Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 1017 DRUGS, COSMETICS, FORENSIC SCIENCES Chromatographic Methods for Analysis of Aminoglycoside Antibiotics ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 NINA ISOHERRANEN University of Helsinki, Department of Chemistry, Laboratory of Analytical Chemistry, PO Box 55, Helsinki 00014, Finland STEFAN SOBACK 1 Kimron Veterinary Institute, National Residue Control Laboratory, PO Box 12, Beit Dagan, Israel Aminoglycosides are antimicrobial agents used frequently in treatment of human and animal diseases caused by aerobic, gram-negative bacteria. Because of the toxicity of these compounds, considerable effort has been attributed to analysis of aminoglycoside content in drug preparations, in serum and urine specimen in therapeutic drug monitoring, and in edible animal tissues in residue control. The present review emphasizes the analytical problems associated with aminoglycoside analysis. Screening methods based on microbiological and immunological procedures were briefly discussed. Gas chromatography and especially high-performance liquid chromatography appeared the most widely used chemical methods for the analysis of these compounds. Due to lack of volatility, chromophore, and hydrophility of aminoglycosides, most methods applied derivatization for enhancement of their chromatographic characteristics. The applicability and advantages of the various derivatization procedures were discussed in detail. A wide variety of detection methods, including mass spectrometry have been used. Packed column separation was generally used for gas chromatographic separation. In liquid chromatography, reversed phase, ion pair, ion exchange, and normal phase separation has been employed. Mass spectrometry, as a detection method, was discussed in detail. Extraction procedures from body fluids and tissues were emphasized. The performance and the operational conditions of the methods were described and detailed information of the data was provided also in table format. minoglycosides have been used to treat infections caused by aerobic gram-negative and some gram-positive bacteria (1). They resemble each other in chemical structure, antimicrobial activity, pharmacokinetic A Received November 2, 1998. Accepted by JM March 24, 1999. 1 Author to whom correspondence should be addressed. characteristics, and toxicity (2). The first aminoglycoside, streptomycin, was identified in 1944 during the search for water-soluble and stable compounds active against gram-negative bacteria (1). Kanamycin was isolated in 1957, and at the time, it was active against streptomycin-resistant bacteria (1). Subsequently, bacteria resistant to kanamycin were isolated and the mechanism of resistance was resolved. Understanding the mechanism of resistance led to the development of semisynthetic aminoglycosides, of which more than 150 have been described (1). Amikacin, dibekacin, gentamicin, kanamycin, neomycin, netilmicin, paromomycin, sisomicin, streptomycin, dihydrostreptomycin, and tobramycin are currently in therapeutic use. In veterinary therapy, neomycin, gentamicin, kanamycin, streptomycin, and dihydrostreptomycin are the most common aminoglycosides (3). The antimicrobial activity of aminoglycosides is based on their ability to inhibit bacterial protein synthesis (4). Current primary use of these drugs in human medicine is for treatment of infections caused by aerobic gram-negative bacteria (4). The most important pathogens treated with aminoglycosides are pseudomonads, enterococci, coliforms, and salmonellae. Aminoglycosides are also used substantially in the treatment of tuberculosis. Aminoglycosides are rapidly absorbed from an injection site (2) but poorly absorbed after oral or rectal administration (4). They are not inactivated in the intestine and are eliminated quantitatively in feces (4). Systemically available aminoglycosides are excreted almost entirely as parent compound by glomerular filtration (2, 4). Aminoglycosides bind to tissue proteins and macromolecules via ionic bonds, but binding to plasma proteins is low (<25%). Aminoglycosides in tissues are usually found in low concentrations and unbound, except in the renal cortex where they tend to concentrate. Half-lives of aminoglycosides in plasma are 2–3 h but in tissues, bound aminoglycosides range from 30 to 700 h. The major concern in aminoglycoside therapy is their toxicity. All members of the group are both nephrotoxic and ototoxic (1, 2). The problems are pronounced in patients with impaired kidney function. Because of the narrow therapeutic range of aminoglycosides, therapeutic drug monitoring is essential. 1018 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 Figure 1. Structures of the therapeutically used aminoglycosides (1). ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 1019 Table 1. Organisms that produce specific aminoglycosides Aminoglycoside Organism(s) producing the compound Finder and year Gentamicin Micromonospora purpurea and M. echinospora Schering Plough, 1963 Kanamycin Streptomyces kanamyceticus Umezawa, 1957b Neomycin S. fradiae and S. albogriseolus Umezawa and Waksman, 1949 S. rimosus Parke Davis, 1959 M. inoyoensis Schering Plough, 1970 Streptomycin S. griseus Waksman and Schatz, 1944 Tobramycin S. tenebrarius Eli Lilly, 1967 S. griseus and S. humidus Takeda Chemicals, 1957 Paromomycin Sisomicin Dihydrostreptomycin a b From reference 1, unless indicated otherwise. From reference 4. Structures and Chemical Characteristics The word aminoglycoside is used to describe an amino-function-containing carbohydrate linked via a glycoside bond to an aminocyclitol (3). Therapeutically used aminoglycosides usually contain a 1,3- or 1,4-diaminocyclitol (1). The aminoglycosides are divided into 2 groups according to the aminocyclitol: streptamine and 2-deoxystreptamine. The deoxystreptamine group, the larger group (1, 3), is further divided into 2 subgroups according to the number and position of the substituents in the deoxystreptamine moiety. The best known subgroups are (1) neomycins and paromomycins, in which substituents are in adjacent positions (positions 4 and 5), and (2) gentamicins and kanamycins, in which the sub- stituents are in nonadjacent