Validation of a method for the detection and ...

Analytica Chimica Acta 586 (2007) 383–393

Validation of a method for the detection and confirmation of nitroimidazoles and the corresponding hydroxy metabolites in pig plasma by high performance liquid chromatography–tandem mass spectrometry St´ephanie Fraselle ∗ , Veerle Derop, Jean-Marie Degroodt, Joris Van Loco Scientific Institute of Public Health, Food Department, 14 J. Wytsmanstreet, 1050 Brussels, Belgium Received 19 June 2006; received in revised form 18 December 2006; accepted 8 January 2007 Available online 13 January 2007

Abstract Nitroimidazoles (Ronidazole, Dimetridazole, Metronidazole, Ipronidazole) and their hydroxy metabolites are banned substances with antibiotic and anticoccidial activity. They are suspected to be carcinogenic and mutagenic. Since nitroimidazoles showed an inhomogeneous distribution and a rapid degradation in incurred muscle samples, plasma is the preferred target matrix for residue analysis. The analytical method of Polzer et al. [J. Polzer, C. Stachel, P. Gowik, Anal. Chim. Acta 521 (2004) 189] was adapted for liquid chromatography–tandem mass spectrometry detection and was validated in house according to the Commission Decision 2002/657/EC. The method is specific for all nitroimidazole except for Ipronidazole and its metabolite, due to interferences at their retention times in chromatograms of blank plasma and reagents samples. The absence of a matrix effect enables the use of a (linear) calibration curve in solution for quantitation. The apparent recovery (obtained after correction with a deuterated internal standard) is between 93% and 123%, except for the metabolite of Metronidazole (58–63%). The repeatability (CVr = 2.49–13.39%) and intralaboratory reproducibility (CVRW = 2.49–16.38%) satisfy the Horwitz equation. The obtained values for the detection capacity (CC␤) range from 0.25 to 1 ␮g L−1, while values obtained for the decision limit (CC␣) are below CC␤. © 2007 Elsevier B.V. All rights reserved. Keywords: Nitroimidazoles; Plasma; Liquid chromatography–tandem mass spectrometry; Validation; Commission Decision 2002/657/EC

1. Introduction Ronidazole (1-methyl-2-[(carbamoyloxy)methyl]-5-nitroimidazole, RNZ), Dimetridazole (1,2-dimethyl-5-nitroimidazole, DMZ), Metronidazole (1-(2-hydroxyethyl)-2-methyl5-nitroimidazole, MNZ), Ipronidazole (2-isopropyl-1-methyl5-nitroimidazole, IPZ) and their corresponding hydroxy metabolites 2-hydroxymethyl-1-methyl-5-nitroimidazole (HMMNI, the metabolite of both RNZ and DMZ), 1-(2-hydroxyethyl)2-hydroxymethyl-5-nitroimidazole (MNZ-OH) and 1-methyl2-(2 -hydroxyisopropyl)-5-nitroimidazole (IPZ-OH) are banned substances with antibiotic and anticoccidial activity. They were used to prevent and treat histomoniasis and coccidiosis in poultry. They have been also used for the treatment of genital trichomoniasis in cattle and haemorrhagic enteritis in pigs. They are suspected to be carcinogenic and mutagenic. Therefore,

∗

Corresponding author. Tel.: +32 2 642 52 09; fax: +32 2 642 56 91. E-mail address: [email protected] (S. Fraselle).

0003-2670/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2007.01.009

RNZ, DMZ and MNZ have been included in Annex IV of Council Regulation 2377/90/EC [1], whereas IPZ is a non-authorized substance for veterinary purposes. Many methods for the determination and confirmation of nitroimidazoles in different matrices, mainly in meat and eggs, have been published [2–12], either using LC–MS/(MS) or GC–MS/(MS). Recent studies on the stability and homogeneity of nitroimidazoles in incurred muscle samples have been carried out by the BVL (Bundesamt f¨ur Verbraucherschutz und Lebensmittelsicherheit, Berlin, Germany) which is the Community Reference Laboratory (CRL) responsible for nitroimidazoles [13,14]. It was shown that the analyte distribution in turkey muscle was not homogeneous; moreover a rapid decline in analytes at storage above 4 ◦ C was observed. For plasma and retina samples however, the analytes were stables during storage under the same conditions, found in considerably higher concentrations and could be detected for a longer period of time after withdrawal of the medication. Therefore, plasma and retina have been recommended as target matrices for the residue control of nitroimidazoles, especially in poultry. All the studies performed by the CRL, employing

