selectivity in additional batches of blank whole blood. In contrast, human whole blood samples as obtained from suspects in a DUI investigation by law enforcement represent a much more consistent matrix that corresponds to the blood samples as used for accuracy and precision analyses in this study.

Intrinsic Selectivity Analysis

When analyzing amphetamines, the relevant precursor ion mass range is m/z 130–220. Based on the previously applied databases for RCMP calculation, a probability distribution of the occurrence of product ions and neutral losses was made specifically for the precursor ion mass range relevant for the analysis of amphetamines. These distributions, as well as for the precursor ion mass in the range of m/z 100–1000, are presented in Figure 3. Clearly, the tropylium cation of m/z 91 is the most common product ion followed by the ions of m/z 77, 69, 95, and 105. Neutral losses of 17 (ammonium), 18 (water), and 46 (formic acid) Da are also frequently observed.
Figure 3

Probability distribution of (a) precursor ion masses, (b) product ion masses, and (c) neutral losses. Probability distribution of product ion masses and neutral losses originate from precursor ions of m/z 130–220 (n = 3547)

As stated above, we tried to choose the fragment ion with the lowest RCMP as is illustrated in Table 3. However, for amphetamine and methamphetamine, only three fragment ions are observed: m/z 119, 91, and 65, of which the intensity of the latter is so low that it would result in a 3-fold increase in the LLOQ. Additionally, selection of the fragment ion for MDA and PMMA based on RCMP was limited because of a shared precursor mass and therefore the same fragment ion masses, as stated before. MDMA and MDEA produce the same set of six most abundant fragment ions. They provide a clear example, however, that selection of fragment ions is not solely based on the probability of the fragment ion mass but also on the probability of the neutral loss. Since MDMA and MDEA have different parent ion masses, their equal fragment ion masses are the result of different neutral losses. As a result, the probability of each of the most predominant six transitions of MDMA is not equal to those of MDEA. Optimization of the selectivity (i.e., obtaining the lowest overall RCMP whilst maintaining sufficient sensitivity) thus leads to the selection of different fragment ions for these two drugs. PMA and 4-FA both provide a rather large set of fragment ions. Their signal intensities, however, vary drastically. In both cases the most abundant fragment ions were chosen since the third most abundant fragment ion would have resulted in an increase of the LLOQ by a factor of three.
Table 3

Calculated RCMP for Each Observed Mass Transition Per Analyte and Internal Standard

Analyte

Precursor ion (m/z)

Product Ion (m/z)

Neutral loss

Relative intensity

P(precursor)

P(product)

P(total) ideal

P(total) selected

Amphetamine

136.1

119.2

17

12

8.2E-05

2.7E-01

1.1E-06

3.8E-06

136.1

91.2

45

100

 

1.7E-01

  

136.1

77.3

59

2

 

1.7E-01

  

136.1

65.3

71

29

 

7.7E-02

  

Methamphetamine

150.1

119.2

31

4.95

1.3E-04

9.6E-02

8.5E-07

2.1E-06

150.1

91.2

59

100

 

1.7E-01

  

150.1

65.3

85

28

 

6.6E-02

  

MDA

180.0

163.2

17

54

3.5E-04

2.7E-01

2.6E-06

3.1E-06

180.0

135.1

45

100

 

1.3E-01

  

180.0

133.1

47

72

 

8.4E-02

  

180.0

105.3

75

80

 

1.1E-01

  

180.0

79.3

101

43

 

8.9E-02

  

180.0

77.3

103

61

 

1.2E-01

  

MDMA

194.0

163.2

31

96

4.5E-04

3.8E-02

1.5E-06

1.8E-06

194.0

135.1

59

76

 

1.7E-01

  

194.0

133.2

61

43

 

1.0E-01

  

194.0

105.2

89

79

 

1.1E-01

  

194.0

79.3

115

33

 

8.9E-02

  

194.0

77.3

117

100

 

1.2E-01

  

MDEA

208.0

163.1

45

72

6.3E-04

6.4E-02

1.2E-06

1.6E-06

208.0

135.1

73

87

 

6.9E-02

  

208.0

133.1

75

50

 

4.4E-02

  

208.0

105.2

103

93

 

