Merja Gergov, Timo Nenonen, Ilkka Ojanpera and Raimo A. Ketola*

Journal of Analytical Toxicology 2015;39:359 364 doi:10.1093/jat/bkv020 Advance Access publication March 5, 2015 Article Compensation of Matrix Effects in a Standard Addition Method for Metformin in Postmortem Blood Using Liquid Chromatography Electrospray Tandem Mass Spectrometry Merja Gergov, Timo Nenonen, Ilkka Ojanpera and Raimo A. Ketola* Department of Forensic Medicine, Faculty of Medicine, University of Helsinki, PO Box 40, FI-00014 University of Helsinki, Finland *Author to whom correspondence should be addressed. Email: raimo.ketola@helsinki.fi This work describes a procedure to evaluate matrix effects in a combined dilution and standard addition method (SAM) using liquid chromatography electrospray tandem mass spectrometry. The method was validated and applied to an analysis of metformin in postmortem blood samples. The analytical method included protein precipitation with methanol, followed by liquid chromatographic separation of metformin on Gemini NX-C 18 reversed-phase column using a gradient consisting of methanol and ammonium acetate at ph 3.2. The mass spectrometric analysis was performed with a quadrupole-linear ion trap mass spectrometer equipped with a turbo ion spray interface in a positive ion mode using selected reaction monitoring. Quantitation was performed based on an SAM. Validation for metformin revealed a practical limit of quantification of 0.1 mg/l, a linear range from 0.1 to 3.0 mg/l, average precision 10%, accuracy (bias) 9% and reproducibility 10%. Combined matrix effects were evaluated by k-values (slopes) of calibration plots, postextraction addition approach and a comparison of within- and between-sample precision (relative standard deviation). It was demonstrated that the method contained matrix effects which were fully compensated for using dilution and the SAM. Introduction Proper calibration is essential in quantitative analysis in order to obtain reliable results from complex biological samples. Common calibration methods include external standard, internal standard (ISTD) and matrix-matched calibration methods. The ISTD calibration is commonly used in bioanalysis; however, even ISTDs cannot perfectly correct the variability of sample matrix; therefore, the use of matrix-matched standards is preferable. In a matrix-matched method, the standards are prepared in a similar matrix as the sample itself, such as urine, plasma or blood. If the sample is putrefied, like in some postmortem cases, even the use of matrix-matched standards is not perfect. In putrefied samples, the degree and amount of degradation of cells can vary a lot, as well as the amount and quality of lipids. The fourth choice of calibration method, especially suitable for hydrophilic/polar drugs in putrefied samples, is a standard addition method (SAM), where the actual samples are used to create a calibration plot individually (1). The main advantage of this method is that it can correct the matrix effect because exactly the same sample matrix is present both in calibration standards and the sample itself. A drawback of the method is, for example, that the amount of sample needed is much higher than in other methods because the sample is needed also for calibration standards. Another drawback is that it is more tedious to prepare the calibration plot for each sample separately. However, good results These authors contributed equally. have been obtained with the SAM for biological samples. For example, phosphatidylcholines with varying carbon chain lengths in a rat liver tissue were analyzed by nanoflow LC MS-MS (2). With the SAM, the effect of both sample matrix and the carbon chain length could be effectively compensated. A challenging sample matrix for organic analysis is placenta. The research group of Zafra-Go mez used the SAM for the determination of parabens and benzophenones from placental tissues using LC MS-MS and atmospheric pressure chemical ionization (APCI) technique (3, 4). The methods showed good variability as the inter- and intra-day variabilities were under 13.8 and 5% for parabens and benzophenones, respectively. Kuepper et al. employed an SAM with ion-pairing solid-phase extraction and subsequent HPLC MS-MS analysis for detection of succinylmonocholine from various biological postmortem samples, such as the brain, liver, kidney and vitreous humor (5). The method worked well not only for obtaining reliable quantitative results from the samples but also for confirming possible negative results for some samples. Mitamura et al. also applied an LC MS-MS with the SAM for quantitative determination of pregnenolone-3-sulfate in rat brains (6). This study shows another