Mass spectrometry coupled to capillary GC constitutes the ideal method, in classi- cal electron ionization (EI) mode, for the analysis of compounds able to be gas chromatographed, i.e., sufficiently volatile and nonthermally degradable com- pounds. In recent years, increasing development of new modes of atmospheric pressure ionization processes has created greater opportunities for the coupling of high-performance liquid chromatography (HPLC) to MS. Nevertheless, this very promising technique is limited by the fact that the ionization process results in quasi–chemical ionization (CI) spectra, from which the structural information is poor compared with that obtained by the standard EI fragmentation process.
Mass spectrometry takes on particular importance in metabolism studies as re- flected by the large amount of literature published in the last decades. Gas chro- matography–mass spectrometry is an extremely useful tool in EI mode, which provides valuable structural information, or in CI mode, which allows one to obtain complement data on the pseudomolecular ion. Indeed, the value of MS lies in its capacity to identify compounds arising from parent drugs able to undergo a wide variety of enzymatic transformations leading to, for example, hydroxylation, N-dealkylation, reduction, or the introduction of a novel group. The GC–MS technique can be implemented in two ways: full-scan mode or selective ion moni- toring (SIM) mode.
3.1. Full-Scan Mode
The classical full-scan mode permits the identification and detection of parent molecules and metabolites by examination of the spectrum of each peak in the total ion chromatogram (TIC) with very good sensitivity in current instrumenta- tion. The examples given bellow illustrate the basic technique of detection and identification by GC–MS.
3.2. Basic Use of Full-Scan Mode
An example of the use of GC–MS for local anesthetic detection is discussed by Fox et al. [1]. It concerns articaine used in dentistry. The investigation demon- strates the power of this detection technique by proving the presence of the intact molecule in the urine 9 hr after the administration.
270 De´sage and Guitton
The most common biotransformations of volatile anesthetics involve oxida- tion (either cleavage or dehalogenation) except for halothane, which undergoes reduction as well as oxidation. Nevertheless, there is great variability in the extent of transformation of the drugs. Enflurane biotransformation was investigated in human urine using GC–MS [2]. In this study, the presence of difluoromethoxy- 2,2-difluoroacetic acid was demonstrated as its spectrum exhibited fragment ions at m/z 51, 67, 95, and 111. This metabolite was previously described as the result of the oxidative dechlorination at theβ-C atom.
Propofol provides a simple example of metabolite detection and identifica- tion. This drug and its metabolites are mainly eliminated in vivo in a conjugated form, i.e., as glucuronate or sulfate, requiring a deconjugation reaction prior to GC–MS analysis. The metabolic pathway is simple, since only three phase-I metabolites were described in different species. Identification of 2,6-diisopropyl- 1,4-quinol (hydroxylation at thepara position on the phenyl ring) was carried out with GC–MS in EI mode, leading to a mass spectrum with a molecular ion at m/z 194 and a major fragment at m/z 179 due to the loss of a methyl group [3]. Recently two other hydroxymetabolites [(2-ω-propanol)-6-isopropylphenol, and (2-ω-propanol)-6-isopropyl-1,4-quinol] were identified in humans by using full-scan mode [3a].
Many years ago, the identification of ketamine metabolites was performed in biological fluids from rats using GC–MS [4]. In this study, Kochhar [4] de- scribed two metabolites: norketamine (NK) and 5,6-dehydronorketamine (DHNK). No derivatization step was performed and GC was carried out on a glass column packed with 1% Carbowax 20 M. In EI mode, the main fragments were observed at m/z 180, 166, and 153 for ketamine, NK, and DHNK, respec- tively. Using CI with methane as reagent gas, the main ions were observed at m/z 238 [M⫹ H]⫹, m/z 207 [(M⫹ H)⫺ NH3]⫹, and m/z 205 [(M⫹ H)⫺ H2O]⫹for ketamine, NK, and DHNK, respectively.
