Identifying and Quantifying Contaminants Contributing to Endogenous

Anal. Chem. 2010, 82, 7835–7841

Identifying and Quantifying Contaminants Contributing to Endogenous Analytes in Gas Chromatography/Mass Spectrometry Dimitrios Tsikas* Institute of Clinical Pharmacology, Hannover Medical School, Hannover, Germany Contaminants from various sources including medical devices, laboratory materials and the environment, and analytical apparatus may contribute to their endogenous congeners at different stages of the analytical process. Here, an approach is reported for the identification and quantification of contaminating analytes in biological fluids by stable-isotope dilution gas chromatography/mass spectrometry (GC/MS) and gas chromatography/tandem mass spectrometry (GC/MS/MS) methods. This approach is based on the analysis of different sample volumes and determination of the peak area ratio (PAR) of the endogenous analyte to the stable-isotope labeled analogue serving as the internal standard. The PAR is obtained by selected-ion monitoring or selected-reaction monitoring of appropriate ions. Generation of PAR values that correlate inversely with the sample volume subjected to analysis reveals the existence of contamination. The extent of contamination is obtained by plotting the PAR of endogenous analyte to internal standard versus the reciprocal of the sample volume analyzed. Examples are given for uncontaminated and contaminated endogenous analytes in biological samples, including nitrite and nitrate analyzed by GC/MS, and the fatty acid metabolites oleic acid oxide, oleic acid ethanol amide, and arachidonic acid ethanol amide analyzed by GC/MS/MS. Dependence of the PAR of endogenous analyte to its internal standard upon derivatization time reveals a unique kind of contamination that was identified in the GC/MS analysis of nitrate in plasma as pentafluorobenzyl ester. This kind of contamination occurs at the latest stage of GC/MS analysis and cannot be controlled by reference to the internal standard. Numerous substances that occur in the human body are also present in the human environment. For instance, nitrite and nitrate are widespread throughout the human body, they are constituents of drinking water and foods, and are present in the air as acids and nitrogen gases (NOx). Part of the nitrite and nitrate pool in blood, urine, tissue, saliva, and other body fluids as well as in the exhaled air are oxidative metabolites of nitric oxide (NO) produced from L-arginine by the catalytical action of NO * To whom correspondence should be addressed. Institute of Clinical Pharmacology Hannover Medical School, Carl-Neuberg-Strasse 1, 30625 Hannover, Germany. Phone: +49 511 532 3959. Fax: +49 511 532 2750. E-mail: [email protected]. 10.1021/ac101826t  2010 American Chemical Society Published on Web 08/26/2010

synthases present in virtually all types of cells.1 The concentration of substances in body fluids often varies between health and disease. Reference intervals and values allow quantitative assessment of the effects of nutrition, environment, physical exercise, lifestyle, and medication on the analyte of interest. Determination of the actual concentration of such substances in body fluids may be considerably compromised by contaminating congeners. Contaminating analytes may be abundantly present in medical devices such as syringes and vacutainers used in blood sampling and in containers used to collect urine or in other devices used for sampling such as microdialysis catheters. In addition, components of the analytical apparatus used for separation and detection, for instance gas chromatography/mass spectrometry (GC/MS) and gas chromatography/tandem mass spectrometry (GC/MS/MS) instruments, may also be a source of contaminants. The contribution of such contaminants to their congeners in body fluids may render it difficult to accurately quantify, define reference values and intervals, and compare different analytical methods to each other. Noteworthy consequences may be the overestimation of concentration, wrong diagnosis, improper therapy, and incorrect conclusions and decisions in basic and clinical research. An outstanding example (but certainly not the only example) of the importance of a thorough understanding of contamination in analytical chemistry is exemplified by the quantitation of prominent members of the L-arginine/NO pathway (and related pathways) such as nitrite and nitrate,2,3 S-nitrosothiols,3,4 3-nitrotyrosine,5 and dimethylamine.6,7 In consideration of the potentially strong impact of contamination on the analytical result, we developed a method for the identification and quantification of contaminants. In the present work, we exemplify this method for nitrite, nitrate, and fatty acid metabolites in human plasma and provide further examples for uncontaminated and contaminated endogenous analytes in biological samples. An unusual kind of contamination is reported for the GC/MS analysis of nitrate as its pentafluorobenzyl (PFB) derivative. The protocol of the GC/ MS method for nitrite and nitrate in human plasma was chosen (1) Boudko, D. Y. J. Chromatogr., B 2007, 851, 186–210. (2) Helmke, S. M.; Duncan, M. W. J. Chromatogr., B 2007, 851, 83–92. (3) MacArthur, P. H.; Shiva, S.; Gladwin, M. T. J. Chromatogr., B 2007, 851, 93–105. (4) Giustarini, D.; Milzani, A.; Dalle-Donne, I.; Rossi, R. J. Chromatogr., B 2007, 851, 124–139. (5) Ryberg, H.; Caidahl, K. J. Chromatogr., B 2007, 851, 160–171. (6) Chobanyan, K.; Mitschke, A.; Gutzki, F. M.; Stichtenoth, D. O.; Tsikas, D. J. Chromatogr., B 2007, 851, 240–249. (7) Tsikas, D. Anal. Biochem. 2008, 379, 139–163.