benefit of the SAM, namely it is possible to detect lower concentrations of analytes than actual instrumental limits of quantitation (LOQ), if the method is precise enough. For example, in their study, the instrumental LOQ was 4 ng/ml; but by the standard addition, it was possible to accurately quantitate concentrations at 0.5 ng/ml (the difference between the actual and the added concentrations). In a comparison of different calibration approaches using multivariate curve resolution for LC MS analysis of biocides from environmental samples, it was noticed that the SAM with ISTDs gave the most precise results for the analytes (7). Peters et al. have presented a comprehensive evaluation of validation procedures and criteria, which could also be used as a guideline in validation of postmortem analytical methods (8). Besides calibration, another critical issue in LC MS analysis is the matrix effect as already stated earlier. Different aspects of matrix effects in LC MS analysis for forensic and clinical toxicology were recently comprehensively reviewed by Peters and Remane (9). The review highlights the importance of studying matrix effects in forensic toxicology during method development. It was concluded that postmortem samples usually possess more matrix effects than clinical samples, and it is practically impossible to remove matrix effects completely in bioanalysis using LC MS. Several procedures are available to measure the degree of matrix effects, such as post-column standard infusion and postextraction addition (PEA) methods (10, 11), of which the latter is now commonly used by researchers as it gives a measure of efficiency of the whole analytical process. However, matrix effects have been very rarely studied in SAMs. Basilicata et al. applied two calibration procedures, the ISTD and # The Author 2015. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com

the SAMs, for the determination of urinary benzene of workers in occupational exposure studies (12). The conclusion of the study was that the SAM effectively compensated the matrix effects of individual urine samples, and a three-point standard addition protocol was enough for obtaining reliable quantitative results. The relationship between matrix concentration and suppression of electrospray ionization was investigated by Stahnke et al. using pesticides present in QuEChERS extracts as an example (13). They noticed that dilution of extracts by a factor of 25240 reduces ion suppression to,20% if the initial suppression is 80%. Five different methods (matrix-matched calibration, standard addition, post-column standard infusion, extrapolative dilution and post-column flow splitting) were compared in accounting for or overcoming matrix effects in LC ESI MS determination of pesticides from food samples (14). Extrapolative dilution is a new form of dilution method where the sample is first diluted prior to sample treatment and analysis to reduce matrix effect, and the final result is obtained by extrapolating to infinite dilution. In this study, extrapolative dilution and SAMs were found to give results that were not statistically different from the correct values. The main aim of the study was to investigate the efficiency of the method in compensating matrix effects encountered in postmortem blood samples and hence to improve the accuracy of the method. A validation procedure of the method is described here, for the determination of metformin in postmortem blood samples using protein precipitation and dilution combined with standard addition and LC ESI MS-MS. Experimental Reagents Metformin was purchased from Weifa AS (Kragero, Norway). Methanol Fluka LC MS Chromasolv w (99.9%) was purchased from Sigma Chemical (St. Louis, MO, USA) and the other solvents from Merck (Darmstadt, Germany). Ammonium acetate, formic acid and sodium phosphate were supplied by Merck. All solvents and reagents were of HPLC or analytical grade. Laboratory water was purified with a Milli-Q Integral 5 system (Millipore, Bedford, MA, USA). Bovine whole blood (blank blood) and authentic whole blood samples from routine medico-legal autopsies served for method development. The time interval between death and autopsy was normally from 1 to 6 days, thus the degree of putrefaction varied from sample to sample. Sample preparation for validation Whole blood sample (0.5 ml) was initially diluted with laboratory water using a dilution factor of 10. The diluted sample was pipetted into five test tubes (0.5 ml each), four of which were fortified with metformin at concentrations of 0.1, 0.3, 1.0 and 3.0 mg/l for calibration. Samples (0.5 ml) were precipitated with 1 ml of methanol by vortexing for 1 min and centrifuged at 4,330g for 