positions (positions 4 and 6) of the aminocyclitol moiety (3). The streptamine group contains streptomycin and dihydrostreptomycin (3). Structures of therapeutically used aminoglycosides are presented in Figure 1. Aminoglycosides are produced by Streptomyces or Micromonospora fungi. Semisynthetic aminoglycosides, such as dibekacin, amikacin, and netilmicin were developed to reduce toxicity and increase antimicrobial activity (1). Dibekacin (1) and amikacin (6) are synthesized from kanamycin B, and netilmicin, from sisomicin (7). Tobramycin can be synthesized from kanamycin or extracted from Streptomyces (1). The fungi producing the various aminoglycosides are presented in Table 1. Aminoglycosides Table 2. Physical and chemical characteristics of aminoglycosidesa Molecular formula Molecular weight Melting point, EC Optical rotation [α]D LD50 (mg/kg) in mice Amikacin C22H43N5O13 585.6 203–204 +99E 340–560b Aminoglycoside Dihydrostreptomycin C21H41N7O12 583.6 — Levorotatory 200 Dibekacin C18H37N5O8 451.54 — +132E 63–72 Gentamicin C1 C21H43N5O7 — 94–100 +158E 81 (iv)c Gentamicin C2 C20H41N5O7 — 107–124 +160E 183 (iv)c Gentamicin C1a C19H39N5O7 — — +165.8E — Kanamycin A C18H36N4O11 — — +146E 583 Kanamycin B C18H37N5O10 — 178–182 +114E 136 Kanamycin C C18H36N4O11 — — +126E — Neomycin C23H46N6O13 — — +80E (B) +120E (C) 36 Netilmicin C21H41N5O7 475.6 — +164E 40 Sisomicin C19H37N5O7 447.55 198–201 +189E — Streptomycin C21H39N7O12 581.58 — Levorotatory 200 Tobramycin C18H37N5O9 467.54 — +128E 118 a b c Data from references 12 and 101. Depends on pH. Intravenous. 1020 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 produced by Streptomyces have the suffix “mycin” and the ones produced by Micromonospora have the suffix “micins” (3). Aminoglycosides are water soluble and heat-, acid-, and base-stable polar compounds. Their solubility in methanol is limited, and they are practically insoluble in hydrophobic organic solvents. In water solutions, aminoglycosides are usually positively charged because of their amino groups. The number of amino groups varies from 4 in kanamycin to 6 in neomycin, and the pKa values of the amino groups vary between 7 and 8.8. The physical and chemical characteristics of the therapeutic aminoglycosides are presented in Table 2. Aminoglycosides are typically drug complexes formed by several closely related components. The neomycin complex is formed by 2 stereoisomers: neomycin B and C (8). Only neomycin B is antimicrobial. Paromoamine and paromomycins I and II occur as impurities of the neomycin complex (8). Commercially available neomycin products include 85–90% neomycin B (8). The gentamicin complex comprises at least 4 main components and several minor components (9). Separation of up to 7 different components of gentamicin has been reported (10), and proportions of gentamicin components in various products differ (11). Paromomycin is a complex of 2 stereoisomers: paromomycin I and II. Kanamycin is a mixture of 3 isomers, A, B, and C (1). The kanamycin components differ markedly in their toxicity and, therefore, commercial mixtures are required to contain at least 75% kanamycin A and not more than 5% kanamycin B (1). Screening of Aminoglycosides Aminoglycosides have been analyzed in tissues and urine by microbiological, radioenzymatic assay (REA), and radioimmunoassay (RIA) methods and by paper chromatography (12). These methods are still widely used but often lack quantitative or qualitative performance. Microbiological assays use methods based on agar diffusion of the drug and concentration-dependent growth inhibition (inhibition zone) of the test organism inoculated in the agar (12). The assays require 12–48 h incubation, after which inhibition of bacterial growth is measured. Several factors influence the accuracy of these methods (2). Incubation temperature, agar pH and ion concentration, depth of the agar on the plate, test strain, incubation time, and presence of other antibiotics in the sample can affect the assay (2). Additionally, different agar pHs need to be used for different aminoglycosides. Although microbiological methods are versatile, simple, and relatively cheap, they are inaccurate and subject to interferences caused by nonspecific inhibitors or other antimicrobial drugs (13, 14). Additionally, repeatability and reproducibility of results are generally poor (12), and the assays usually lack sensitivity below 2 µg/mL (2). RIA represents a major improvement over microbiological assays (2). RIA methods are sensitive (2, 12) and specific, but other aminoglycosides might cause interferences (12). Gentamicin, tobramycin, amikacin, netilmicin, and sisomicin can be determined by RIAs (12). Analysis using an RIA method requires complicated parameter optimization and specialization of the analyst (12). Selection and preparation of suitable specific antibodies is difficult and time consuming (12). In addition, handling of radioactive materials and radioactive waste and high cost are inhibitory factors (2, 12). Correlations between RIA and other methods are variable (2). REA offers similar advantages as do RIAs. The instability of enzymes and the presence of other antibiotics in the sample Table 3. GLC methods used for aminoglycoside analysis Derivatization Column phase, length (m) × id (mm) Temperature programa Carrier gas, flow rate (mL/min) Matrix Neomycin TMSDEA+ TrisilZ 3% OV-1 GasChromQ, 0.61 × 3 it 290EC He, 70 Standard FID 19 Neomycin TMSDEA+ TrisilZ 0.75% OV-1 GasChromQ, 1.83 × 3 it 290EC or tg 10EC /min 150–310EC He, 40 Standard FID 20 Kanamycin, paromomycin TMSDEA+ TrisilZ 0.75% OV-1 or 3% OV-1 GasChromQ, 1.83 × 3 it 290EC or it 300EC He, 70 Standard FID 21 Neomycin TMSDEA+ TrisilZ 3% OV-1 GasChromQ, 0.61–0.91 × 3 it 300EC He Standard FID 18 Kanamycin , gentamicin, tobramycin TMSI/HFBI 3% OV-101, 2 × 3 it 272EC N2, 60 Plasma ECD 22 DNT, tobramycin, netilmicin TMSI/HFBI 3% OV-101 ChromosorbW, 2 × 3 it 272EC N2, 60 Plasma ECD 23 Amikacin, paromomycin TMSI/HFBI 1% OV-17, 2 × 3 it 277EC N2, 80 Plasma ECD 23 Kanamycin HFBI/TMSI 4% OV-101, 1.5 × 3 it 270EC — Muscle ECD 24 Analyzed compounds b a b it = isothermal; tg = thermal gradient. Internal standard. Detector Ref. ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 1021 can cause false results (2, 12). An REA method has been developed for kanamycin, tobramycin, amikacin, and sisomicin (12). However, occasionally poor correlation with results obtained with other methods has been observed (12). The 2 most common assays for aminoglycosides are the homogenous enzyme immunoassay and the fluorescence polarization immunoassay (FPIA; 2). These methods have similar sensitivity and precision as RIAs but are less expensive and do not require handling of radioactive material (2). Chromatographic Analysis Chromatographic methods for aminoglycoside analysis are needed for simultaneous qualitative and quantitative determinations. However, separations are difficult to achieve because of the structural similarity of aminoglycosides (15, 16). Paper chromatography of gentamicin components was the first reported chromatographic method for aminoglycosides (16). Gas Chromatography Gas chromatography (GC), with its high separation capacity and efficiency, is a popular technique for volatile, heat-stable compounds. However, direct analysis of intact aminoglycosides by GC is impossible because of the hydrophilic, basic, and nonvolatile nature of these molecules (17). By derivatizing the amino and hydroxyl groups, volatile derivatives can be produced (18–24). The gas–liquid chromatographic (GLC) methods for aminoglycosides are presented in Table 3. Trimethyl silyldiethyl amine (TMSDEA), a reagent that silylates both amino and hydroxyl groups, has been used as a derivatizing agent (18–21). The products are moisture sensitive and, therefore, unstable (18–21). Consequently, results are nonlinear and poorly repeatable, and the derivatization yield is low. Freeze drying of samples prior to derivatization has been used to eliminate variations in sample moisture content and solubility (18). Sealed sample vials, removal of metal parts from the chromatographic system, and on-column injection have been tried to improve repeatability and quantitation (18). Nonetheless, linearity and derivatization yield remain poor. It is important to know the ratio of drug components or stereoisomers that differ in their pharmacological characteristics in a pharmaceutical product. Stereoisomers of neomycin B and C (18, 20) and paromomycin I and II (19, 21) have been separated by GLC as TMSDEA–silylated derivatives. Stereoselective separation using a nonchiral column has been suggested to result from stronger retention of the equatorial than the axial epimer (21). On the basis of this assumption, it has been suggested that paromomycin I must elute before paromomycin II and neomycin B before C. The kanamycin components A, B, and C have been separated as their trimethylsilyl (TMS) derivatives (21). The TMS derivatives of neomycin (19, 20), kanamycin (21), and paromomycin (21) also have been identified by mass spectrometry (MS). Derivatization results in silylation of all amino and hydroxyl groups. A 2-step derivatization using trimethyl silylimidazole (TMSI) for silylation of hydroxyl groups and heptafluorobutyric imidazole (HFBI) for heptafluorobutyrylation of amino groups has been reported (22–24). Using this technique, Mayhew and Gorbach (23) analyzed concentrations of gentamicin, tobramycin, netilmicin, amikacin, and paromomycin in serum. For each aminoglycoside, another one was used as an internal standard. Results are acceptable in accuracy and precision, and the method is linear for the concentration range investigated. TMS– heptafluorobutyryl (HFB) derivatives are more stable than TMSDEA derivatives, but protection of the samples from moisture still is crucial. Gentamicin elutes from the column as 2 peaks: the first is gentamicin C1, and the second, as a combination of gentamicins C1a and C2. All the other aminoglycosides analyzed produce 1 peak. Mayhew and Gorbach (22) also reported an improved GLC method for gentamicin with kanamycin as internal standard. The accuracy and repeatability of the method are improved, and the correlation coefficient (r) for the calibration curve is 0.979 (22). GLC analysis of kanamycin in muscle tissue has been reported (24). In this study, the 2-step TMSI–HFBI derivatization described by Mayhew and Gorbach (23) is used in reversed order. This procedure produces better results than the original TMSI–HFBI derivatization. In comparison, the 2-step derivatization with TMSI and HFBI appears superior to silylation with TMSDEA, the main advantage being the higher derivatization yield and the more stable derivatives. The 2-step derivatization method is also applicable to larger number of aminoglycosides and more internal standards are available compared with the TMSDEAbased method. Unlike in TMSDEA methods, individual aminoglycosides do not disturb the assays of other aminoglycosides in the TMSI–HFBI methods. The HFBI derivatization allows the use of an electron capture detector (ECD), providing increased selectivity and sensitivity of the method compared with methods using a flame ionization detector (FID). The main weakness of GC methods for aminoglycosides is the lack of capillary column applications. Liquid Chromatography As water-soluble, polar, and relatively large compounds, aminoglycosides typically would be analyzed by liquid chromatography (LC; 1). Most high-performance liquid chromatographic (HPLC) methods concern gentamicin analysis (25), but because of the chemical similarity within the group, these methods are often applicable to other aminoglycosides. The most significant characteristics of aminoglycosides relevant to HPLC analysis are the lack of a chromophore and the high hydrophilicity. To improve detectability and separation, the aminoglycosides are usually derivatized prior to or after chromatographic separation. Methods used to separate aminoglycosides by HPLC are presented in Table 4. Reversed-Phase Chromatography Because of their polar and ionic nature, aminoglycosides usually are not partitioned on reversed-phase columns (15, 26). 