Analytical Chemistry, Vol. 82, No. 18, September 15, 2010

7835

The PAR of analyte + blank to the internal standard is given by formula 1. Considering the volume V of the sample subjected to analysis, formula 1 is converted to formulas 2a and 2b. Formula 2b describes a straight line with the regression equation y ) y0 + mx, with the y-axis intercept y0 being CA/CIS and the slope m being NB/CIS. Formulas 2a or 2b can be rearranged into formula 3. With the newly calculated NB value set into formula 3, the true analyte concentration CA in the sample can be calculated. Consequently, identification and quantification of contamination should be possible by subjecting different volumes of the biological sample to analysis under conditions in which all other parameters are held constant. In the case of no contamination, i.e., NB ) 0, a constant PAR with the value of CA/CIS would result. Formula 2b predicts that in the case of contamination by NB, plotting of PAR (y) against 1/V (x) would result in a straight line with the slope value NB/CIS, from which the total contaminant amount NB can be calculated. The formulas are

Figure 1. Schematic of the procedure for the quantitative determination of nitrite and nitrate in human plasma samples by GC/MS using their 15N-labeled analogues as internal standards after derivatization with pentafluorobenzyl bromide (PFB-Br) in aqueous acetone and extraction of pentafluorobenzyl derivatives by toluene.8 Potential sources of contaminating nitrite and nitrate throughout the whole analytical process are indicated. SIM, selected-ion monitoring. The + and - over the arrows indicate the presence and absence of contamination, respectively; the number of + signs and the size of the arrow illustrate the extent of contamination.

as an example to explain potential sources and the extent of contamination in GC/MS-based analyses of endogenous analytes using stable-isotope labeled internal standards (Figure 1). EXPERIMENTAL SECTION Proposal and Derivation of Formulas. The measured peak area of the internal standard (IS) is produced by injection (e.g., 1 µL of toluene extract) of the known amount NIS. The measured peak area of the unlabeled analyte is due to the sum of the unknown amount of the endogenous analyte, i.e., NA, and of the unknown amount of the contaminating or blank analyte, i.e., NB. The contaminating amount NB itself may be the sum of several different contaminating analyte amounts, i.e., due to contamination originating from the stable-isotope labeled internal standard (NBIS) and from other sources that may contribute contaminants at different stages of the analytical process (e.g., NB1, NB2, NB3): NB ) NBIS + NB1 + NB2 + NB3. It is reasonable to assume that NB is fairly constant under similar analytical conditions. (8) Tsikas, D. Anal. Chem. 2000, 72, 4064–4072.