5 min. Finally, 100 ml of the aqueous layer was transferred to an autosampler vial and diluted with 900 ml of laboratory water by vortexing. For validation, blank blood samples were first fortified with metformin to produce metformin concentration of 1, 10 or 100 mg/l prior to dilution. Samples with high concentrations of metformin (10 and 100 mg/l) were diluted before standard additions using higher dilution factors than 10, namely 50 or 500. Thus, in the final extract for the LC MS-MS analysis, the original samples were diluted by an absolute factor of 300 15,000. Instrumental analysis LC separations were carried out with a Shimadzu Prominence HPLC system consisting of two LC-20AD pumps, SIL-20AHT autosampler, DGU-20A5 vacuum degasser and CTO-20A column oven (Shimadzu USA Manufacturing, Inc., Canby, OR, USA). A Phenomenex Gemini NX-C 18 column (100 2.1 mm; 3 mm) and Phenomenex C 18 guard column (4 2 mm) were used in chromatographic separations (Phenomenex, Torrance, CA, USA). The mass spectrometric analysis was performed using an AB Sciex 3200 Q TRAP w LC MS-MS System instrument equipped with a Turbo V TM source and a TurboIonSpray w probe (AB Sciex, Concord, ON, Canada) in triple quadrupole mode. The software was Analyst 1.5.1. The analytical column was stabilized at 408C. The mobilephase gradient consisted of 10 mm ammonium acetate buffer (ph 3.2) and methanol (both containing 0.1% of formic acid) as follows: the methanol proportion was held at 10% during the equilibrium time of 7 min. After injection, the methanol proportion was increased to 30% over 4 min, then to 95% in 3 min, kept at 95% for 1 min to clean the column and finally decreased back to 10% in 2 min. The total flow rate through the column was 0.25 ml/min. The injection volume was 10 ml. The total flow from the LC was directed to the ion source without splitting. Needle voltage was 5.0 kv, declustering potential 30 V and ion spray heater temperature 4508C. The curtain gas (nitrogen) was set at 12 and collision gas (nitrogen) pressure in the collision cell at 5 in the Analyst control software. The nebulizer gas (nitrogen) was set at 60 psi (414 kpa) and turbo heater gas (nitrogen) at 70 psi (482 kpa). A selected reaction monitoring (SRM) mode was used for metformin with transitions m/ z 130! 71 and 130! 85, and the corresponding values for collision energy were 30 and 20 ev. A dwell time of 80 ms was used for the transitions. Quantitation method An SAM was applied for quantification. Metformin was fortified to the samples at known concentrations, and signal intensities of the non-fortified and fortified samples were measured. A calibration plot of the peak areas vs. standard additions of the spikes yielded a straight line with linear regression (Figure 1f). The concentration of metformin was calculated from the point (x-axis) at which the extrapolated linear regression line crosses the concentration axis at zero signal. Results Method validation Validation included assessment of selectivity, linearity, lower and higher limits of quantification, precision, accuracy, reproducibility and combined matrix effects, and it was based on the protocol proposed by Peters et al. (8). Summary of the validation results is presented in Table I. 360 Gergov et al.

Figure 1. A negative autopsy blood sample diluted with laboratory water (a dilution factor of 10) with standard additions of (A) 0, (B) 0.1, (C) 0.3, (D) 1.0 and (E) 3.0 mg/l (SRM chromatograms of a transition of m/z 130! 71 for metformin). (F) Calibration plots prepared from the negative sample and a positive sample containing 1.9 mg/l metformin (a dilution factor of 10), and the structure of metformin. Table I Validation Results for Metformin in Postmortem Blood Using a Dilution and SAM with Protein Precipitation and LC ESI MS-MS Parameter Value Accuracy (bias%) 9 + 3 Precision (%) 10 + 2 Linear range (mg/l) 0.1 3.0 Lower limit of quantification (mg/l) a 0.1 Reproducibility (bias%) 10 + 5 a A practical lower LOQ. Selectivity/specificity Selectivity was assured by using a triple quadrupole system in the MS-MS mode. Selectivity was evaluated by analyzing 48 autopsy blood samples known to be negative for metformin but after routine drug screening found to contain other drugs. These other drugs did not give any signal interfering with the SRM transitions of metformin. Possible interfering signals were monitored. Selectivity was proven by the absence of any interfering signals at the monitored SRM traces of the studied 48 authentic autopsy samples (Figure 1A). Linearity Blank blood samples were first fortified at 0 (no fortification), 1, 10 and 100 mg/l concentrations of metformin. Five-point calibration plots were prepared in duplicates at