Metabolic pathways of fentanyl (FTI) have been studied in vivo in several mediums, such as blood, saliva, and urine, and in vitro using isolated hepatocytes and microsomes from different species. Fentanyl is metabolized according to three reactions: oxidative N-dealkylation, aliphatic hydroxylation, and/or aro- matic hydroxylation, leading to 10 metabolites. An eleventh phase-I metabolite was described resulting from hydrolysis of the amide bond. Structures of FTI metabolites described in several studies are shown in Figure 1. Throughout this chapter we use FTI for numerous examples because the metabolism of this drug has been extensively studied with techniques based on several multidimensional detection strategies. First, GC–MS was employed to attempt to identify fentanyl metabolites in rat urine [5]. In this study, the authors concluded that fentanyl was metabolized mainly by a hydrolytic pathway leading to FTIII as main metabo- lite, as confirmed by the mass spectra.
Analysis of Anesthetics 271
Figure 1 Structures of fentanyl and its metabolites. (a) Introduction of a hydroxyl group into fentanyl at the propionyl moiety (FTVI), at the piperidine ring (FTVII), at the alpha position of phenetyl group (FTVIII), and at theparaposition of phenyl ring (FTIX). (Com- pounds resulting from dihydroxylation of FTI are not shown.) (b) Introduction of a hy- droxyl group into norfentanyl (FTII) at the propionyl moiety (FTIV) and at the piperidine ring (FTV). (c) Despropionylfentanyl (FTIII).
At last, surface ionization coupled to GC was reported by a Japanese re- search group (see Chap. 2) and applied to lidocaine.
3.3. Derivatization
In many cases, the detection of metabolites requires a derivatization step since these compounds are usually too polar. This process may be considered as a tedious step before analysis. However, it can also be considered as an optimiza-
272 De´sage and Guitton
tion strategy for MS. Indeed, this constraint may be turned into an advantage if a derivatization reaction is performed to create high-molecular-mass derivatives, which exhibit large molecular ions or fragment ions in the high-mass scale. Sev- eral studies from different anesthetic drugs support this point. N-dealkylated metabolites of fentanyl, alfentanil, and sufentanil contain a secondary nitrogen that can be derivatized. Valaer et al. [6] compared the derivatization with either pentafluoropropionic acid anhydride (PFPA) or with pentafluorobenzoyl chloride (PFB-Cl) for the screening of these metabolites in urine. In both cases, metabolite derivatives exhibited a high-mass fragment. However, with PFPA derivatives, lower background interferences were observed, particularly with cholesterol and steroids.
The necessity of a derivatization step for the determination of hydroxylated metabolites of MDZ on a GC column was previously indicated [7]. In this way, the determination of MDZ and its two hydroxyl metabolites, 1-hydroxymidazo- lam (1-OH-MDZ) and 4-hydroxymidazolam (4-OH-MDZ), was also performed with GC–MS after a derivatization with N-methyl-N-(tert-butyldimethylsilyl) trifluoroacetamide (MTBSTFA). The two derivatized metabolites exhibited the same three fragments in the EI mode at m/z 398, 400, and 440, respectively [8].
Furthermore, some specific derivatization of NH group, such as fluoroacy- lation permit the introduction of electrophilic groups in the molecule (COCF3, COC2F5), thus yielding a great sensitivity in negative-ion chemical ionization (NCI), which is generally more sensitive than EI. Rubio et al. [9] have described a method for the determination of MDZ, 1-OH-MDZ, and desmethylmidazolam (DMMDZ) using NCI. In this assay, the authors chose to analyze 1-OH-MDZ after derivatization with N-O-bis(trimethylsilyl)trifluoroacetamide (BSTFA), since MDZ carries a Cl atom. Mass spectra of MDZ and DMMDZ exhibit only pseudomolecular ions at m/z 325 and m/z 311, respectively. 1-OH-MDZ exhibits a mass spectrum with a pseudomolecular ion at m/z 413 in addition to a fragment ion at m/z 323 corresponding to [M-((CH3)3-SiOH)]⫹.