7836


PAR ) (NA + NB)/NIS

(1)

PAR ) {(CAV + NB)/CIS}V

(2a)

PAR ) {(CA/CIS) + (NB/CIS)}(1/V)

(2b)

CA ) {(PAR)(CIS)} - (NB /V)

(3)

where NB is the total amount of the contaminant or blank (B) that contributes to the analyte (A), NA is the amount of the analyte A in the biological sample, NIS is the amount of the stable-isotope labeled IS in the biological sample, CA is the concentration of the analyte A in the biological sample, CIS is the concentration of the stable-isotope labeled IS in the biological sample, PAR is the peak area ratio of analyte (e.g., m/z 46 for [14N]nitrite and m/z 62 for [14N]nitrate) to stableisotope labeled IS (e.g., m/z 47 for [15N]nitrite and m/z 63 for [15N]nitrate) measured in the biological sample, and V is the volume of the biological sample subjected to analysis. Derivatization Procedure and GC/MS Analysis of Nitrite and Nitrate in Human Plasma. With the use of [15N]nitrate and [15N]nitrite as internal standards, nitrate and nitrite in human plasma were analyzed by GC/MS after derivatization with pentafluorobenzyl bromide (PFB-Br) as described elsewhere in this journal.8 All materials, solvents, and chemicals including the sodium salts of [15N]nitrate and [15N]nitrite were the same as described previously.8 The analytical procedure of this method is schematically shown in Figure 1. Typically, to 100 µL aliquots of plasma, 400 µL aliquots of acetone and 10 µL aliquots of PFB-Br were added and the samples were heated at 50 °C for 5 min (for nitrite) or 60 min (for nitrate and nitrite). When smaller volumes of plasma (e.g., 10, 20, 30, 50 µL) were used, a 4-fold volume of acetone was added, in order to maintain a constant acetone-to-plasma volume ratio of 4:1. After derivatization, acetone was removed and analytes were extracted with the same volume of toluene (300 µL). Analysis of quality control (QC) samples was performed on 100 µL aliquots of plasma samples as described.8 These data were used to determine the precision and accuracy of the method for unspiked

Table 1. Quality Control for Human Plasma Nitrite and Nitratea nitriteb QC sample QC1 QC1 - blanke QC2 QC2 - blank QC3 QC3 - blank

added (nM)c 0 2000 2613 4000

nitrateb

measured (nM)c

recovery (%)

precision (RSD, %)

added (µM)c

measured (µM)c

recovery (%)

precision (RSD, %)

750 ± 33 553 2810 ± 47

n.a.d n.a.d 103 103 105 105

4.4 n.a.d 1.7 n.a.d 1.2 n.a.d

0

73.1 ± 1.0 69.3 95.9 ± 2.1 92.1 118 ± 0.16 114.2

n.a.d n.a.d 114 114 112 112

1.4 n.a.d 2.2 n.a.d 0.14 n.a.d

4940 ± 59 4743

20 40

a All analyses were performed in duplicate using 100 µL plasma aliquots. b The derivatization time was 5 min for nitrite and 60 min for nitrate. The concentrations of [15N]nitrite and [15N]nitrate were 4 and 40 µM, respectively. d n.a., not applicable. e Contaminating (blank) concentrations of nitrite and nitrate were determined using formula 3 as 197 nM and 3.8 µM, respectively. c