384

S. Fraselle et al. / Analytica Chimica Acta 586 (2007) 383–393

gas chromatography–mass spectrometry (GC–MS), required derivatisation prior to analyses. In the course of the applied silylation, RNZ and HMMNI formed an identical derivative and it was therefore not possible to distinguish between these two compounds. As far as we know, no methods coupling extraction of plasma samples and LC–MS/MS detection have been published to confirm nitroimidazoles. Such a method, using the extraction protocol of the CRL [13] and a confirmatory analysis by liquid chromatography coupled to atmospheric pressure chemical ionisation mass spectrometry (LC–APCI-MS/MS), has been validated according to the Commission Decision 2002/657/EC [15] and is presented here. We used pig plasma to perform this validation study due to the difficulty to obtain large amounts of poultry plasma.

in 0.002 mol L−1 HCl. The solid phase extraction (SPE) step was performed using Chromabond XTR-cartridges® 45 mL, 8300 mg from Macherey-Nagel (Filter Service, Eupen, Belgium). The LC mobile phase used to reconstitute the dried extracts before injection was prepared by mixing 93% of mobile phase A (0.1% acetic acid in water) with 7% of mobile phase B (acetonitrile). Individual stock standard solutions at 1 mg mL−1 in methanol were prepared and stored at 4 ◦ C for 1 year. Individual intermediate standard solutions (10 and 1 ␮g mL−1 ) and working standard solutions (mixture of (deuterated) nitroimidazoles) were prepared in the LC mobile phase (93:7) and stored at 4 ◦ C for 6 months.

2. Experimental

A quaternary delivery LC pump (HP 1100, Hewlett-Packard) linked to an automatic sampler (Gilson 231 XL, Gilson, The Netherlands) and coupled to a Quattro II triple quadrupole mass spectrometer (Micromass Inc., Manchester, UK) was used for the measurements. The separation was achieved using a reversed phase (C18) Genesis column from Jones Chromatography (4 ␮m; 250 mm × 3 mm I.D.) provided by Grace Vydac (USA) and maintained at 30 ◦ C using an oven. The chromatographic separation was performed in a gradient mode using water acidified with 0.1% acetic acid (mobile phase A) and acetonitrile (mobile phase B), at a flow rate of 0.6 mL min−1 . The initial conditions (0–16 min) were 93% acidified water and 7% acetonitrile. Then the conditions changed to 30% acidified water and 70% acetonitrile between 16 and 21 min and these proportions were maintained up to 25 min. Finally, the conditions returned to the initial ones in 2 min (25–27 min), and were maintained until the chromatographic run of 30 min ended. The ionisation mode used was the positive APCI mode, with the corona pin discharge

2.1. Chemicals and reagents All analytical standards of nitroimidazoles, including deuterated substances, were provided by the CRL (BVL, Berlin, Germany). Pig plasma was purchased from Centre d’Economie Rurale (CER, Laboratoire d’Hormonologie, Marloie, Belgium). Water was of Milli-Q quality (Millipore Corp., Bedford, MA, USA). All organic solvents were of LC analytical grade and purchased from VWR (Leuven, Belgium). A sodium chloride/potassium dihydrogen phosphate buffer (NaCl/KH2 PO4 ) was prepared by dissolving 5.84 g of NaCl (VWR, p.a. quality) and 13.61 g of KH2 PO4 (VWR, p.a. quality) in 950 mL of water, adjusted to pH 3 with 25% hydrochloric acid (HCl) and filled up to 1000 mL with water. For the enzymatic hydrolysis, Protease Type XVIII (EC no. 232-642-4) from Sigma (Bornem, Belgium) was used at a concentration of 80 mg mL−1

2.2. LC–MS/MS instrumentation

Table 1 Monitored transitions in MRM mode, collision energies, cone voltages and retention times of nitroimidazoles, hydroxynitroimidazoles and deuterated internal standards (IS) Analyte