each of the concentrations, using a standard addition set of 0, 0.1, 0.3, 1.0 and 3.0 mg/l of metformin. To evaluate linearity also in authentic sample matrix, calibration plots were prepared in 48 laboratory s routine autopsy samples, known to be negative for metformin. For determination of the linear range, the absolute peak area was plotted against the drug concentration. The following criteria were applied for linear range: linear regression with a correlation coefficient (r 2 )of.0.995, bias from the calibration plot of,20% for all individual calibration points and precision as relative standard deviation (RSD),20%. The linear range both in blank blood and in the authentic blood samples was from 0.1 to 3 mg/ L. The average correlation coefficient (r 2 ) was 0.9994 in blank blood and 0.9979 in the autopsy samples. Higher metformin concentrations, 10 and 100 mg/l, resulted in calibration plots with quadratic regression, and therefore, samples with higher concentration than 3 mg/l must be diluted with laboratory water using a higher dilution factor than 10 to reach the linear range from 0.1 to 3 mg/l. Compensation of Matrix Effects by Standard Addition Method 361

Accuracy, precision, recovery and lower and upper limits of quantification (LLOQ and ULOQ) Precision, accuracy, recovery and LLOQ were evaluated using blank blood and six authentic negative autopsy blood samples fortified at metformin concentrations of 1, 10 and 100 mg/l. Five-point calibration plots were prepared using standard addition set of 0, 0.1, 0.3, 1.0 and 3.0 mg/l. Two replicate calibration plots were prepared for all three fortified concentrations. Precision was calculated as within-sample intraday RSD of the replicates, and it was averagely 8, 11 and 8% at metformin concentrations of 1, 10 and 100 mg/l, respectively. Accuracy was expressed as average bias between fortified and measured concentrations, and it was 11, 6 and 8% at metformin concentrations of 1, 10 and 100 mg/l, respectively. In this case, the bias served also as an indicator of the recovery; therefore, the recovery was 89, 94 and 92% at 1, 10 and 100 mg/l, respectively. LLOQ was established as the lowest added concentration fulfilling the following criteria: bias from the calibration plot,20%, precision (RSD),20%, signal-to-noise ratio at least 10 and a symmetrical peak shape. The lowest standard addition was 0.1 mg/l, which served as a practical LLOQ. At LLOQ, bias from the calibration plot varied between 2 and 10% and precision 8 and 15% in the autopsy samples. Signal-to-noise ratio at 0.1 mg/l was larger than 45 in all authentic samples. ULOQ (3 mg/l) was defined as the upper limit of the linear range when the initial dilution factor of 10 was used. Examples of calibrations plots, prepared in autopsy blood samples, are presented in Figure 1. Intra-laboratory reproducibility Metformin was considered stable in blood samples and on a freeze thaw cycle. Therefore, reproducibility was evaluated from the re-analysis of 10 autopsy blood samples after a few days (six samples) to several months (four samples) interval of time from the first analysis. Metformin concentrations in the original analyses varied from 1.3 to 52 mg/l. Reproducibility was calculated as bias of the new analytical result from the original one, and on average, it was 9.6%, ranging from 0.5 to 17.3%. Combined variation from matrix In matrix effect tests, three strategies were applied: (i) variation in slopes (k-values) of calibration plots (10), (ii) PEA approach (9) and (iii) a comparison of within- and between-sample RSD (15). Results of the matrix effect tests are summarized in Table II. Table II Summary of Matrix Effect Results Using (a) Variation in the k-values and r 2 of the Calibration Plots, (b) Post-Extraction Addition and (c) Comparison of Within- and Between-Sample RSDs Calibration plots a Post-extraction addition b Comparison of RSD (%) c k-value RSD (%) 42 ME% RE% PE% Wi/Bw (area) Wi/Bw (conc.) r 2 RSD (%) 0.43 1 mg/l 70 93 65 4/7 8/10 10 mg/l 100 79 78 4/8 11/6 100 mg/l nd nd nd 4/8 8/11 nd, not determined. a RSD of the k-values and correlation coefficient r 2 of the calibration plots in 48 autopsy samples. b Matrix effect (ME), recovery (RE) and process efficiency (PE) in six autopsy samples. c Within (Wi) and between (Bw) sample RSD of metformin peak areas and concentrations in six autopsy samples. Variation in k-values of the calibration plots The matrix effects were estimated from the k-values of calibration plots in 48 authentic negative autopsy blood samples using a dilution factor of 10 prior to standard addition. Straight calibration plots were obtained by using linear regression. Variability of the slopes served as an indicator