plasma samples (QC1) and for plasma samples spiked with 2 µM nitrite and 20 µM nitrate (QC2) and with 4 µM nitrite and 40 µM nitrate (QC3). In all samples, [15N]nitrite and [15N]nitrate were used at a fixed concentration of 4 and 40 µM, respectively. Aliquots (1 µL) of toluene extracts were injected into the GC/ MS apparatus (model DSQ from ThermoFisher; Dreieich, Germany) in the splitless mode. Quantification was performed by selected-ion monitoring (SIM) of m/z 46 for [14N]nitrite, m/z 47 for [15N]nitrite, m/z 62 for [14N]nitrate, and m/z 63 for [15N]nitrate using a dwell time of 50 ms for each ion. The peak area of unlabeled and labeled nitrite and nitrate and PAR of unlabeled to labeled nitrite or nitrate were used for calculations. Derivatization Procedure and GC/MS/MS Analysis of Fatty Acid Metabolites in Human Plasma. GC/MS/MS analyses were performed using a model TSQ 7000 Finnigan MAT (San Jose, CA). cis-9,10-Epoxyoctadecanoic acid (cis-EODA) was analyzed in pooled plasma from a healthy volunteer by GC/MS/MS as described previously9 using cis-[9,10-2H2]-epoxyoctadecanoic acid (D2-cis-EODA) as internal standard at a final concentration in plasma of 50 nM. The procedure consisted of solid-phase extraction, derivatization with PFB-Br, and selected-reaction monitoring (SRM) of m/z 171 from m/z 297 and of m/z 172 from m/z 299. Analyses were performed in duplicate using plasma volumes of 100, 200, 300, 400, 500, 750, and 1000 µL. Arachidonic acid ethanol amide (AEA; anandamide) and oleic acid ethanol amide (OEA) were analyzed simultaneously in pooled plasma of a healthy volunteer by GC/MS/MS as described elsewhere.10 Internal standards were (5Z,8Z,11Z,14Z)-N-([1,1,2,2-2H4]2-hydroxyethyl)eicosa-5,8,11,14-tetraenamide (D4-AEA) for AEA and (9Z)-N-(2-hydroxyethyl)-[11,11-2H2]-9-octadecenamide (D2-OEA) for OEA at final plasma concentrations of 1 and 10 nM, respectively. The analytical procedure comprised of solvent extraction with toluene and derivatization first with pentafluorobenzoyl chloride and then with pentafluoropropionic anhydride. SRM of m/z 303 from m/z 540 and m/z 303 from m/z 544 was performed for AEA and D4-AEA and of m/z 282 from m/z 518 and m/z 284 from m/z 520 for OEA and D2OEA, respectively. Analyses were performed in duplicate using plasma volumes of 50, 100, 250, 500, and 1000 µL. All materials, solvents, and chemicals including the stable-isotope labeled analogues were the same as described previously.9,10 Safety Considerations. It should be noted that PFB-Br, pentafluorobenzoyl chloride, and pentafluoropropionic anhydride are corrosive. PFB-Br is an eye irritant. Inhalation and contact

with skin and eyes should be avoided. All work should be performed in a well-ventilated fume hood. RESULTS AND DISCUSSION Application of the Proposal to the Analysis of Nitrate and Nitrite in Human Plasma. In the newly performed investigations reported here, we considered the results of two recent studies indicating that NOx from laboratory air may be absorbed by biological samples including human plasma and contribute abundantly to endogenous nitrite and nitrate.11,12 Thus, in the present study appropriate measures were taken to minimize sample contamination by NOx gases from laboratory air as described recently.12 The results from the QC sample analysis for nitrite and nitrate in human plasma samples (100 µL) are summarized in Table 1. These data indicate that added nitrite and nitrate were measured accurately in the QC plasma samples. However, because of the ubiquity of nitrite and nitrate and the potential for contamination, these QC data do not ensure that the basal concentrations measured in the QC1 sample for nitrite (i.e., 750 ± 33 nM) and nitrate (i.e., 73.1 ± 1 µM) are the true concentrations of these anions in the plasma sample analyzed (see below). Nevertheless, it should be emphasized that in our basic and clinical research we use optimum conditions for nitrite and nitrate analysis as outlined previously.8 Figure 2 shows GC/MS chromatograms obtained for the simultaneous analysis of nitrite and nitrate in human plasma in the absence of added unlabeled nitrite and nitrate and derivatization for 60 min in plasma volumes of 10 (A) and 100 µL (B). The peak areas of unlabeled and labeled analytes were higher in the 100 µL plasma sample than in the 10 µL plasma sample. This is expectable because derivatization products were extracted using the same volume of toluene. However, Figure 2 indicates that different plasma concentrations of nitrite and nitrate are found depending upon the plasma volume analyzed. Thus, the concentration of nitrite is 2560 nM in the 10 µL aliquot of plasma, whereas (9) Tsikas, D.; Mitschke, A.; Gutzki, F. M.; Meyer, H. H.; Fro¨lich, J. C. J. Chromatogr., B 2004, 804, 403–412. (10) Zoerner, A. A.; Gutzki, F. M.; Suchy, M. T.; Beckmann, B.; Engeli, S.; Jordan, J.; Tsikas, D. J. Chromatogr., B 2009, 877, 2909–2923. (11) Fang, Y. I.; Hatori, Y.; Ohata, H.; Honda, K. Anal. Biochem. 2009, 393, 132–134. (12) Tsikas, D.; Mitschke, M.; Engeli, S.; Jordan, J. Anal. Biochem. 2010, 397, 126–128.