5.4E-02

  

208.0

79.3

129

42

 

4.5E-02

  

208.0

77.3

131

100

 

5.9E-02

  

4-FA

154.0

137.2

17

11

2.0E-04

2.7E-01

4.5E-07

1.9E-06

154.0

109.2

45

100

 

1.3E-01

  

154.0

83.2

71

27

 

7.7E-02

  

154.0

63.3

91

4

 

1.0E-01

  

154.0

59.4

95

2

 

4.0E-02

  

154.0

57.3

97

8

 

5.6E-02

  

PMA

166.0

149.2

17

41

2.6E-04

2.7E-01

6.9E-07

1.3E-06

166.0

121.2

45

100

 

1.3E-01

  

166.0

91.2

75

30

 

1.7E-01

  

166.0

78.3

88

7

 

4.0E-02

  

166.0

77.3

89

14

 

1.2E-01

  

166.0

65.3

101

10

 

6.6E-02

  

PMMA

180.0

149.2

31

15

3.5E-04

3.3E-02

7.7E-07

1.9E-06

180.0

121.2

59

100

 

1.7E-01

  

180.0

91.2

89

39

 

1.7E-01

  

180.0

77.3

103

14

 

1.2E-01

  

180.0

65.4

115

13

 

6.6E-02

  

Amphetamine-D8

144.2

127.3

17

16

1.4E-04

2.7E-01

3.3E-07

2.4E-06

144.2

97.2

47

100

 

6.5E-02

  

144.2

96.2

48

52

 

6.1E-02

  

144.2

69.3

75

28

 

1.1E-01

  

144.2

68.3

76

12

 

3.7E-02

  

Methamphetamine-D8

158.2

124.3

34

9

2.0E-04

5.2E-02

4.0E-07

1.0E-06

158.2

93.2

65

100

 

9.6E-02

  

158.2

67.4

91

12

 

1.0E-01

  

158.2

66.3

92

20

 

3.8E-02

  

MDA-D5

185.2

168.2

17

69

3.8E-04

2.7E-01

6.5E-07

1.2E-06

185.2

138.2

47

66

 

3.5E-02

  

185.2

137.1

48

61

 

5.8E-02

  

185.2

110.2

75

100

 

8.7E-02

  

185.2

80.2

105

25

 

4.9E-02

  

MDMA-D5

199.2

165.2

34

100

4.6E-04

5.2E-02

6.3E-07

1.8E-06

199.2

136.2

63

41

 

1.7E-01

  

199.2

135.2

64

48

 

8.4E-02

  

199.2

107.2

92

52

 

7.7E-02

  

199.2

81.3

118

11

 

6.6E-02

  

199.2

78.3

121

33

 

2.6E-02

  

MDEA-D5

213.1

163.2

50

100

5.8E-04

1.9E-02

4.4E-07

1.3E-06

213.1

135.1

78

52

 

4.0E-02

  

213.1

133.2

80

41

 

4.2E-02

  

213.1

105.2

108

57

 

5.4E-02

  

213.1

79.3

134

22

 

4.5E-02

  

213.1

77.3

136

53

 

5.9E-02

  

4-FA-D5

159.0

142.2

17

18

2.3E-04

2.7E-01

5.5E-07

1.0E-06

159.0

112.2

47

19

 

4.7E-02

  

159.0

111.2

48

100

 

5.1E-02

  

159.0

110.2

49

40

 

5.8E-02

  

159.0

84.2

75

26

 

8.7E-02

  

Finally, under the chosen conditions, the intrinsic probability of an interfering compound for the selected amphetamines ranges from 1.0*10–6 to 3.8*10–6. Note that the selectivity has to be set fit for purpose, but as previously discussed [37], in residue analysis a probability below 2*10–7 is preferred. In this case, none of the amphetamines comply with this guideline and therefore it is suggested, based on the intrinsic selectivity calculations, that the selectivity should be further improved to achieve additional certainty on the confirmation of the identity of the amphetamines in whole blood, e.g., by monitoring an additional transition. The relatively low selectivity is caused by the omission of chromatography in combination with the very generic molecular composition of the amphetamines having common functional groups.