1022 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 Table 4. HPLC methods used for separation of aminoglycosides Column phase, id (mm) × length (cm) × particle size (µm) Flow rate of mobile phase, L/min Mobile phasea Derivatizationb Detector, wavelengths (nm)c Aminoglycosidesd Ref. LiChrospher 100 RP18, 4.0 × 12.5 × 5 0.2 MeOH–H2O, 55 + 45, 0.05 cs-EDTA OPA pc fl, ex 340, em 440 gnt (1), neo, kan, ami, dhs 5 PLRP-S 1000Å, 4.6 × 25 × 8 1.0 70 g/L Na2SO4 1.4 g/L ocs fos pH 3 — PED neo 8 Cellulose phosphate PII, 0.9 × 15 × — 0.2 2.1M NaCl — Polarography gnt (4) 10 Lichrosphere C8, 3.0 × 25 × 7 1.5 0.015M pns aca OPA pc fl, ex 350, em 450 gnt, ntl, neo, sis, kan 13 µPartisil, 3.9 × 30 × — 1.0 H2O–MeOH–DEA, 60 + 40 + 0.5 OPA pc RI gnt, ntl, neo, kan, sis 13 Ultrasphere C18, 4.6 × 25 × 5 1.0 MeOH–H2O HFBA — RI str, kan, sis, gnt 15 ODS II, 4.6 × 10 × 3 1.0 MeOH–H2O TFA — MS TSP gnt (4) 16 Hypersil C18, 5 × 12.5 × 5 or Spherisorb ODS, 5 × 25 × 5 0.25–3.0 MeOH–H2O, 70:30 aca, .02M hps OPA UV, 330 gnt (4) 63 Zorbax 18, 4.6 × 15 × 5 0.8 25 mM tls, perchloric acid pH 2.0 OPA pc fl, ex 360, em — sis, ntl, neo, par 14 Spherisorb ODS-2, 2.0 × 10 × 5 0.2 ACN–H2O, 5 + 95, 20 mM PFPA — MS/MS str, dhs, neo, gnt 17 µBondapak C18, 3.9 × 30 × 5 2.0 MeOH–H2O, 3 + 97, 0.1% aca 0.02M pns 0.2M Na2SO4 OPA pc fl, ex 340, em 418 gnt, sis, ntl 50 Supelcosil LC-8-DB, 4.6 × 15 × 5 — 1.5% MeOH, 0.01M pns, 0.056M Na2SO4, 0.007M aca OPA pc fl, ex 340, em 455 neo 51 Supelcosil LC-8-DB, 4.6 × 15 × 5 — 0.011M pns 0.007M aca, 1.5% MeOH OPA pc fl, ex 340, em 455 neo 25 PhaseSep Sperisorb ODS2, 4.6 × 15 × 5 1.5 MeOH–H2O, 82 + 18 0.01M pns, 0.0056M Na2SO4, 0.1% aca OPA pc fl, ex 340, em 420 gnt (4) 52 Nucleosil C18, 4.6 × 12.5 × 5 1.0 MeOH–H2O, 80 + 20 aca OPA on line fl nea, ami, dib, gnt (4), sis, tob, ntl 26 LiChrosorb RP18, 4.6 × 25 × 10 4.0 ACN–DCM–H2O–MeOH, 80 +10 + 8 + 2 BSC UV, 230 nm gnt (1), ntl 27 µBondapak C18, 4.0 × 30 × — 1.0 ACN–H2O, 95 + 5 DansCl fl, ex 220, em 470 gnt (2) 28 µBondapak C18, 3.9 × 30 × 10 3 ACN–tris, 70 + 30 FDNB UV, 365 gnt (3) 29 µBondapak C18, 3.9 × 30 × 10 1.5 ACN–tris (pH 7), 70 + 30 FDNB UV, 365 gnt (3), sis, tob 30 LiChrosorb RP8, 3.0 × 25 × 5 0.5 ACN–tris (pH 3), 70 + 30 OPA fl, ex 340, em 418 gnt (3) 31 µBondapak C18, 4.0 × 30 × — — MeOH–H2O, 79 + 21 2g/L EDTA OPA fl, ex 360, em 430 gnt (3) 32 µBondapak C18, — — MeOH–H2O, 79 + 21 2g/L EDTA OPA fl gnt (3) 33 µBondapak C18, — — MeOH–H2O, 75 + 25 tris, TEA, H2SO4 pH 7 OPA fl tob, ntl 33 Hypersil ODS, 4.6 × 20 × 3 1.0 ACN–H2O, 90 + 10 FMOC–Cl fl, ex 260, em 315 gnt (3) 34 Partisil5 ODS-3, 4.6 × 10 × 5 1.5 ACN–H2O, 40 + 60 DNBC UV, 254 ami, tob, kan 35 Partisil5 ODS-3, 4.6 × 25 × 5 1.5 MeOH–H2O, 70 + 30 DNBC UV, 254 gnt (4) 35 µBondapak C18, 3.9 × 30 × 10 3 ACN–H2O–aca, 60 + 39 + 1 FDNB UV, 365 tob, gnt 36 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 1023 Table 4. (continued) Flow rate of mobile phase, L/min Mobile phasea Derivatizationb Detector, wavelengths (nm)c Aminoglycosidesd Ref. Merck RP18, 3.9 × 30 × 10 1.3 ACN–H2O–aca, 70 + 29 + 1 FDNB UV, 365 tob 37 LiChrosorb RP18, 3.9 × 30 × 10 2.5 ACN–H2O–aca, 47 + 52 + 1 FDNB UV, 365 ami, kan 38 Ultrasphere ODS C18, 4.6 × 25 × 5 — ACN–H2O, 68 + 32 FDNB UV, 360 ami 39 µBondapak, 3.9 × 30 × 10 2.0 MeOH–fos (pH 6), 78 + 22 TEA OPA fl, ex 260, em 418 gnt (3) 71 LiChrosorb RP8, 4.6 × 25 × 5 1.5 0.02M fos (pH 7.5), KOH, ACN, MeOH TNBS UV, 350 gnt, tob, sis, ami, kan 40 Ultrasphere octyl C8, 4.6 × 25 × 5 2.0 ACN–fos, 52 + 48 TNBS UV, 340 ami, kan 41 Ultrasphere octyl, 4.6 × 25 × — 3.0 ACN–fos (pH 3.5), 70 + 30 TNBS UV, 340 tob, sis 42 Column phase, id (mm) × length (cm) × particle size (µm) µBondapak C18, — 1.0 ACN–H2O, 95 + 5 DansCl fl, ex 220, em — ntl 43 µBondapak C18, 4.0 × 30 × — 1.0 ACN–H2O–MeOH, 5 + 30 + 65 2g/L EDTA OPA fl, ex 350, em 450 ami 44 Hisep, 4.0 × 15 × 5 1.7 MeOH–H2O TCA, EDTA OPA fl, ex 340, em 418 neo 45 Amine bonded, 4.7 × 25 × 5 0.7 ACN–H2O–aca, 70 + 30 + 0.1 — MS moving belt kan, tob, nea 46 Hitachi gel ODS, 4.0 × 15 × 5 1.0 100 mM NH4 acetate — MS APCI kan 47 Hitachi gel ODS, 4.0 × 15 × 5 — MeOH–H2O NH4 acetate — MS APCI kan, gnt (3) 48 Ultraspher C18, 4.6 × 25 × 5 1.0 MeOH–H2O different ion pairs — MS tai RI gnt, sis, kan, str, neo 69 Spherisorb ODS-2, 2.0 × 10 × 5 0.2 ACN–H2O, 8–16 + 92–84 40 mM HFBA — PAD or MS str, dhs 49 Spherisorb ODS-2, 4.6 × 15 × 5 — 0.0056M Na2SO4, 0.007M aca 0.011M pns, 18.5% MeOH OPA pc fl, ex 340, em 455 neo 53 Supelcosil LC-8-DB, 4.6 × 15 × 5 0.9 0.01M pns, 0.056M Na2SO4, 0.007M aca, 1.5% MeOH OPA pc fl, ex 340, em 455 neo, par, str, dhs 54 µBondapak C18, 3.9 × 30 × 10 1.0 0.05M pns, 0.2M Na2SO4, 0.1% aca — Electrochemical gnt (3) 55 Nucleosil C18, 4.6 × 15 × 10 1.2 0.02M pns aca OPA pc fl, ex 340, em 418 ami, tob 56 µBondapak C18, 3.9 × 30 × — 2.0 0.2M Na2SO4, 0.02M pns, 17.4M aca, 3% MeOH OPA pc fl, ex 340, em 418 gnt, ami, tob 57 Ultrasphere ODS, 4.6 × 25 × 5 1.5 MeOH–H2O, 80 + 20 5% aca, 5 g/L hxs OPA UV, 330 gnt (4) 59 LiChrosorb RP8, 3.2 × 25 × 10 1.7 MeOH–H2O–aca, 9 + 90 + 1, 5 g/L hxs OPA fl, ex 340, em 418 gnt (3) 59 Lichrosorb RP18, 4.0 × 25 × 5 1.0 ACN–fos (pH 3), 8 + 92 3.76 g/L hxs — UV, 195 str, dhs 61 µBondapak C18, 3.9 × 30 × 5 1.0 ACN–H2O, 8 + 92, 0.02M hxs, 0.025M fos — UV, 195 str, dhs 58 Supelcosil LC8-DB, 4.6 × 25 × 5 0.5 ACN–H2O, 17 + 83 hxs NQSA fl, ex 347, em 418 str, dhs 89 Supelcosil LC8-DB, 4.6 × 25 × 5 0.5 ACN–H2O, 17 + 83 hxs NQSA fl, ex 365, em 418 str, dhs 60 1024 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 Table 4. (continued) Flow rate of mobile phase, L/min Mobile phasea Derivatizationb Detector, wavelengths (nm)c Aminoglycosidesd Ref. Ultrasphere ODS, 4.6 × 15 × — 1.5 MeOH–H2O, 70 + 30, aca, hps OPA UV, 330 gnt (4) 64 Zorbax SB C18, 2.1 × 15 × — 0.5 MeOH–H2O, 70 + 30, aca, hps OPA UV, 330 gnt (4) 65 ODS Hypersil, 5.0 × 10 × 5 1.0 MeOH–H2O–aca, 70 + 25 + 5, 5 g/L hps OPA UV, 330 gnt (4) 66 Nucleosil C18, 4.0 × 20 × 5 0.9 MeOH–H2O, 80 + 16, aca, hps OPA fl, ex 340, em 450 sis, ntl 86 Ultremex C18, 4.6 × 25 × 5 1.6 MeOH–H2O, 70 + 30 aca, hps OPA fl, ex 340, em 418 gnt (3) 67 Hypersil ODS, 4.6 × 10 × 5 1.0 MeOH–H2O, 72 + 28 hps OPA UV, 350 gnt 11 Supelcosil LC-8, 4.6 × 25 × 5 1.8 MeOH–H2O 10–60% MeOH, aca, hps OPA fl, ex 340 em 448 gnt (3) 68 Radicalpak C18, 8 × 10 × 10 1.5 ACN–H2O, 80 + 20, aca, ocs-eds NQS fl, ex 351, em 420 str 88 µBondapak C18, 3.9 × 30 × 10 1.5 ACN–H2O, 15 + 85 aca, 0.1M eds-ocs OPA pc fl, ex 365, em 440 gnt (3) 62 PRP-1 polymer, 4.6 × 25 × 10 — ACN–H2O, 7 + 1, different ion pairs — RI tob, kan, str, neo 70 