7837

Figure 2. GC/MS chromatograms from the simultaneous analysis of nitrite and nitrate in pooled human plasma derivatized with pentafluorobenzyl bromide for 60 min for sample volumes of 10 (A) and 100 µL (B). The internal standards [15N]nitrate and [15N]nitrite were added to the plasma samples at a final concentration of 40 and 4 µM, respectively. Numbers above the solid peaks indicate the peak area in arbitrary units. The open peaks in front of the nitrite peak (m/z 46 and m/z 47 traces) are due to nitrate.8

its concentration is only 752 nM in the 100 µL aliquot of plasma; the corresponding concentrations for nitrate were 116 and 79.6 µM (see Figure 2). Figure 3 shows the results from the simultaneous GC/MS analysis of nitrite and nitrate in different volumes of unspiked human plasma (i.e., 10, 20, 30, 50, and 100 µL) subjected to derivatization for 5 or 60 min. In accordance with previous findings,8 the peak area of the nitrate derivative is smaller than that of nitrite and depends on the derivatization time to a greater degree compared to nitrite (Figure 3A). Figure 3B shows that the PAR of m/z 46 to m/z 47 for nitrite and the PAR of m/z 62 to m/z 63 for nitrate are not constant but decrease with increasing plasma volume. Plotting the measured PAR versus the reciprocal of the plasma volume resulted in straight lines with positive slope values (Figure 3C). Figure 3B,C indicates that nitrite and nitrate analysis is complicated by contamination. From the slope values and the known concentrations of the internal standards used, the mean amounts NB of contaminating nitrite and nitrate are calculated to be 19.7 and 392 pmol, respectively. Thus, the contamination-corrected mean plasma concentrations are calculated by means of formula 3 as 553 nM for nitrite and 69.3 µM for nitrate (for V ) 100 µL). Table 1 shows that basal nitrite and nitrate concentrations, but not the method accuracy, for the 7838


spiked QC2 and QC3 plasma samples are affected by the contaminating nitrite and nitrate anions. Application of the Proposal to the Analysis of Fatty Acid Metabolites in Human Plasma. Figure 4A shows that the PAR of m/z 171 to m/z 172 for cis-EODA decreases with increasing plasma volume. Plotting the measured PAR versus the reciprocal of the plasma volume resulted in a straight line (Figure 4B). Thus, Figure 4 indicates that cis-EODA analysis in plasma is associated with contamination. The mean amount NB of contaminating cisEODA is calculated to be 620 fmol, which corresponds to a contaminating cis-EODA concentration of 0.62 nM with respect to a plasma volume of 1000 µL. Figure 5 shows that the PAR of m/z 281 to m/z 283 for OEA, but not the PAR of m/z 303 (from m/z 540) to m/z 303 (from m/z 544)10 for AEA, decreases with increasing plasma volume subjected to analysis and indicates that OEA but not AEA analysis in plasma is associated with contamination. The mean amount NB of contaminating OEA is calculated to be 100 fmol, which corresponds to a contaminating OEA concentration of 0.400 nM with respect to a plasma volume of 250 µL. Application of the Proposal to Further Analytical Examples. We have applied the proposed procedure to test the potential contribution of contamination to endogenous compounds,