Application-Based Selectivity Analysis

For each of the eight selected amphetamines we performed an extensive search for false positive candidates on the basis of chemical structure and CID fragment ions using PubChem. The procedure and results are described in detail in Section S2 of the Supplementary Information. Compounds were screened and selected in four consecutive steps:
  1. 1.

    Selection of Pubchem compounds with an identical molecular formula as the amphetamine considered.

     
  2. 2.

    Sub-selection of Pubchem compounds that contain a phenyl or benzodioxol moiety similar to the amphetamine considered.

     
  3. 3.

    Manual selection of candidates on the basis of structural similarities that could lead to similar CID fragments as the amphetamine considered.

     
  4. 4.

    Using Pubchem and Internet information and forensic expert input to investigate whether the false positives candidates are likely to occur in human whole blood, e.g., as drugs of abuse, medicines, doping agents, metabolites, food stuff, natural compounds, or abundant man-made chemicals.

     

This methodology is based on a couple of assumptions to effectively scan millions of compounds to find potential false positive candidates. We assume that no compounds exist with a different molecular formula that could result in a false positive outcome. This assumption will be true for most small molecules but will become more problematic when dealing with molecules with higher molecular masses. High resolution MS with accurate mass measurements would add a lot of certainty in this respect but is not always an option in terms of cost and required speed of analysis in PS-MS analysis. We also assume that no compounds exist with the same molecular formula that do not possess an aromatic functionality but still result in a false positive outcome. Given the selected fragment ions that are all based on the phenyl moiety, this seems to be a very reasonable assumption. Finally, if in PubChem only references are made to chemical specialty companies that provide the material as a standard or reference and the material is unknown in forensic toxicological and illicit drug case work, we assume that finding such a compound in human whole blood is a priori very unlikely.

This search strategy resulted in four potential false positive candidates out of a total of over 91 million registered compounds in the Pubchem platform (Figure 4). This exceptional selectivity clearly shows the power of MRM in increasing the reliability of separation-free ambient mass spectrometry.
Figure 4

(a) para-Methoxymethamphetamine (b) ortho-methoxymethamphetamine (methoxyphenamine), (c) methamphetamine, (d) phenpromethamine, (e) 2-fluoroamphetamine, (f) 3-fluoroamphetamine, (g) 4-fluoroamphetamine

Compounds that have been identified as potential false positive sets with a reasonable a-priori probability of being present in human whole blood samples are phenpromethamine versus methamphetamine, methoxyphenamine, also known as ortho-methoxymethamphetamine versus PMMA, and 2-fluoroamphetamine (2-FA) and 3-fluoroamphetamine (3-FA) versus 4-FA. These compounds might theoretically interfere although to our knowledge they have not been reported in forensic case work. Actual analysis of these compounds, using the developed PS-MS method, should show whether we indeed can predict a response similar to the analyte of interest on the basis of molecular structure. The outcomes of the application based selectivity yields a more accurate estimation of the probability of encountering an interfering substance. The finding of potentially interfering substances for the studied amphetamines is in agreement with not meeting the residue analysis guideline as determined by the intrinsic selectivity calculations.

A feature that has not been considered thus far is the ion ratio, which might play a crucial role in the extent of interference. Johansen et al. demonstrated how 4-FA can be discriminated from 2-FA and 3-FA by variations in a selective ion ratio [30]. Therefore, this feature could also be applicable in our PS-MS method. To the best of our knowledge, no studies utilizing MS for the analysis of phenpromethamine or methoxyphenamine have been published. Further research is therefore warranted to determine the ion ratio of these compounds to define the extent of interference with methamphetamine and PMMA, respectively.

Conclusion

We have successfully developed and validated a rapid, sensitive, and sufficiently selective separation-free method for the quantification of eight amphetamines in whole blood based on PS-MS. Lower limits of quantification were below typical physiological and toxicological levels of all amphetamines. All considered compounds were quantified simultaneously in a single PS-MS analysis of only 1.3 min. We have also combined three independent approaches to ensure proper selectivity in the absence of chromatographic separation. In addition to the conventional approach, a previously published method for the theoretical calculation of the intrinsic selectivity and a new iterative process to determine the application-based selectivity were applied. Since new drugs, e.g., NPS, are continuously emerging, the list of possible interfering compounds should be kept up-to-date and selectivity validation should also be repeated regularly over time. To allow for on-site analysis, e.g., on crime scenes or at drug laboratories, we are currently transferring the herein presented method to a miniaturized mass spectrometer that also utilizes paper spray ionization.