Parent ion (m/z)

Daughter ions (m/z)

Collision energy (eV)

Cone voltage (V)

MNZ-OH

188 188 158 158 172 172 201 201 142 142 186 186 170 170 161 175 204 145 189 173

144 126 140 110 128 82 140 110 96 81 168 122 124 109 143 131 143 99 171 127

15 20 15 20 15 20 15 20 20 25 15 20 20 25 10 15 15 15 20 15

25 25 20 20 25 25 20 20 30 30 20 20 25 25 15 20 20 25 15 20

HMMNI MNZ RNZ DMZ IPZ-OH IPZ HMMNI-d3 (IS) MNZ-d3 (IS) RNZ-d3 (IS) DMZ-d3 (IS) IPZ-OH-d3 (IS) IPZ-d3 (IS)

Bold figures refer to the daughter ion transitions giving the most intense response and used for quantitation.

Retention time (min) 6.05 8.23 9.12 11.06 13.32 22.28 23.46 8.14 9.07 10.97 13.09 22.28 23.46

S. Fraselle et al. / Analytica Chimica Acta 586 (2007) 383–393

set at 3.5 kV. The different transitions of the nitroimidazoles were recorded in multiple reaction monitoring (MRM) mode for 30 min. The individual MRMs with their respective collision energies and cone voltages are listed in Table 1. Nitrogen was used as desolvation gas, with flows of 75 L h−1 (sheat gas) and 350 L h−1 (drying gas), while argon was used as collision gas at a pressure of 3 × 10−3 mbar. The source temperature was set at 150 ◦ C and the APCI probe temperature at 300 ◦ C. Dwell time and interchannel delay were 0.1 and 0.03 s, respectively. 2.3. Sample preparation One day before the extraction occurred, the whole plasma sample was first centrifuged to remove all suspended solid particles and 5 mL of the supernatant was weighted in a 50 mL propylene tube (Falcon® ). The internal standards (200 ␮L of a 50 ng mL−1 mixture of six deuterated nitroimidazoles, ISd3 solution) were added and the sample was shaken for 30 s using a Vortex® . After 15 min equilibration time, 2 mL of NaCl/KH2 PO4 buffer and 1 mL of protease solution were added. The pH value was adjusted to 3 with 25% HCl and the mixture was hydrolysed overnight at 37 ◦ C. The next day, the sample was centrifuged (10 min, 10,000 rpm) and the supernatant was transferred into a new 50 mL propylene tube (Falcon® ). The pH was adjusted to 3 with 25% HCl (if necessary) and the sample was defatted by a careful extraction with hexane. To perform this defattening step, 5 mL of hexane were added, the mixture was shaken for 1 min using a Vortex® and then centrifuged at 4 ◦ C (10 min, 3000 rpm). The lower aqueous phase was collected, by passing through the fat layer with a glass pipette, in a new 50 mL propylene tube (Falcon® ) of and then adjusted to pH 6 with 5 mol L−1 sodium hydroxide (NaOH). The mixture was transferred in a 15 mL propylene tube (Falcon® ), filled up to 8 mL with water and applied on a Chromabond XTR cartridge® without any preconditioning step. After 20 min equilibration time, 9 mL of a mix ethyle acetate/tertiary butylmethylether, 1:1 (v/v) were added. After another 15 min equilibration time, the Chromabond XTR cartridge® was eluted slowly (1 drop s−1 ) using twice 9 mL of ethyle acetate/tertiary butylmethylether, 1:1 (v/v) in a glass conical flask. The combined eluates were transferred into a glass tube and evaporated to dryness at 30 ◦ C under a stream of nitrogen. Just before the evaporation finished, 100 ␮L of a mix methanol/ethyleneglycol, 99:1 (v/v) were added