Figure 4. GC/MS/MS analysis of cis-EODA in different volumes of human plasma spiked with D2-cis-EODA (50 nM) as the internal standard. Plot of the peak area ratio of m/z 171 for cis-EODA to m/z 172 for D2-cis-EODA versus the plasma volume V (A) or the reciprocal plasma volume 1/V (B). GC/MS/MS analysis was performed as described elsewhere.9

Figure 3. Simultaneous GC/MS analysis of nitrite and nitrate in human plasma derivatized with pentafluorobenzyl bromide for 5 or 60 min for different plasma volumes spiked with [15N]nitrate (40 µM) and [15N]nitrite (4 µM). (A) Plot of the peak area of [15N]nitrate and [15N]nitrite against the plasma volume for samples derivatized for 60 and 5 min. Plot of the peak area ratio of nitrite and nitrate for plasma samples derivatized for 60 min against the plasma volume V (B) or the reciprocal plasma volume 1/V (C).

for which varying sample volumes had been utilized in method validation in previous work from our group. These compounds include S-nitrosoalbumin in human plasma,13 the F2-isoprostane 15(S)-8-iso-PGF2R in human urine,14 creatinine in human urine,15 and nitrite in saliva.16 In these studies, stable-isotope labeled analogues and GC/MS13,14 or GC/MS/MS15 methods (13) Tsikas, D.; Sandmann, J.; Gutzki, F. M.; Stichtenoth, D. O.; Fro ¨lich, J. C. J. Chromatogr., B 1999, 726, 13–24. (14) Tsikas, D.; Schwedhelm, E.; Suchy, M. T.; Niemann, J.; Gutzki, F. M.; Erpenbeck, V. J.; Hohlfeld, J. M.; Surdacki, A.; Fro ¨lich, J. C. J. Chromatogr., B 2003, 794, 237–255.

Figure 5. Simultaneous GC/MS/MS analysis of anandamide (AEA) and oleic acid ethanol amide (OEA) in different volumes of plasma spiked with the internal standards D4-AEA (1 nM) and D2-OEA (10 nM). Plot of the peak area ratio of m/z 281 for OEA to m/z 283 for D2-OEA and of m/z 303 for AEA to m/z 303 for D4-AEA versus the plasma volume V. GC/MS/MS analysis was performed as described elsewhere.10

had been used. Expectedly, because S-nitrosoalbumin in human plasma is actually analyzed after its conversion to nitrite, contamination due to nitrite was identified for this S-nitrosothiol. No contamination was found for urinary 15(S)-8-iso-PGF2R (at a (15) Tsikas, D.; Wolf, A.; Mitschke, A.; Gutzki, F. M.; Will, W.; Bader, M. J. Chromatogr., B 2010, DOI: 10.1016/j.jchromb.2010.04.025. (16) Tsikas, D.; Bo ¨hmer, A.; Mitschke, A. Anal. Chem. 2010, 82, 5384–5390.


7839

Figure 6. Relationship between the peak area ratio (PAR) of m/z 63 to m/z 62 and incubation time at two derivatization temperatures (25 °C, 50 °C) obtained for the simultaneous GC/MS analysis of nitrite and nitrate in phosphate buffer, pH 7.4. The buffer was spiked with 50 µM each of [14N]nitrite, [15N]nitrite, [14N]nitrate, and [15N]nitrate. The PAR of m/z 47 to m/z 46 was close to the theoretical value of 1.0 (indicated by the horizontal dotted line) at both derivatization temperatures and for all incubation times (not shown here for the sake of simplicity). This Figure was constructed from the data of Figure 3B of previous work.8 The inset shows the reaction of nitrate with pentafluorobenzyl bromide in aqueous acetone to form the nitric ester. N depicts 15N as well as 14N.