Notes

Acknowledgements

Karen Yannell is gratefully acknowledged for her scientific input in the method development. The authors thank the Brazilian Science Foundation FAPESP (process No. 2016/01683-2) and the National Science Foundation (NSF CHE-1307264) for financial support.

Supplementary material

13361_2017_1790_MOESM1_ESM.docx (382 kb)
ESM 1 (DOCX 381 kb)

References

  1. 1.
    Wang, H., Liu, J., Cooks, R.G., Ouyang, Z.: Paper spray for direct analysis of complex mixtures using mass spectrometry. Angew. Chem. Int. Ed. 49, 877–880 (2010)CrossRefGoogle Scholar
  2. 2.
    Manicke, N.E., Bills, B.J., Zhang, C.: Analysis of biofluids by paper spray MS: advances and challenges. Bioanalysis. 8, 589–606 (2016)CrossRefGoogle Scholar
  3. 3.
    Zhang, Z., Cooks, R.G., Ouyang, Z.: Paper spray: a simple and efficient means of analysis of different contaminants in foodstuffs. Analyst. 137, 2556–2558 (2012)CrossRefGoogle Scholar
  4. 4.
    Garrett, R., Rezende, C., Ifa, D.: Coffee origin discrimination by paper spray mass spectrometry and direct coffee spray analysis. Anal. Methods. 5, 5944–5948 (2013)CrossRefGoogle Scholar
  5. 5.
    Wang, Q., Zheng, Y., Zhang, X., Han, X., Wang, T., Zhang, Z.: A silica coated paper substrate: development and its application in paper spray mass spectrometry for rapid analysis of pesticides in milk. Analyst. 140, 8048–8056 (2015)CrossRefGoogle Scholar
  6. 6.
    Deng, J., Yang, Y.: Chemical fingerprint analysis for quality assessment and control of Bansha herbal tea using paper spray mass spectrometry. Anal. Chim. Acta. 785, 82–90 (2013)CrossRefGoogle Scholar
  7. 7.
    Taverna, D., Di Donna, L., Mazzotti, F., Tagarelli, A., Napoli, A., Furia, E., Sindona, G.: Rapid discrimination of bergamot essential oil by paper spray mass spectrometry and chemometric analysis. J. Mass Spectrom. 51, 761–767 (2016)CrossRefGoogle Scholar
  8. 8.
    Jjunju, F., Maher, S., Damon, D., Barrett, R., Syed, S., Heeren, R., Taylor, S., Badu-Tawiah, A.: Screening and quantification of aliphatic primary alkyl corrosion inhibitor amines in water samples by paper spray mass spectrometry. Anal. Chem. 88, 1391–1400 (2016)CrossRefGoogle Scholar
  9. 9.
    Maher, S., Jjunju, F., Damon, D., Gorton, H., Maher, Y., Syed, S., Heeren, R., Young, I., Taylor, S., Badu-Tawiah, A.: Direct analysis and quantification of metaldehyde in water using reactive paper spray mass spectrometry.  Sci. Rep. 6, 35643 (2016).  https://doi.org/10.1038/srep35643
  10. 10.
    Oradu, S., Cooks, R.: Multistep mass spectrometry methodology for direct characterization of polar lipids in green microalgae using paper spray ionization. Anal. Chem. 84, 10576–10585 (2012)CrossRefGoogle Scholar
  11. 11.
    Liu, W., Wang, N., Lin, X., Ma, Y., Lin, J.: Interfacing microsampling droplets and mass spectrometry by paper spray ionization for online chemical monitoring of cell culture. Anal. Chem. 86, 7128–7134 (2014)CrossRefGoogle Scholar
  12. 12.
    Ferreira, P., Silva, D., Augusti, R., Piccin, E.: Forensic analysis of ballpoint pen inks using paper spray mass spectrometry. Analyst. 140, 811–819 (2015)CrossRefGoogle Scholar
  13. 13.
    Jurisch, M., Augusti, R.: Detection of signature forgery with erasable pens using paper spray mass spectrometry (PS-MS). Anal. Methods. 8, 4543–4546 (2016)CrossRefGoogle Scholar