of matrix effect. Precision of the slopes should not exceed 3 4% for a method to be considered free of matrix effect. In the case of metformin, comparison of the plots revealed intensive matrix effects, while variability of the k-value was 42% (Figure 2). These samples were analyzed within a time period of 6 months; thus, variation in the performance of the LC-ESI MS-MS instrument (shifts in the signal intensities) was included in the variation in the k-values. However, a comparison of correlation coefficients (r 2 ) revealed very good linearity with an average value of 0.9979 and an RSD of 0.43% (Figure 2). PEA approach The second strategy was to estimate matrix effect (ME), recovery (RE) and overall process efficiency (PE) by preparing three sets of calibration plots: pure metformin fortified in laboratory water (Set 1), authentic autopsy blood samples fortified after sample preparation (Set 2) and the same samples fortified before the sample preparation (Set 3). ME (%), RE (%) and PE (%) were calculated by comparing the corresponding absolute peak areas of metformin as a ratio of Set 2 to Set 1, Set 3 to Set 2 and Set 3 to Set 1, respectively. Six samples from the k-value evaluation were fortified with 1 and 10 mg/l. As metformin concentrations in authentic postmortem blood samples can be much.3 mg/l, all samplesweredilutedpriortostandardadditionandthereby the effect of dilution was studied. Two dilution factors for different analyte concentrations were tested: 10 for 1.0 mg/l and 50 for 10 mg/l. Calculations showed that the ME on absolute peak areas depended strongly on the dilution factor. At 1 mg/l concentration (dilution factor of 10), suppression was detected and ME was 70%, whereas at 10 mg/l concentration (dilution factor of 50), there was no matrix effect (ME 100%). The corresponding RE and PE percentages were 93 and 65% at 1 mg/l of metformin and 79 and 78% at 10 mg/l, respectively. Comparison of within- and between-sample precision For the third approach, six authentic autopsy blood samples were fortified in two replicates with metformin at concentrations of 1, 10 and 100 mg/l. The samples were known to be negative for metformin but after routine drug screening found to contain other drugs. The within-sample precision was calculated as the average RSD of the replicates, whereas the betweensample RSD was calculated from the averaged values of the six samples. Calculations were performed both from the measured concentrations and absolute peak areas. In comparison of withinand between-sample RSD, a significantly higher between-sample RSD was tentatively regarded as an indication of combined effects of matrix and variation in extraction efficiency. Calculation from concentrations revealed a within-sample RSD of 8 11% and a between-sample RSD of 6 11%, whereas the corresponding values from absolute peak areas were 4.2 4.4 and 6.6 8.4%, respectively. Statistical evaluation was done using one-way analysis of variance (ANOVA) with 0.05 as the critical P-value. Calculations revealed that differences in the within- and between-sample RSD 362 Gergov et al.

Figure 2. Calibration plots prepared for 48 authentic negative autopsy blood samples using a dilution factor of 10, standard addition, protein precipitation and LC ESI MS-MS. A transition of m/z 130! 71 was monitored. values were significant in the absolute peak areas, but not in the measured concentrations. Moreover, the differences in RSDs were systematically more significant at higher metformin concentrations and standard additions at which the peak areas were greater. This was considered as an indication of matrix effects in absolute peak areas, which was demonstrated also by k-value and PEA approaches. However, matrix had no significant effect on the measured concentrations, as can be concluded also from the obtained accuracies, which in this test were 11, 6 and 8%, expressed as bias from the fortified metformin concentrations of 1, 10 and 100 mg/l, respectively. Discussion Metformin is a very polar compound (Figure 1), and therefore, a fast, easy and rugged protein precipitation method was applied. Unfortunately, protein precipitation is reported to suffer from serious matrix effects (16). In the present study, protein precipitation was combined with dilution and an SAM for quantification. The accuracy and the matrix effects of this combined analytical method were evaluated in three different ways: comparison of the k-values (slopes) of the calibration plots, the PEA approach, and within- and between-sample precision ( peaks areas and concentrations). All approaches revealed matrix effects to a similar extent, 30 40%. The PEA approach showed a PE of 65 78%, but the accuracy could not