concentration of 1.48 nM and using urine volumes of 100, 200, 500, 1000 µL) after immunoaffinity column extraction. Also, no contamination was detected for urinary creatinine (at a concentration of 8.47 mM and urine volumes of 1, 20, and 100 µL). Finally, no contamination was found for endogenous nitrite in human saliva (100, 250, 500, 750, and 1000 µL). Thus, the concentration of nitrite in the saliva sample was determined as 92.4 ± 1.52 µM (RSD, 1.65%).16 It is worth mentioning that contamination is a serious problem in other research areas such in clinical microbiology17,18 and genome sequencing19 and may thus affect clinical decisions. Interestingly, the rate of blood culture contamination was found to be higher with lower blood volumes17,18 and that blood culture contamination rate was inversely correlated with the blood volume (P < 0.001),17 resembling the findings of the present study. Identification of an Uncommon Type of Contamination in Nitrate Analysis. Previously,8 we observed that derivatization of [14N]nitrite, [15N]nitrite, [14N]nitrate, and [15N]nitrate (each at a concentration of 50 µM in phosphate buffer, pH 7.4) for various incubation times at 25 °C did not result in a constant PAR for m/z 63 to m/z 62, unlike the situation at 50 °C (Figure 6). The expected value for the PAR of m/z 63 to m/z 62 is about 1.0 (50 µM/50 µM) at 50 °C. Interestingly, the PAR of m/z 63 to m/z 62 increased with increasing incubation time and reached the expected value after about 120 min of derivatization. This is a surprising finding and seems to be contradictory to the general thought that stable-isotope labeled analogues correct for all changes occurring to their congeners during the whole analytical process. We therefore interpreted our previous findings on the (17) Bekeris, L.; Tworek, J. A.; Walsh, M. K.; Valenstein, P. N. Arch. Pathol. Lab. Med. 2005, 129, 1222–1225. (18) Gonsalves, W. I.; Cornish, N.; Moore, M.; Chen, A.; Varman, M. J. Clin. Microbiol. 2009, 47, 3482–3485. (19) Koboldt, D. C. Brief. Bioinf. 2010, DOI: 10.1093/bib/bbq016.