385

as “keeper” solution to ensure good recoveries. The dried extract was then dissolved in 400 ␮L of the LC mobile phase (mobile phase A/mobile phase B, 93:7 (v/v)) and filtered through a PVDF filter (0.45 ␮m, 4 mm; Alltech). An aliquot of 100 ␮L was injected into the LC–MS/MS system and the rest was kept at −20 ◦ C for maximum 48 h. 2.4. Validation study As nitroimidazoles are banned substances (Annex IV of Council Regulation (EEC) No. 2377/90 [1], group A6, Council Directive 96/23/EC [16]), the method was validated in-house according to the Commission Decision 2002/657/EC with regard to the “recommended validation levels” proposed by the CRL (BVL, Berlin, Germany) in 2002. These levels were for all matrixes (except fish) 1 ␮g L−1 (or ␮g kg−1 ) for MNZ, DMZ and IPZ; 2 ␮g L−1 (or ␮g kg−1 ) for RNZ and HMMNI while no recommendations were made for MNZ-OH and IPZ-OH due to a lack of available data. To perform their validation study, the authors used both recommended levels and 2.5 ␮g L−1 for compounds without recommended levels. However, these recommended levels are not minimum required performance limits (MRPL) strictly speaking, the abbreviation “MRPL” has been used in the text, as well as in all tables and figures for more clarity. According to the Commission Decision 2002/657/EC, the following parameters were determined: matrix effect, specificity, linearity, recovery, repeatability, intralaboratory reproducibility, decision limit (CC␣) and detection capacity (CC␤). 2.4.1. Matrix effect procedure To determine the (non) presence of a matrix effect the following protocol was followed. Solution standards were injected in parallel with matrix standards (corresponding to blank plasma samples fortified at the end of the sample preparation). The working concentrations were fixed to 0, 1/4, 1/2, 1, 1.5 and 2 times the “MRPL” for all nitroimidazoles and to 2 ␮g L−1 for the deuterated internal standards. These concentration levels, used to prepare both solution and matrix standards, are described in Table 2. Each standard, except standard 0, was prepared by mixing 200 ␮L of a 50 ng mL−1 IS-d3 solution, 100 ␮L of the LC mobile phase (93:7) and 100 ␮L of the A, B, C, D or E solution, respectively. The standard 0 was prepared by mixing 200 ␮L of a 50 ng mL−1 IS-d3 solution with 200 ␮L of the LC mobile phase

Table 2 Concentration levels (in ␮g L−1 ) of nitroimidazoles and hydroxy-metabolites used in the matrix effect procedure

MNZ DMZ IPZ RNZ HMMNI MNZ-OH IPZ-OH

1/4 “MRPL” (solution A)

1/2 “MRPL” (solution B)

1× “MRPL” (solution C)

1.5× “MRPL” (solution D)

2× “MRPL” (solution E)

0.25 0.25 0.25 0.5 0.5 0.625 0.625

0.5 0.5 0.5 1 1 1.25 1.25

1 1 1 2 2 2.5a 2.5a

1.5 1.5 1.5 3 3 3.75 3.75

2 2 2 4 4 5 5

“MRPL” refers to the recommended validation levels. a Working level for compounds without recommended validation levels.

386

S. Fraselle et al. / Analytica Chimica Acta 586 (2007) 383–393

Table 3 Summary of the procedure followed to determine all validation parameters. Each sample was performed in triplicate (replicate # 1, 2 and 3) at each concentration level. All samples analyzed on the same day (day # 1, 2 or 3) were injected with a solution standard calibration curve and with a week in between

1/4 “MRPL”

½ “MRPL” 1× “MRPL”

1.5× “MRPL”

2× “MRPL” Standard 0 ␮g L−1 (+2 ␮g L−1 IS-d3) Standard 0.5 ␮g L−1 (+2 ␮g L−1 IS-d3) Standard 1 ␮g L−1 (+2 ␮g L−1 IS-d3) Standard 2 ␮g L−1 (+2 ␮g L−1 IS-d3) Standard 5 ␮g L−1 (+2 ␮g L−1 IS-d3)

Day #1

Day #2

Day #3

Replicate #1 Replicate #2 Replicate #3 Replicate #1 Replicate #2 Replicate #3 Replicate #1 Replicate #2 Replicate #3 Replicate #1 Replicate #2 Replicate #3 Replicate #1 Replicate #2 Replicate #3 Calibration curve #1

Replicate #4 Replicate #5 Replicate #6 Replicate #4 Replicate #5 Replicate #6 Replicate #4 Replicate #5 Replicate #6 Replicate #4 Replicate #5 Replicate #6 Replicate #4 Replicate #5 Replicate #6 Calibration curve #2

Replicate #7 Replicate #8 Replicate #9 Replicate #7 Replicate #8 Replicate #9 Replicate #7 Replicate #8 Replicate #9 Replicate #7 Replicate #8 Replicate #9 Replicate #7 Replicate #8 Replicate #9 Calibration curve #3

“MRPL” refers to the recommended validation levels.