Dowex AG 1×2, 0.9 × 15 × — 4.5 H2O — — neo 74 Carbopak PA1, — 1.0 0.5–50M NaOH — PAD tob 75 MPIC–NSI polystyrene, 4.0 × 25 × — 0.6 0.25M NaOH — PAD tob, kan 76 Partisil SCX, 3.6 × 25 × 10 2.0 ACN–fos, 70 + 30 Fluorescamine fl, ex 275, em 418 gnt 77 Zipak SCX, 2.1 × 100/50 × 37-44 0.8 0.01M EDTA (pH 9.5) OPA pc/ fluorescamine fl and RI kan (3) 78 Amberlite IRC 50, 13.9 × 29 × — — For elution, 0.5M Na2SO4 Dihydrolutidine fl, ex 421, em 488 tob, neo, sis, kan, ami 79 Zorbax Sil, 4.6 × 25 × — — Dichloroethane–heptane–MeOH– H2O–DEA, 79 + 15 + 5.5 + 3.6 + 1.5 FDNB UV, 365 neo (3) 80 P-EHS5 Silica, 4.6 × 12.5 × 5 1.5 Chl–MeOH–aca, 95 + 2 + 3 NSCl UV, 254 neo, gnt, kan 81 LiChrosorb SI-100, 4.6 × 25 × 5 — Chl–THF–H2O, 60 + 39 + 1 FDNB UV, 254 or 350 neo, gnt, kan, par 82 Nucleosil C18, 4.0 × 20 × 5 0.6 MeOH–H2O, 60 + 40 EDTA OPA fl, ex 340, em 450 sis 87 Column phase, id (mm) × length (cm) × particle size (µm) a b c d OCS = octane sulfonate; fos = phosphate buffer; pns = pentane sulfonate; aca = acetic acid; DEA = diethylamine; HFBA = heptafluorobutyric acid; TFA = trifluoroacetic acid; hps = heptane sulfonate; t/s = toluene sulfonate; ACN = acetonitrile; PFPA = pentafluoropropionic acid; DCM = methylene chloride; TEA = triethylamine; TCA = trichloroacetic acid; hxs = hexane sulfonate; eds = ethane disulfonate; chl = chloroform. OPA = o-phthalaldehyde; pc = postcolumn derivatization; NQSA = $-naphthoquinone-4-sulfonic acid; NQS = $-naphthoquinone-4-sulfonate; FDNB = 1-fluoro-2,4-dinitrobenzene; NSCl = naphthalene sulfonyl chloride; BSC = benzenesulfonyl chloride; DansCl = dansyl chloride; FMOC–Cl = 9-fluorenylmethoxycarbonyl chloride; DNBC = dinitrobenzoylchloride; TNBS = 2,4,6-trinitrobenzene sulfonic acid. fl = fluorescence detector; ex = excitation wavelength; em = emission wavelength; PED = pulsed electrochemical detector; RI = refractive index detector; MS = mass spectrometry; TSP = thermospray; UV = ultraviolet; MS/MS = tandem mass spectrometry; APCI = atmospheric pressure chemical ionization; PAD = pulsed amperometric detector. Numbers in parentheses indicate the number of components; gnt = gentamicin; neo = neomycin; kan = kanamycin; ami = amikacin; dhs = dihydrostreptomycin; ntl = netilmicin; sis = sisomicin; str = streptomycin; par = paromomycin; tob = tobramycin; dib = dibekacin; nea = neamine. ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 1025 Partition to C18 columns can be improved by using acetate buffer in the mobile phase. The acetate ion acts as a counterion, forming ion pairs with aminoglycosides (26). Usually aminoglycosides are derivatized with a nonpolar group prior to reversed-phase analysis to improve the partition to column and the separation characteristics. The most commonly used derivatization reagents are o-phthalaldehyde (OPA) and 1-fluoro-2,4-dinitrobenzene (FDNB). The separation columns usually contain C8 or C18 materials, and mobile phases consist of acidic buffers with methanol and/or acetonitrile. The reversed-phase method reported by Essers in 1984 (26) apparently remains the method able to simultaneously separate the largest number of aminoglycosides. On-line precolumn derivatization of the aminoglycosides is used. The mobile phase consists of methanol–acetic acid (80 + 20). Dibekacin, sisomicin, tobramycin, and 4 components of gentamicin can be separated in one chromatographic run. The gentamicin components elute in the order C1, C1a, C2a, and C2, with the stereoisomers C2 and C2a separated from each other. By reducing the proportion of methanol in the mobile phase, the more polar aminoglycosides amikacin and neamine can be separated. Netilmicin requires an incresed proportion of methanol. For each aminoglycoside, one of the others was used as an internal standard. The separation of gentamicin components is of interest in the quality control of pharmaceutical preparations, pharmacokinetics, and toxicological studies. Gentamicin components can separate to 1–5 peaks in reversed-phase columns depending on derivatization and separation conditions (27–35). When gentamicin is derivatized with benzene– sulfonyl chloride (BSC), the derivatives elute as a single peak (27). After dansyl chloride (28) or FDNB (29, 30) derivatization, gentamicin elutes as 2 separate peaks. The elution orders of dansyl and dinitrophenyl derivatives are similar but separation of components differ. Retention of gentamicin components is assumed to increase with the number of methyl groups in the gentamicin structure (30). As dansyl derivatives, gentamicins C1a and C2 elute together, followed by gentamicin C1 as a separate peak (28). As 2,4-dinitrophenyl derivatives, gentamicin C1a elutes before the coeluting C1 and C2 components (29, 30). Both dansyl chloride and FDNB derivatize all amino groups in gentamicin and, therefore, the relative functionalities do not change as a result of the derivatization reagent used. In the separation of 2,4-dinitrophenyl derivatives, the lack of a methyl group in gentamicin C1a appears to be the major factor affecting separation. The mechanism of separation of dansyl derivatives has not been discussed in the literature. It can be assumed that the separation of the dansylated gentamicins is based on the functionality of the amine on C-6′. After derivatization, the components (C1a and C2) that have similar amino groups, HNR2, elute together before the gentamicin component C1 which contains an NR3 amino group. Three gentamicin components are separated as OPA derivatives on C8 or C18 columns by use of acetonitrile–tris (31) or methanol–ethylenediamine tetraacetic acid (EDTA)–water (32, 33) mobile phases, correspondingly. In these studies, identification of the different gentamicin components is based on standards of individual components but the structures of the derivatives are not discussed. Strangely, the elution order of the gentamicin components is C1a, C2, and C1 with a C8 column but C1, C1a, and C2 with a C18 column. The retention mechanism is not discussed. The retention of gentamicin C1 relative to other gentamicin components is of special interest because as an OPA derivative, it contains one derivatizable amine less than the other components. The derivatized C1 component is much smaller and has one basic