7840


basis of the unexpected apparent derivatization time-dependent increase of PAR of m/z 63 to m/z 62 for the derivatization temperature of 25 °C (Figure 6). Obviously, under these conditions the injected amount of the analyte is much larger than that of the contamination, i.e., NA . NB. The unexpected finding shown in Figure 6 for the derivatization temperature of 25 °C is interesting and suggests that additional yet unknown kinds of interference/ contamination may occur in the GC/MS analysis of nitrate when analyzed as the PFB derivative (i.e., PFB-ONO2). We therefore re-examined this issue in more detail. We may assume that in a mixture of [14N]nitrate (50 µM; 5 nmol/100 µL) and [15N]nitrate (50 µM; 5 nmol/100 µL) equimolar amounts of PFB-O14NO2 (NA) and PFB-O15NO2 (NIS) will be formed at any derivatization temperature and any incubation time, i.e., NA ) NIS, resulting in a PAR of about 1.0, if one considers the isotopic purity (>98 at % at 15N) of the commercially available sodium salt of [15N]nitrate (Sigma, Munich, Germany). However, Figure 6 shows that the PAR was only about 0.67 at 25 °C and 5 min. As in this particular case PAR ) NIS/(NA + NB) ) 0.67 and NA ) NIS, it follows that NB ) 1 /2NA. With increasing derivatization time, NA and NIS also increase at the same rate, whereas NB remains relatively constant. Finally, after 90 and 120 min of derivatization at 25 °C, NIS ) NA . NB, so that the PAR approaches the theoretically value of 1.0, i.e., 50 µM [15N]nitrate/50 µM [14N]nitrate. Assuming complete derivatization of [15N]nitrate and [14N]nitrate and quantitative extraction of PFB-O15NO2 and PFBO14NO2 into toluene and considering an injection volume of 1 µL in the splitless mode, it can be estimated that about 5 pmol each of PFB-O15NO2 and PFB-O14NO2 were injected from the toluene extract of the sample derivatized for 120 min at 25 °C. This calculation indicates that contaminating 14NOx from the GC/MS system contributes to PFB-O14NO2 and that the extent of the contribution is much less than 5 pmol per injection. Figure 6 suggests that at 25 °C and 5 min contaminating NOx species do contribute to [14N]nitrate to an extent comparable to the amount of PFB-ONO2 formed from nitrate present in the samples, i.e., NA ∼ NB. Because the internal standard [15N]nitrate apparently does not correct for this contribution, we have assumed that contamination occurs at a stage of the analytical process at which the internal standard is not present in the sample (Figure 1). As unreacted [14N]nitrate and [15N]nitrate anions remain in the aqueous phase of the reaction mixture, whereas their derivatives, i.e., PFB-O14NO2 and PFBO15NO2, and excess PFB-Br are quantitatively extracted by toluene, this kind of sample-contamination by NOx may originate with the separation of the PFB-O15NO2/PFB-O14NO2containing toluene phase from the biological sample (Figure 1). The large amounts of unreacted PFB-Br and other PFB derivatives and PFB-OH, the hydrolysis product of PFB-Br,8 should be able to react with NOx, for instance during the injection of the toluene extract in the hot injector (200 °C) of the gas chromatograph, to mainly form PFB-O14NO2 thus contributing to m/z 62. Recently, we identified an in-injector reaction of nitrous acid (HONO)16 and of a lipophilic S-nitrosothiol,20 i.e., a thionitrite, with solvents such as toluene and ethyl acetate. (20) Tsikas, D.; Dehnert, S.; Urban, K.; Surdacki, A.; Meyer, H. H. J. Chromatogr., B 2009, 877, 3442–3455.

Potential sources for this kind of contamination could be all the parts of the GC/MS system that are involved in sample injection (i.e., autosampler and injector, including the toluene extract-containing glass vials and septa), the solvent used for syringe clean up (toluene in the present case), and NOxcontaminated gases present within the injector port. CONCLUSIONS Commonly, analytical methods based on GC/MS consist of several steps, and contaminating analytes may contribute to their endogenous congeners in biological samples throughout the whole analytical approach, from sampling the biological matrix to separation and detection of the native and much more frequently of the derivatized analyte. Contamination at a late stage of the analytical process may be serious and in some cases not rectifiable by the use of stable-isotope labeled analogues of the target analytes. Quantitative analysis by varying the sample volume is a simple and adequate approach to identify sources of contamination, to determine the extent of contamination, and thus to calculate correctly the concentration of the target analyte in a biological sample. The approach proposed here should be

incorporated into method validation for endogenous compounds in biological fluids. Despite high sensitivity of GC/MS approaches, use of larger sample volumes than theoretically required may minimize contaminant contribution to endogenous substances. Especially for contaminations contributing at late stages of the analytical process, smaller volumes of organic solvents for sample extraction, for instance toluene in the case of nitrite and nitrate analysis as PFB derivatives, should be used. Similar considerations may also apply to other types of analysis including clinical microbiology and genomics. ACKNOWLEDGMENT The excellent laboratory assistance by Bibiana Beckmann, Anja Mitschke, and Maria-Theresia Suchy is gratefully acknowledged. Frank-Mathias Gutzki is thanked for performing GC/MS and GC/ MS/MS analyses in this study. The author sincerely thanks A. J. L. Copper for his constructive criticism. Received for review April 1, 2010. Accepted August 10, 2010. AC101826T


7841

Identifying and Quantifying Contaminants Contributing to Endogenous

Recommend Documents