The formation of derivatized metabolites was very useful in some cases to avoid misinformation about a metabolic pathway. As an example, Adams et al.
[16] used pentafluoropropionyl (PFP) derivatives to avoid the thermal degrada- tion of hydroxylated metabolites during GC–MS. This degradation applied to DHNK indicated that it is an artifact of the analytical process rather than a major metabolite as previously described [4,10]. Another example is provided by the study of Van Rooy et al. [11], who detected and identified FTII and FTIII using GC–MS in the plasma of patients treated with fentanyl. Unlike Maruyama and Hosoya [5], Van Rooy et al. [11] introduced a derivative-forming step using acetic anhydride as reagent. Acetyl derivatives of the metabolites happened to be less volatile and no significant losses of the compound during evaporation were observed.
The type of derivatization chosen, however, induces a shift in retention time
Analysis of Anesthetics 273
of the peaks obtained by this process, the intensity of the shift being dependent on the characteristics of the derivatizing group, allowing adjustment of the chromato- graphic resolution. Some other examples below illustrate the use of the derivatiza- tion strategy for the analysis of some anesthetics.
Meuldermans et al. [12] presented a complete study on the metabolic path- ways of SUF and ALF from the excreta of rats and dogs. Among other techniques, they used GC–MS in EI and PCI modes for the characterization of major metabo- lites after derivatization by silylation or acylation. The same laboratory [13] has identified the two major metabolites of ALF in human urine. Lavrijsen et al. [14]
have studied the biotransformation of SUF in vitro in different species using GC–
MS in EI mode, but also with desorption CI and Townsend discharge ionization.
Several metabolites were identified unambiguously on the basis of their MS data, in comparison with the mass spectra of authentic reference compounds.
In some cases, GC–MS offers the possibility of identification by analogy with related structures. Stereoselective metabolism of enflurane to difluorometh- oxy-2,2-difluoroacetic acid (DFMDA) was studied in vitro using GC–MS [15].
The DFMDA was determined after derivatization step using ethanolamine. How- ever, authentic DFMDA was not available. Therefore, the authors used mass spectral data of a related ethanolamide derivative structure, i.e., 2-(N-2-chloro- 2,2-difluoro-acetamido)ethanol, to predict the mass spectrum of the derivatized DFMDA. In this study, the authors concluded that (R)-enflurane metabolism was more important than the (S)-enflurane metabolism.
Although derivatization was not necessary for analyzing the metabolites of propofol, silylation was achieved. In this case, shifts in retention time and in the m/z of the molecular ion of the major peak were observed. Two prominent ions at m/z 338 and m/z 323 were attributed to (di-TMS) and (M-15) of the 1,4-quinol derivative, respectively. Another unshifted peak was observed and identified as 2,6- diisopropyl-1,4-quinone (1,4-quinone) either in vivo or in vitro. Its mass spectrum exhibited a molecular ion at m/z 192 and a major fragment ion at m/z 149 due to the loss of a propyl group. It seems that 1,4-quinone was not a metabolite of propofol but corresponds to a chemical conversion of 1,4-quinol [3].
3.4. Multiple Strategy: Derivatization and High-Resolution Mass Spectrometry
An example of multistrategy analysis is provided by Adams et al. [16] who per- formed a thorough study on the biotransformation of ketamine in rat liver micro- somes. In this investigation, eight metabolites (products of alicyclic ring hydrox- ylation of ketamine) were described using low-resolution (m/∆m⫽ 600) mass spectra and high-resolution (m/∆m⫽10,000) mass spectra of derivatized metab- olites. High-resolution mass spectral data of ketamine, NK, and DHNK, used as pure compounds, allowed the investigators to establish the fragmentation path-
274 De´sage and Guitton
ways of ketamine metabolites under EI, and finally to obtain accurate knowledge on the sites of the compounds involved in the metabolic reactions. Ketamine metabolites were analyzed in this study as their pentafluoropropionyl (PFP) deriv- atives, while trifluoroacetyl derivatives were used for quantitative analysis.