(93:7). All standards, in matrix as well as in solution, were performed in triplicate and 100 ␮L of each were injected into the LC–MS/MS system during the same run of analyses. 2.4.2. Validation procedure All other validation parameters but matrix effect (specificity, linearity, recoveries, repeatability, intralaboratory reproducibility, decision limit and detection capacity) were determined by the procedure described below. Blank plasma samples were fortified at the beginning of the extraction protocol at concentrations corresponding to 1/4, 1/2, 1, 1.5 and 2 times the “MRPL” (see Table 2) for nitroimidazoles and with a constant concentration of 2 ␮g L−1 for the deuterated internal standards. Analyses of samples were performed in triplicate over three different weeks, with at least a week in between. The spiked samples were injected with a five points calibration curve using solution standards and ranging from 0 to 5 ␮g L−1 . The three independent runs of LC–MS/MS analyses were also performed with a week in between. The complete applied procedure is described in Table 3. 3. Results and discussion 3.1. Specificity As depicted in Fig. 1, the method was specific for all nitroimidazoles, except for IPZ and IPZ-OH due to a very high noise at the retention time of both daughter ions for these two compounds in chromatograms of blank matrix and blank reagents samples. As an enzymatic hydrolysis using a protease was performed to avoid any adverse matrix effects in MS analysis, we suspected that these interfering “peaks” were coming from the clean-up step, either from the XTR phase itself or from interferences with plastic of the cartridges. But despite the use of

other lots of reagents and a different SPE phase (Hydromatrix® from Varian) coated in glass columns, these interferences were always present. This problem prevented any quantitation of IPZ and IPZ-OH, and allowed only a qualitative analysis for these two compounds. 3.2. Linearity To evaluate the (non) linearity of the regression model used to quantify, the Mandel’s Fitting test was applied [17]. The Mandel’s Fitting test evaluates whether an alternative regression model, in our case a quadratic model, better fits the data than a straight line regression model. The responses (area of each analyte divided by area of the corresponding IS-d3) of the three calibration curves in solution described in Section 2.4.2 were plotted versus the concentration of the five standards (0, 0.5, 1, 2 and 5 ␮g L−1 ). An “F-value” was calculated for each nitroimidazole according to the equations below: F=

(n − 2)Sy21 − (n − 3)Sy22 Sy22

with n the number of calibration levels, Sy1 and Sy2 the standard errors of the straight-line and the quadratic regression model, respectively calculated as followed:    2 (a + b Xi + c Xi2 − Yi ) Sy1 = n−2  (a + bXi − Yi )2 Sy2 = n−3 The obtained “F-values” were compared with a tabulated Fvalue, corresponding to the F-distribution with 1 and n-3 degrees of freedom and a probability of 99%. The F0.99 equals 9.33028

S. Fraselle et al. / Analytica Chimica Acta 586 (2007) 383–393

387

Fig. 1. MRM chromatogram of a blank plasma sample with traces for (top to bottom) IPZ-OH, IPZ, DMZ, RNZ, MNZ, HMMNI and MNZ-OH. The arrows point out the expected retention time of each compound. The chromatogram exhibits interfering “peaks” at the expected retention times of IPZ and IPZ-OH.

for all compounds. According to the Mandel’s fitting test, when the calculated “F-value” is below the F0.99 , the alternative regression model is not significantly better than the straight-line regression model. Based on the obtained “F-values” described in Table 4, a straight-line regression model was preferred. All calibration parameters obtained during the Mandel’s fitting test are summarized in Table 4. 3.3. Matrix effect As depicted in Fig. 2, the presence of a matrix effect was tested by comparing the slopes of both solution and

Table 4 Equation curves (for linear regression model), coefficients of determination (R) and calculated F-values obtained during the Mandel’s fitting test

HMMNI RNZ MNZ MNZ-OH IPZ IPZ-OH DMZ

Linear regression curve

R

F-value

y = 0.2969x + 0.0085 y = 0.4457x − 0.0002 y = 0.6653x + 0.0203 y = 0.1235x + 0.0022 y = 0.3683x + 0.038 y = 1.3295x + 0.0456 y = 0.2197x + 0.0159

0.996 0.999 0.996 0.994 0.991 0.999 0.997

0.61644 0.09412 0.53902 0.18542 0.00025 0.79832 0.00421

With F 0.99 = 9.330279 for all compounds.