  14. 14.
    Li, M., Zhang, J., Jiang, J., Zhang, J., Gao, J., Qiao, X.: Rapid, in situ detection of cocaine residues based on paper spray ionization coupled with ion mobility spectrometry. Analyst. 139, 1687–1691 (2014)CrossRefGoogle Scholar
  15. 15.
    de Paula, C., Lordeiro, R., Piccin, E., Augusti, R.: Paper spray mass spectrometry applied to the detection of cocaine in simulated samples. Anal. Methods. 7, 9145–9149 (2015)CrossRefGoogle Scholar
  16. 16.
    Zheng, Y., Zhang, X., Bai, Z., Zhang, Z.: Characterization of the surface properties of MgO using paper spray mass spectrometry. Rapid Commun. Mass Spectrom. 30, 217–225 (2016)CrossRefGoogle Scholar
  17. 17.
    Banerjee, S., Basheer, C., Zare, R.: A study of heterogeneous catalysis by nanoparticle-embedded paper-spray ionization mass spectrometry. Angew. Chem. Int. Ed. 55, 12807–12811 (2016)CrossRefGoogle Scholar
  18. 18.
    de Paula, C., Valadares, A., Jurisch, M., Piccin, E., Augusti, R.: Paper spray mass spectrometry applied in the monitoring of a chemical system in dynamic chemical equilibrium: the redox process of methylene blue. Rapid Commun. Mass Spectrom. 30, 1176–1180 (2016)CrossRefGoogle Scholar
  19. 19.
    Hamid, A., Jarmusch, A., Pirro, V., Pincus, D., Clay, B., Gervasi, G., Cooks, R.: Rapid discrimination of bacteria by paper spray mass spectrometry. Anal. Chem. 86, 7500–7507 (2014)CrossRefGoogle Scholar
  20. 20.
    Hamid, A., Wei, P., Jarmusch, A., Pirro, V., Cooks, R.: Discrimination of Candida species by paper spray mass spectrometry. Int. J. Mass Spectrom. 378, 288–293 (2015)CrossRefGoogle Scholar
  21. 21.
    Bain, R., Pulliam, C., Raab, S., Cooks, G.: Chemical synthesis accelerated by paper spray: the Haloform reaction. J. Chem. Educ. 93, 340–344 (2016)CrossRefGoogle Scholar
  22. 22.
    Correa, D.N., Santos, J.M., Eberlin, L.S., Eberlin, M.N., Teunissen, S.F.: Forensic chemistry and ambient mass spectrometry: a perfect couple destined for a happy marriage? Anal. Chem. 88, 2515–2526 (2016)CrossRefGoogle Scholar
  23. 23.
    Espy, R.D., Teunissen, S.F., Manicke, N.E., Ren, Y., Ouyang, Z., van Asten, A., Cooks, R.G.: Paper spray and extraction spray mass spectrometry for the direct and simultaneous quantification of eight drugs of abuse in whole blood. Anal. Chem. 86, 7712–7718 (2014)CrossRefGoogle Scholar
  24. 24.
    Li, L., Chen, T.C., Ren, Y., Hendricks, P.I., Cooks, R.G., Ouyang, Z.: Mini 12, miniature mass spectrometer for clinical and other applications–introduction and characterization. Anal. Chem. 86, 2909–2916 (2014)CrossRefGoogle Scholar
  25. 25.
    United Nations Office on Drugs and Crime, World Drug Report. 2016. Vienna, United Nations publication. http://www.unodc.org/ (2016) Accessed May 2016
  26. 26.
    Ambach, L., Hernández Redondo, A., König, S., Weinmann, W.: Rapid and simple LC-MS/MS screening of 64 novel psychoactive substances using dried blood spots. Drug Test Anal. 6, 367–375 (2014)CrossRefGoogle Scholar
  27. 27.
    Rohanova, M., Balikova, M.: Studies on distribution and metabolism of para-methoxymethamphetamine (PMMA) in rats after subcutaneous administration. Toxicology. 259, 61–68 (2009)CrossRefGoogle Scholar
  28. 28.