3.5. Selective Ion Monitoring
Selective ion monitoring (SIM) detection in GC–MS allows the detection of al- ready known compounds at a lower concentration than the full-scan technique does. It combines enhanced sensitivity due to the use of a smaller number of measured ions, and enhanced specificity due to the choice of more specific high- mass ions.
3.6. Classical Use of Selective Ion Monitoring
A study of a local anesthetic by Tahraoui et al. [17] compared assays of bupiva- caine by HPLC and GC–MS, both using pentacaine as internal standard. The GC–MS was performed with SIM at m/z 140 and 154, for the molecule measured and the internal standard, respectively. These ions represent nearly the complete ionic current in the mass spectra of these molecules, thus assuring high sensitivity.
In this case, the simplicity of the mass spectrum of the bupivacaine and the inter- nal standard does not permit any modulation of the specificity of the detection, since only one ion is available. Moreover, this ion appears at a low m/z value, which does not provide a high specificity.
An assay for ketamine was described by Feng et al. [18]. The ion-trap detector, used by these authors, confines ions in the ionization chamber before expelling them, which induces ion–molecule reactions. The spectra obtained in this way are often different from conventional EI spectra. For the ketamine, for instance, which should show a molecular ion at m/z 237, an unexpected [M⫹ 1]⫹ ion at m/z 238 is observed. This artifact appears to result from CI, but in the present case it permits a specific detection of ketamine. The linearity ranges from 25 to 250 ng/ml and the yield of recovery was very good and in an accept- able concentration range.
Two assays for the quantification of propofol have been described. The first one was developed by Stetson et al. [19] using a silylation reaction and monitoring the [M–15]⫹ ion for quantification, and the molecular ion for the confirmation of each of the silylated derivatives. A liquid–liquid extraction was performed and thymol was used as internal standard. The method was applied in the range 1 to 3000 ng/ml. The second assay was performed in human whole blood after a simple extraction. The propofol and the internal standard thymol were quantified by means of the [M–15]⫹ion from the nonderivatized compound.
A linear response was obtained in the range of 10 to 10,000 ng/ml [3].
Analysis of Anesthetics 275
3.7. Indirect Measurement
As previously observed in relation to propofol, the observation of 1,4-quinone in the samples revealed a chemical transformation of 1,4-quinol. Therefore, for the assay of the hydroxy metabolite, the conversion of quinol into quinone, which can be achieved in basic medium, was applied to measure the metabolite, thereby avoiding all supplementary derivatization reactions. The metabolite is measured by monitoring the m/z-149 [M–43]⫹ion of quinone and the m/z-135 [M–15]⫹ ion of thymol used as the internal standard. This fast and appreciable method proved to be linear between 50 and 2500 ng/ml. It enables the estimation of propofol hydroxylation at the phenyl ring during in vitro experiments, using rat microsomes of animals pretreated by different inducers [20].
The acidic GHB, having a hydroxyl group in the gamma position, is suscep- tible to internal cyclization in acidic medium, resulting in butyrolactone. Ferrara et al. [21], after verification of the absence of this lactone in the plasma of patients treated with GHB, took advantage of this reaction to transform the acid into the lactone in order to avoid the necessity of derivatization. The formed lactone was measured by GC–MS in SIM mode using valerolactone as the internal standard.
The separation was performed on a nonpolar column and with split injection.
The method is applicable to concentrations in the range of 2 to 200 àg/ml in plasma, and 2 to 150àg/ml in urine.