secondary amine more than the other components. The retention of the gentamicin C1 component is more sensitive to changes in mobile phase pH than the other components, and its retention increases rapidly when the pH of the mobile phase increases (31). This or derivatization differences may explain why gentamicin C1 elutes as the last component from a C8 column and as the first component from a C18 column. Addition of EDTA to the mobile phase yields sharper peaks (32). The gentamicin 9-fluorenyl methoxycarbonyl chloride (FMOC–Cl) derivatives are separated as 4 separate peaks on a C18 column with an acetonitrile–water mobile phase (34). The elution order is C1a, C2, x, and C1. The component x is believed to be the stereoisomer of gentamicin C2 and gentamicin C2a. The reported elution order corresponds to the increasing methylation of gentamicin components. On the basis of the elution order of the components, it is assumed that all the primary and secondary amino functions of the gentamicins are derivatized. Otherwise, the component C1 is expected to elute first (34). After 3,5-dinitrobenzoyl chloride (DNBCl) derivatization, the gentamicin components are separated on a C18 column with methanol–water (45 + 55) elution (35). The 3,5-dinitrophenyl derivatives elute as 5 peaks, of which only 4 could be identified. The fifth peak is assumed to be one of the minor gentamicin components. The elution order of the gentamicins is C2a, C1, C1a, and C2. No explanation for the separation of the stereoisomers C2a and C2, in such a manner that the 2 other components eluted in between, is provided. The reported elution order is different from that reported (29, 30) for the nitrophenyl derivatives of gentamicin. Amikacin, tobramycin, and kanamycin (35) also have been analyzed as 3,5-dinitrophenyl derivatives, while tobramycin (30, 36, 37), sisomicin (30), and amikacin (38, 39) have been analyzed as 2,4-dinitrophenyl derivatives. C18 columns have been used with a mobile phase containing acetonitrile and water for 3,5-dinitrophenyl derivatives or acetonitrile and acetic acid for 2,4-dinitrophenyl derivatives. The separation of different aminoglycosides as their 3,5-dinitrophenyl derivatives has not been investigated. As 2,4-dinitrophenyl derivatives, tobramycin and gentamicin components can be separated in one chromatographic run (36), but when sisomicin, a dehydro analogue of gentamicin C1, is added to the same run, it does not separate from gentamicin (30). Gentamicin has been used as the internal standard for tobramycin (36, 37). Apparently sisomicin can be quantitated more easily as an internal standard than gentamicin because it produces only one chromatographic peak. 1026 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 The separation of different 2,4-dinitrophenyl derivatives of aminoglycosides is assumed to be related to the different amounts of hydroxyl groups in the different aminoglycosides (30). Tobramycin has 2 hydroxyl groups more than sisomicin and gentamicin and is more polar. Thus, tobramycin should elute first. Sisomicin elutes before gentamicin, probably because of the polar double bond in its structure, and the gentamicins elute according to increasing degree of methylation (30). The dinitrophenyl derivative of amikacin is more polar than the corresponding derivatives of other aminoglycosides (38). As a consequence, in reversed-phase chromatography (RP), it requires a weaker eluent than the others. By reducing the ratio of acetonitrile in the mobile phase, separation of amikacin has been achieved (38). Kanamycin is a suitable internal standard for amikacin, with its dinitrophenyl derivative eluting after the amikacin derivative (38). 2,4,6-Trinitrobenzene sulfonic acid (TNBS) has also been used for precolumn derivatization of aminoglycosides (40–42). TNBS forms trinitrophenyl derivatives with aminoglycosides that are very similar to dinitrophenyl derivatives in their polarity and in derivatized groups. Amikacin (41) and tobramycin (42) have been analyzed as their trinitrophenyl derivatives with a C8 column, an acetonitrile– phosphate buffer mobile phase, sisomicin as the internal standard for tobramycin, and kanamycin as the internal standard for amikacin. Because of their different polarities, amikacin and tobramycin are analyzed with different mobile phases. Kanamycin and amikacin elute in the solvent front in tobramycin analysis (42). The elution order of the aminoglycosides in these methods is amikacin, kanamycin, tobramycin, and sisomicin, similar to the elution order of dinitrophenyl derivatives. Therefore, the number of hydroxyl groups can be assumed to determine also the elution order of the trinitrophenyl derivatives. The mobile phase is optimized by increasing the proportion of acetonitrile, adding tetrahydrofuran (THF), or changing the pH. The separation of tobramycin and sisomicin is insufficient when acetonitrile exceeds 75% (42). Addition of THF reduces tailing (41, 42), but the same results can be achieved also by reducing the pH (41). When the pH of the mobile phase is 3 or less, no tailing is observed and optimal conditions can be achieved without THF (41). Tobramycin, sisomicin, amikacin, the 4 components of gentamicin, and kanamycins A, B, and C have been analyzed as their trinitrophenyl derivatives with a C8 column and an acetonitrile–phosphate buffer–methanol mobile phase (40). The solvent ratio of the mobile phase has been varied according to the polarity of the analyzed aminoglycosides. The ratio of the organic phase is highest for gentamicin and lowest for amikacin. Kanamycin elutes before tobramycin when they are in the same run. When kanamycin is added to the amikacin analysis, it elutes clearly after amikacin. The possibility of separating all aminoglycosides in one chromatographic run or the possible interference caused by other aminoglycosides has not been investigated. Netilmicin has been detected as its dansyl derivative with a C18 column and an acetonitrile–water (95 + 5) mobile phase (43). Netilmicin elutes after gentamicin, suggesting that derivatized netilmicin is relatively nonpolar. Gentamicin C1 elutes almost simultaneously with netilmicin and interferes in the analysis. The separation of tobramycin and netilmicin