388

S. Fraselle et al. / Analytica Chimica Acta 586 (2007) 383–393

Fig. 2. Comparison of matrix-matched (dotted line) and solution (continuous line) calibration curves obtained for each compound during the matrix effect procedure. The response corresponds to the analyte area divided by the corresponding IS-d3 area (except for MNZ-OH for which MNZ-d3 was used as internal standard) and has consequently no units.

S. Fraselle et al. / Analytica Chimica Acta 586 (2007) 383–393

389

Table 5 Equation curves, coefficients of determination (R) and calculated t-factors obtained during the matrix effect procedure (for both solution and matrix-matched curves)

HMMNI RNZ MNZ MNZ-OH IPZ IPZ-OH DMZ

Solution curve

R solution

Matrix-matched curve

R matrix

t-factor

y = 0.1121x + 0.0224 y = 0.4359x + 0.0147 y = 0.3933x + 0.0294 y = 0.0467x + 0.0157 y = 0.168x + 0.0135 y = 0.9584x + 0.1543 y = 0.085x + 0.0089

0.989 0.992 0.996 0.987 0.993 0.981 0.994

y = 0.1084x + 0.0154 y = 0.4705x − 0.0269 y = 0.3919x + 0.0174 y = 0.0493x + 0.0114 y = 0.1205x + 0.252 y = 1.0494x + 0.1259 y = 0.0837x + 0.0127

0.986 0.996 0.997 0.993 0.916 0.988 0.997

0.5896 1.9822 0.1154 1.0385 3.3011 1.4121 0.4417

With tabulated t = 2.0369 for all compounds.

matrix-matched calibration lines described in the “matrix effect procedure” (see Section 2.4.1) by means of a Student’s t-test, using equations below [18]: t(b) =  Sp  Sp =

1/



|bs − ba |     ¯ s 2 + 1/ ¯a 2 Xi,s − X Xi,a − X

(na − 2)Ss2 + (na − 2)Sa2 ns + n a − 4

matched) together with the calculated t-factors. According to the calculated t-factors and by comparison to the tabulated t (2.0369 for all compounds), no matrix effect was present except for IPZ. Since the method was not specific for IPZ and IPZ-OH (see Section 3.1), it was concluded that calibration curves in solution could be used to quantify all other nitroimidazoles. 3.4. Apparent recoveries

with b the intercept of the equation curve, X the concentration, n the number of replicates (in this case 18) and S the standard deviation of the slope. The indexes s (solution) and a (addition) refer to the standard solution and matrix-matched equation curves respectively. There is no significant difference between the slopes of both calibration curves with a probability of 95%, meaning that no matrix effect is present, when the Student t is below the tabulated t with ns + na − 4 degrees of freedom. Curves equations and correlation coefficients are summarized in Table 5 for each compound and for both calibration lines (solution and matrix-

The apparent recovery, or recovery obtained after correction with a deuterated standard (different from extraction yield [19]) was calculated for all compounds according to the procedure described in Section 2.4.2. An average (of the nine replicates described in Table 3) apparent recovery was evaluated at each level of concentration. The obtained values are presented in Table 6. All apparent recoveries obtained were within the range of 50–120% authorized for a multi residues method, except for DMZ at 1/4 “MRPL”. The lower values obtained for MNZ-OH were probably due to the use of MNZ-d3 as internal standard since no deuterated compound corresponding to the hydroxy

Table 6 Apparent recoveries (average of nine replicates) obtained at each level of concentration

MNZ DMZ RNZ HMMNI MNZ-OH

Recommended validation levels “MRPL” (␮g L−1 )

1/4 “MRPL” (%)