3.8. Enhanced Specificity
Podkowik and Masur [22] used two ions each for midazolam and for clinazolam used as the internal standard. One of the two ions was used for quantification, while the second was used as a qualifying ion in order to increase the specificity of the detection. This method is linear between 10 and 500 ng/ml when a volume of only 40àl of plasma is used. The lower limit of quantification can be reduced to 0.25 ng/ml when a 500-àl sample is used.
3.9. Multiple Internal Standards
In another midazolam assay, Martens and Banditt [8] combined the advantages of a high-mass derivative and the choice of an internal standard as closely related to the target molecule as possible. The authors applied two distinct internal stan- dards, i.e., underivatized medazepam to quantify midazolam, and temazepam, which carries a hydroxyl group amenable to derivatization. The second internal standard was used for quantification of the two hydroxy metabolites, optimizing the conditions of quantification. This technique permits the detection of the minor metabolite 4-OH-MDZ in the range of 0.1 to 5 ng/ml, of 1-OH-MDZ from 1 to 50 ng/ml, and of midazolam itself from 2.5 to 125 ng/ml.
276 De´sage and Guitton
3.10. Use of Labeled Molecules
MS detection also permits the use of stable isotope labeling, a technique that can be used in several ways. In thetracer technique, metabolites are detected by the presence of isotope clusters after the in vivo absorption or in vitro incubation of an equimolar mixture of unlabeled and labeled molecules. The labeling can be done with one or more atoms of a heavy isotope, generally2H,13C,15N or18O, avoiding the use of radioactive tracers. Under these conditions, the mass spectrum of the parent drug exhibits ions of both the unlabeled and the labeled molecules, producing doublets in the isotope cluster pattern, the intensities of which corre- spond to the isotopic enrichment in the labeled material. Consequently, all metab- olites retaining the labeled atoms will also exhibit spectra containing the isotope cluster, proving unambiguously their metabolic origin. In this way, the analyst can focus only on the spectra showing this particularity. This so-called ‘‘isotope cluster,’’ ‘‘ion doublet,’’ or ‘‘twin-ion’’ technique greatly facilitates the identifi- cation of metabolites.
Goromaru et al. [23] used the stable isotope tracer technique in their study of the metabolism of fentanyl. An equimolar mixture of FTI and fentanyl labeled with deuterium on the aniline ring (Ft I-d5) was orally administered to rats at a dose of 20 mg/kg. Urinary extracts were derivatized with BSTFA and subse- quently analyzed by GC–MS. Five components were found to exhibit the isotope clusters in their mass spectra and were consequently recognized as metabolites of FTI. The mass spectrum of the main peak showed cluster ions at m/z 304:
309 [M]⫹, m/z 289:294 [M–15]⫹, m/z 275:280 [M–C2H5]⫹, m/z 247:252 [M–
COC2H5]⫹, m/z 231:236, and a single ion at m/z 155. The authors identified this compound as trimethylsilylated FTII, corresponding toN-dealkylation of FTI (see Fig. 1). Two other peaks exhibited ion clusters in their mass spectra with doublet m/z 377:382 [M]⫹. This value is 88 Da higher than that of FTII, indicating the introduction of an additional trimethylsilyl ether group (O-TMS) in the FTII struc- ture. The study of the fragmentation allows identification of the position of hy- droxylation. From the ion at m/z 117 [C2H4OTMS]⫹, it could be concluded that the substitution occurs on the propionyl moiety (FTIV). The absence of the m/z-117 ion would indicate that the hydroxylation occurs at the piperidine ring (FTV). Moreover, the presence of isotope clusters separated by 5 Da indicates that the aniline group is not metabolized. Similarly, two other metabolites were identified corresponding to the hydroxylation of FTI at the propionyl moiety (FTVI) and at the piperidine ring (FTVII). Due to these results, Goromaru et al.
[24] succeeded in the identification of those metabolites in humans. Another study from isolated hepatocytes of rat and guinea pig allowed the identification of sev- eral other metabolites of fentanyl (FTVIII, FTIX, FTX, FTXI), also with the use of the ion cluster technique [25].