as their OPA derivatives has been investigated with a C18 column and a methanol–EDTA–water mobile phase. Retention times are too long, and the separation is unsuccessful (33). Reversed-phase separation of amikacin as its OPA derivative with an EDTA–methanol–water–acetonitrile mobile phase has been reported (44). Neomycin has been analyzed as an OPA derivative with a methanol–EDTA–water mobile phase and a Supelco HISEP column that has a silica-based stationary phase containing hydrophobic regions shielded by a hydrophilic network (45). Nonderivatized kanamycin, tobramycin, and neamine have been separated by RP using amino-bonded column, an acetonitrile–acetate buffer (70 + 30) mobile phase, and a mass spectrometer as a detector (46). Underivatized kanamycin also has been separated with a C18 column and a methanol–acetate buffer mobile phase (47, 48). The underivatized kanamycin isomers A and C (47) and 3 gentamicin components (48) can be separated and identified by RP and MS. Ion-Pair Chromatography Ion-pair chromatography (IP) is well-suited to aminoglycosides because of their polar, charged, and basic characteristics under conditions usually used for this technique. IP seems the most popular chromatographic method for aminoglycoside analysis. The most widely used counterions are the negatively charged pentane, heptane, and hexane sulfonates. However, alkylsulfonates are not suitable for HPLC/MS applications because of their nonvolatile characteristics (16, 49). Instead, the volatile fluorinated carboxylic acids are used as counterions in HPLC/MS applications for aminoglycosides (49), as well as in preparative work (15). The counterions used for IP of aminoglycosides are presented in Table 5. The mechanisms of retention of aminoglycosides by use of alkylsulfonates as ion-pairing reagents have not been discussed in the literature. However, separation and retention are assumed to result from ion-pair formation between positively charged protonated aminoglycoside and anionic alkylsulfonate ions and the different interactions between the ion pair and the hydrophobic column phase (50). Derivatization markedly affects the polarity and acid–base characteristics of the analyzed compound and, it is assumed, also the IP separation. OPA has been the main derivatization reagent in IP of aminoglycosides. In postcolumn derivatization, ion-pair formation is not affected. But in precolumn derivatization, effects are likely to be observed. The protonated primary and secondary amino groups of aminoglycosides responsible for ion-pair formation are the same ones that undergo OPA derivatization. Thus, retention mechanisms and ion-pair formation for precolumn-derivatized and nonderivatized aminoglycosides must differ. Unfortunately, these differences have not been noted in the literature, and no hypothesis of the ion-pair formation between derivatized aminoglycoside and ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 1027 Table 5. Counterions used in ion-pair chromatographic analysis of aminoglycosides Ion-pair reagenta pns hxs hps ocs ocs and eds Column phase Aminoglycosidesa Derivatizationa References C8 gnt (3)b OPA pc 13 C8 neo OPA pc 25, 51 C18 neo OPA pc 53 C8 neo, par, str, dhs OPA pc 54 C18 gnt (3) — 55 C18 gnt (4) OPA pc 52 C18 ami, tob OPA pc 56 C18 gnt (3), sis, ntl OPA pc 50 C18 gnt (3), ami, tob OPA pc 57 C18 gnt (3) OPA pc 59 C8 gnt (4) OPA 59 C18 str, dhs — 58, 61 C8 str, dhs NQS pc 60, 89 C18 gnt (4) OPA 11, 63–66 C18 sis, ntl OPA 86 C18 gnt (3) OPA 67 C8 gnt (3) OPA 68 C8 ami, kan OPA pc 102 PLRPc neo — 8 C18 str NQS pc 88 C18 gnt (3) OPA pc 62 tls C8 sis, ntl, tob OPA pc 14 CS C18 gnt (1), neo, kan, ami, dhs, str OPA pc 5 TEA TFA HFBA PFPA a b c C18 gnt (3) OPA 71 C18 tob, ntl OPA 33 C18 str, kan, sis, gnt (4) — 15 gnt (4) — 16 str, dhs — 49 str, dhs, neo, gnt — 17 C18 For definitions of abbreviations, see footnotes to Table 4. Numbers in parentheses indicate the number of gentamicin components separated. Poly(styrenevinylbenzene) copolymer column. alkylsulfonate has been presented. Different mercaptans are used in OPA derivatizations, but their effect on the charge and polarity of the derivatives and on ion-pair formation has not been discussed. In most methods, Na2SO4 is added with the alkylsulfonate to the mobile phase (8, 13, 29, 50–57). The Na2SO4 concentration is the most important factor affecting the retention of aminoglycosides (8). In IP, the sulfate ion reduces retention times of aminoglycosides (13) and decreases k′ (8), apparently because the sulfate ion is more hydrophilic than the sulfonates used as counterions (8). The ionic strength of the mobile phase is also affected by the sulfate ion, and without sulfate in the mobile phase, an increase in the alkylsulfonate concentration reduces the retention of aminoglycosides, resulting in poor separation (50). Pentane, hexane, and heptane sulfonates have been compared in their ability to separate streptomycin and dihydrostreptomycin (58). The retention of streptomycin increases with the length of the carbon chain in the alkylsulfonate. Hexane sulfonate provides optimal separation in a relatively short time. For other aminoglycosides, the use of different alkylsulfonates has not been compared. In IP, the mobile-phase buffer should not form ion pairs but should have sufficient buffering capacity to maintain constant pH. Acetate buffer is the most widely used with aminoglycosides (25, 51–57, 59, 60) but phosphate buffer 1028 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 also has been investigated (8). With octane sulfonate as counterion, phosphate buffer is more advantageous than acetate buffer (8). Phosphate buffer is the choice when underivatized aminoglycosides are detected by UV absorption (195 nm) because it does not absorb UV light at this wavelength (58, 61). The proportions of phosphate buffer and alkylsulfonate affect separation characteristics (58). When the alkylsulfonate concentration exceeds the phosphate concentration, the retention of aminoglycosides can be increased by adding alkylsulfonate. When phosphate concentration is greater than alkylsulfonate concentration, the retention cannot be increased and tailing is observed. The effect of the mobile-phase buffer pH on the