1/2 “MRPL” (%)

1× “MRPL” (%)

1.5× “MRPL” (%)

2× “MRPL” (%)

1 1 2 2 2.5a

97 123 94 105 58

97 114 97 109 58

100 112 97 108 63

97 110 93 107 62

101 100 98 105 62

“MRPL” refers to the recommended validation levels. a Working level for compounds without recommended validation levels. Table 7 Relative standard deviations (average of nine replicates) for repeatability and intralaboratory reproducibility obtained at each level of concentration “MRPL” (␮g L−1 )

MNZ DMZ RNZ HMMNI MNZ-OH

1 1 2 2 2.5a

1/4 “MRPL”

1/2 “MRPL”

1× “MRPL”

1.5× “MRPL”

2× “MRPL”

CVr

CVRw

CVr

CVRw

CVr

CVRw

CVr

CVRw

CVr

CVRw

12.39 3.26 6.89 7.85 12.10

12.39 12.63 8.26 9.13 16.38

2.57 5.86 3.90 5.10 6.42

6.37 10.66 11.79 11.52 11.76

2.99 5.50 3.16 4.43 4.33

4.68 9.39 3.79 6.19 14.74

4.38 3.93 12.69 4.53 8.08

4.38 10.01 12.69 5.35 16.39

2.49 6.70 4.52 4.43 6.05

2.49 6.70 6.78 7.93 14.62

“MRPL” refers to the recommended validation levels. a Working level for compounds without recommended validation levels.

390

S. Fraselle et al. / Analytica Chimica Acta 586 (2007) 383–393

Fig. 3. MRM chromatogram of a blank plasma sample fortified at 1 ␮g L−1 (DMZ, MNZ), 2.5 ␮g L−1 (MNZ-OH) and 2 ␮g L−1 (RNZ, HMMNI and all deuterated internal standards).

metabolite of MNZ was available. A chromatogram of a blank plasma sample spiked at the “MRPL” levels and at 2 ␮g L−1 for IS-d3 is depicted in Fig. 3. The concentration values obtained during all the validation protocol were not corrected with apparent recoveries due to isotopic dilution with deuterated internal standards.

The CVr and the CVRw calculated according to the ISO 5725-2 guidelines [20] ranged from 2.49% to 12.69% and from 2.49% to 16.39%, respectively. The Horwitz equation was respected in all cases, with CV below the maximum allowed values (Horwitz equation modified for amounts up to 100 ␮g kg−1 ) of 14.7% for repeatability and 22% for intralaboratory reproducibility (Table 7).

3.5. Repeatability and intralaboratory reproducibility 3.6. Decision limit and detection capacity The repeatability (r) and the intralaboratory reproducibility (Rw) were evaluated by calculating the coefficients of variation (CV) obtained for these two parameters during the procedure described in Section 2.4.2 at each level of concentration and according to the equation CV (%) = 2(1–0.5 log C) , with C the concentration expressed as fraction (e.g. 1 ␮g L−1 = 10−3 ).

The decision limit (CC␣) and the detection capacity (CC␤) were first determined using the data obtained at Section 2.4.2 and according to the ISO 11843-2 guidelines [21]. These two parameters were calculated (see equations below) using a weighted regression linear model assuming that the standard deviation

S. Fraselle et al. / Analytica Chimica Acta 586 (2007) 383–393

391

Table 8 Decision limits (CC␣) and detection capacities (CC␤) obtained using ISO 11843-2 procedure (applied on both diagnostic ion transitions) and based on identification criteria checking “MRPL” (␮g L−1 )

MNZ DMZ RNZ HMMNI MNZ-OH

1 1 2 2 2.5c

ISO 11843a 1st transition

ISO 11843a 2nd transition

Based on ID-criteriab

CC␤ (␮g L−1 )

CC␣ (␮g L−1 )

CC␤ (␮g L−1 )

CC␣ (␮g L−1 )

CC␤ (␮g L−1 )

CC␣ (␮g L−1 )

0.12 0.17 0.10 0.13 0.14

0.07 0.10 0.06 0.08 0.08

0.05 0.22 1.39 0.81 0.23

0.03 0.13 0.82 0.48 0.14

0.25 0.5 1 1 1.25