be deduced from this experiment. Similarly, the comparison of k-values of the calibration plots did not provide straightforward information on accuracy of the method. The comparison of within- and between-sample precision showed clear matrix effects on the analyte peak areas, but not on final analyte concentrations. This proved that the effect of matrix on quantitative results could be compensated using the SAM, thereby giving good accuracy. This is because each sample has an individual calibration plot, and exactly the same matrix is present both in the sample and in all of the calibration points. This result is in accordance with results obtained by Pere -Trepat et al. (7) andbasilicataet al. (12). In another method for the analysis of metformin from postmortem blood, the matrix effects could be eliminated using matrix-matched calibrants with isotopically labeled metformin as an ISTD, protein precipitation and cation-exchange solid-phase extraction followed by hydrophilic interaction LC ESI MS-MS (17). The remaining ion suppression was,5%. However, there was no information on the decomposition status of the postmortem samples and the variability of matrix effects between individual samples. Recovery of the method presented was evaluated in two ways: fortifying in a traditional way and using the PEA method. Both approaches gave similar recoveries, ranging from 76 to 93%, thus proving that both approaches could be used for reliable determination of recovery. There was no direct correlation observed between the degree of putrefaction and validation results, thus indicating that the dilution and SAM method was suitable also for decomposed blood samples. With the present method, the upper limit of linearity range of metformin concentration was 3 mg/l. Samples with a higher metformin concentration resulted in quadratic calibration plots in which accuracy of the lowest addition (0.1 mg/l) did not fulfill the acceptance criteria (bias from calibration plot 20%). Therefore, blood samples were diluted using a dilution factor of 10 500. Dilution factor tests, as well as earlier experience from the laboratory s other SAMs, show that determination of analyte concentration is more accurate when higher dilution factors are used. This is apparently because the matrix compounds are also diluted, thus reducing the matrix effects. When the dilution factor was large enough compared with the original concentration of the analyte, the average correlation coefficient r 2 was better than 0.997 with excellent precision. In routine analysis, where the analyte concentration in a sample is unknown, it is impossible to predict the correct level of dilution factor beforehand. Therefore, critical evaluation of the correlation coefficient r 2 and the accuracy of the lowest additions in the calibration plot is essential. The PEA approach is efficient in determination of ME, RE and PE; however, it leads to a large number of measurements, and Compensation of Matrix Effects by Standard Addition Method 363

therefore, it is time-consuming and laborious. In postmortem studies, the matrix is much more variable than in clinical samples; therefore, we consider at least 20 authentic samples with two replicates to be a minimum number of measurements to represent a postmortem matrix. When authentic postmortem samples are used for validation, the sample amount available may also become a restricting factor for the number of experiments. With the SAM method, even a restricted assessment of the matrix effect (one concentration level, a five-point calibration plot, 2 replicates of 20 authentic samples) would include 410 measurements by PEA, but only 200 measurements by comparison of within- and between-sample precision. Thereby, the comparison of the within- and between-sample precision apparently is an easier approach, revealing the most important parameters of validation, namely the combined matrix effect, accuracy and precision. In the case of metformin, observation of significant matrix effects on absolute peak areas, but not on the measured concentrations, clearly demonstrated the benefits of SAM. In general, if dilution is needed, like in this case, it will increase the number of experiments, especially if different dilution factors will be evaluated. Based on the results obtained in this study and by Stahnke et al. (13); however, we suggest that evaluation of only two different dilution factors is enough as long as the difference of these factors is at least 10-fold. In our method, several levels of additions were used, due to a large range of metformin concentrations in authentic samples and still a limited linear measurement range with the instrument. In the analysis of compounds with