    Röhrich, J., Becker, J., Kaufmann, T., Zörntlein, S., Urban, R.: Detection of the synthetic drug 4-fluoroamphetamine (4-FA) in serum and urine. Forensic Sci. Int. 215, 3–7 (2012)CrossRefGoogle Scholar
  29. 29.
    Al-Saffar, Y., Stephanson, N.N., Beck, O.: Multicomponent LC-MS/MS screening method for detection of new psychoactive drugs, legal highs, in urine-experience from the Swedish population. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 930, 112–120 (2013)CrossRefGoogle Scholar
  30. 30.
    Johansen, S.S., Hansen, T.M.: Isomers of fluoroamphetamines detected in forensic cases in Denmark. Int. J. Legal Med. 126, 541–547 (2012)CrossRefGoogle Scholar
  31. 31.
    [Anonymous]: Scientific Working Group for Forensic Toxicology (SWGTOX) Standard Practices for Method Validation in Forensic Toxicology. J. Anal. Toxicol. 37, 452–474 (2013)CrossRefGoogle Scholar
  32. 32.
    Peters, F., Maurer, H.: Bioanalytical method validation and its implications for forensic and clinical toxicology – a review. Accreditation and Quality Assurance. 7, 441–449 (2002)CrossRefGoogle Scholar
  33. 33.
    Peters, F., Drummer, O., Musshoff, F.: Validation of new methods. Forensic Sci. Int. 165, 216–224 (2007)CrossRefGoogle Scholar
  34. 34.
    Wille, S., Peters, F., Di Fazio, V., Samyn, N.: Practical aspects concerning validation and quality control for forensic and clinical bioanalytical quantitative methods. Accreditation and Quality Assurance. 16, 279–292 (2011)CrossRefGoogle Scholar
  35. 35.
    Bjork, M., Nielsen, M., Markussen, L., Klinke, H., Linnet, K.: Determination of 19 drugs of abuse and metabolites in whole blood by high-performance liquid chromatography-tandem mass spectrometry. Anal. Bioanal. Chem. 396, 2393–2401 (2010)CrossRefGoogle Scholar
  36. 36.
    Bjork, M., Simonsen, K., Andersen, D., Dalsgaard, P., Siguroardottir, S., Linnet, K., Rasmussen, B.: Quantification of 31 illicit and medicinal drugs and metabolites in whole blood by fully automated solid-phase extraction and ultra-performance liquid chromatography-tandem mass spectrometry. Anal. Bioanal. Chem. 405, 2607–2617 (2013)CrossRefGoogle Scholar
  37. 37.
    Berendsen, B., Stolker, L., Nielen, M.: The (Un)certainty of selectivity in liquid chromatography tandem mass spectrometry. J. Am. Soc. Mass Spectrom. 24, 154–163 (2013)CrossRefGoogle Scholar

Copyright information

© American Society for Mass Spectrometry 2017

Authors and Affiliations

  • Sebastiaan F. Teunissen
    • 1
    • 2
  • Patrick W. Fedick
    • 2
  • Bjorn J. A. Berendsen
    • 3
  • Michel W. F. Nielen
    • 3
    • 4
  • Marcos N. Eberlin
    • 1
  • R. Graham Cooks
    • 2
  • Arian C. van Asten
    • 5
    • 6
    • 7
  1. 1.ThoMSon Mass Spectrometry LaboratoryUniversity of Campinas - UNICAMPCampinasBrazil
  2. 2.Department of Chemistry and Center for Analytical InstrumentationPurdue UniversityWest LafayetteUSA
  3. 3.RIKILT, Wageningen ResearchWageningenThe Netherlands
  4. 4.Laboratory of Organic ChemistryWageningen UniversityWageningenThe Netherlands
  5. 5.Netherlands Forensic InstituteThe HagueThe Netherlands
  6. 6.Faculty of Science, Van ’t Hoff Institute for Molecular SciencesUniversity of AmsterdamAmsterdamThe Netherlands
  7. 7.CLHC, Amsterdam Center for Forensic Science and MedicineAmsterdamThe Netherlands

Personalised recommendations