a broad measurement range, it could be possible to use only one addition level to obtain reliable quantitative results. Conclusions Protein precipitation method is a very quick and easy method, but it results in serious matrix effects for postmortem blood samples. A comparison of different approaches for matrix effect assessment showed that all three approaches studied revealed matrix effects in the analytical method, and PEA approach could also estimate the extent of matrix effects. The comparison of within- and between-sample precision not only revealed the presence of matrix effects in absolute peak areas but also proved that these effects were efficiently compensated by dilution and SAM producing good accuracy and reliable quantitative results. The work load in the comparison of within- and between-sample precision approach was also less than in the PEA approach. References 1. Massart, D.L., Vandeginste, B.G.M., Buydens, L.M.C., De Jong, S., Lewi, P.J., Smeyers-Verbeke, J. (1997) Standard addition method. In Vandeginste, B.G.M., Rutan, S.C. (eds), Handbook of Chemometrics and Qualimetrics, Part A, Elsevier, Amsterdam, The Netherlands, p. 207. 2. Ahn, E.J., Kim, H., Chung, B.C., Moon, M.H. (2007) Quantitative analysis of phosphatidylcholine in rat liver tissue by nanoflow liquid chromatography/tandem mass spectrometry. Journal of Separation Science, 30, 2598 2604. 3. Jime nez-dı az, I., Vela-Soria, F., Zafra-Gómez, A., Navalo n, A., Ballesteros, O., Navea, N. et al. (2011) A new liquid chromatography tandem mass spectrometry method for determination of parabens in human placental tissue samples. Talanta, 84, 702 709. 4. Vela-Soria, F., Jiménez-Díaz, I., Rodríguez-Gómez, R., Zafra-Go mez, A., Ballesteros, O., Navalo n, A. et al. (2011) Determination of benzophenones in human placental tissue samples by liquid chromatography tandem mass spectrometry. Talanta, 85, 1848 1855. 5. Kuepper,U.,Musshoff,F.,Madea,B.(2011)Applicabilityofsuccinylmonocholine as a marker for succinylcholine administration comparative analysis of samples from a fatal succinylcholine-intoxication versus postmortem control specimens. Forensic Science International, 207, 84 90. 6. Mitamura, K., Yatera, M., Shimada, K. (1999) Quantitative determination of pregnenolone-3-sulfate in rat brains using liquid chromatography/electrospray ionization-mass spectrometry. Analytical Sciences, 15, 951 955. 7. Peré-Trepat, E., Lacorte, S., Tauler, R. (2007) Alternative calibration approaches for LC MS quantitative determination of coeluted compounds in complex environmental mixtures using multivariate curve resolution. Analytica Chimica Acta, 595, 228 237. 8. Peters, F.T., Drummer, O.H., Musshoff, F. (2007) Validation of new methods. Forensic Science International, 165, 216 224. 9. Peters, F.T., Remane, D. (2012) Aspects of matrix effects in applications of liquid chromatography mass spectrometry to forensic and clinical toxicology a review. Analytical and Bioanalytical Chemistry, 403, 2155 2172. 10. Matuszewski, B.K., Constanzer, M.L., Chavez-Eng, C.M. (2003) Strategies for the assessment of matrix effect in quantitative bioanalytical methods based on HPLC-MS/MS. Analytical Chemistry, 75, 3019 3030. 11. Matuszewski, B.K. (2006) Standard line slopes as a measure of a relative matrix effect in quantitative HPLC MS bioanalysis. Journal of Chromatography B, 830, 293 300. 12. Basilicata, P., Miraglia, N., Pieri, M., Acampora, A., Soleo, L., Sannolo, N. (2005) Application of the standard addition approach for the quantification of urinary benzene. Journal of Chromatography B, 818, 293 299. 13. Stahnke, H., Kittlaus, S., Kempe, G., Alder, L. (2012) Reduction of matrix effects in liquid chromatography electrospray ionization mass spectrometry by dilution of the sample extracts: how much dilution is needed? Analytical Chemistry, 84, 1474 1482. 14. Kruve, A., Leito, I. (2013) Comparison of different methods aiming to account for/overcome matrix effects in LC/ESI/MS on the example of pesticide analyses. Analytical Methods, 5, 3035 3044. 15. Gergov,M.,Nokua,P.,Vuori,E.,Ojanpera,I.(2009)Simultaneous screening and quantification of 25 opioid drugs in post-mortem blood and urine by liquid chromatography tandem mass spectrometry. Forensic Science International, 186, 36 43. 16. Chambers, E., Wagrowski-Diehl, D.M., Lu, Z., Mazzeo, J.R. (2007) Systematic and comprehensive strategy for reducing matrix effects in LC/MS/MS analyses. Journal of Chromatography B, 852, 22 34. 17. Sørensen, L.K. (2012) Determination of metformin and other biguanides in forensic whole blood samples by hydrophilic interaction liquid chromatography electrospray tandem mass spectrometry. Biomedical Chromatography, 26, 1 5. 364 Gergov et al.