separation of neomycin (8), streptomycin, and dihydrostreptomycin (58, 61) has been investigated at pH 3–6 with octane (8) or hexane sulfonate (58, 61) as counterion. Neomycin is positively charged at pH < 5, and pH changes do not affect its retention. Thus, separation depends mainly on the sulfonate concentration (8). When hexane sulfonate is the ion-pairing reagent, streptomycin and dihydrostreptomycin cannot be separated at all at pH 6. Only when pH is reduced to 3 or less is satisfactory separation observed (61). However, another report concludes that pH 6 is optimal for separation (58). The pH and the ion strength of the injected sample have an effect on IP results (17, 25, 54, 57, 62). To achieve the best results, the sample should be dissolved in the mobile phase prior to injection (54, 57). When the sample is injected in basic solution after sample preparation, split peaks of paromomycin (54), neomycin (54), and gentamicin (57) are observed, resulting in nonuniform gentamicin chromatography (62). The inhomogenities, caused by sample addition into the mobile phase, move more slowly than the solvent front and result in nonuniform retention of the sample components (57). To avoid this effect, ion-pair concentrate and mobile-phase buffer are added to the sample before injection. Improved peak shape, separation, and repeatability are recorded (54, 57). Apparently, the effect varies among compounds. For example, using pentafluoropropionic acid (PFPA) as a counterion, sulfuric acid, HCl, and trifluoroacetic acid (TFA) causes peak splitting but formic acid does not (17). All IP methods for neomycin concern separation of the underivatized drug (8, 25, 51, 53, 54). Neomycin stereoisomers B and C have been separated with a polymer column with octane sulfonate as counterion (8). Neomycin B (25, 51, 53, 54) and paromomycin (54) have been analyzed with pentane sulfonate and Na2SO4 in the mobile phase. Separation is done with a C8 column and a methanol–acetate buffer (1.5 + 98.5) mobile phase. The life span of the C8 column is limited with this mobile phase (53). Column stability can be improved by increasing the proportion of methanol to 18%, reducing the concentration of Na2SO4, and changing the column to C18. These changes improve retention time repeatability (53). However, the C18 column cannot be used for simultaneous analysis of multiple aminoglycosides. Derivatization affects separation of gentamicin components with hexane sulfonate as counterion (59). Precolumn derivatization of gentamicin results in separation of 4 components, whereas only 3 components are separated with postcolumn derivatization. Similar results are achieved with other ion-pair reagents. Postcolumn derivatization usually results in 3 gentamicin peaks when pentane or octane sulfonates are used (50, 57, 62), but 4 peaks also are observed (52). Four gentamicin components can be separated after precolumn derivatization when hexane or heptane sulfonate is used as ion-pairing reagent (59, 63–66). The elution order of the OPA-derivatized gentamicins is C1, C1a, C2a, and C2. The separation of 3 gentamicin components is achieved with heptane sulfonate as counterion (11, 67, 68). The elution order of the components is C1, C1a, and C2 when the stereoisomers C2a and C2 elute together. Simultaneous elution of all gentamicin components is advantageous when a quantitative and sensitive method is required. Simultaneous elution has been achieved with the use of camphor sulfonate as the ion pair and methanol as the organic solvent in the mobile phase (5). Use of acetonitrile instead of methanol results in gentamicin eluting as 3 peaks. When camphor sulfonate is the counterion and EDTA is added to the mobile phase, the elution order of the aminoglycosides studied is amikacin, kanamycin, gentamicin, and neomycin (5). Dihydrostreptomycin elutes in the solvent front. The elution order coincides with the order of increasing amount of amino groups. Octane sulfonate, when used as a counterion with 2,2-ethane disulfonate in gentamicin analysis, causes gentamicin to elute after compounds with fewer amino groups. The use of 1,2-ethane disulfonate increases the separation of gentamicin components (62). Toluene sulfonate has been used as a counter ion for sisomicin, netilmicin, and tobramycin analysis, providing selective separation and lacking interference from other aminoglycosides (14). Perfluorocarboxylic acids have been used as volatile ion-pairing reagents in the analysis of aminoglycosides (15, 49, 69, 70). The applicabilities of the different fluorinated counterions for aminoglycoside analysis have been compared, and the separation mechanism has been investigated in detail (15, 69, 70). Inchauspé and Samain (15) concluded that PFPA and heptafluorobutyric acid (HFBA), despite their short carbon chains, are suitable as aminoglycoside counterions, except for gentamicin for which only poor separation could be achieved. TFA is especially selective to gentamicin because other aminoglycosides are not retained on reversed-phase columns with TFA (15, 69, 70). The retention of gentamicin depends on the concentration of TFA, and baseline separation of the different gentamicin components can be achieved (15). With camphor sulfonate as counterion, gentamicin does not separate at all. But with HFBA, gentamicin separates as a wide peak, and with PFPA, the different components are partially separated (69). For other aminoglycosides, TFA, when used with HFBA or PFPA, serves as an inorganic salt reducing the formation of ion pairs and decreasing kN (70). HFBA and TFA have opposite effects, ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 1029 and therefore, the chromatographic mechanisms of these 2 ion pairs are believed to be different (70). Inchauspé et al. assumed that the cause for the ability of TFA to selectively retain gentamicin is the unique methyl group in α-position relative to an amino group in the gentamicin structure, allowing simultaneous interaction of the amino group with the ion pair and the methyl group with the hydrophobic column phase (69). However, the methyl group can have an effect only when short-chain counterions are used (69). When HFBA and PFPA are compared, PFPA can be considered better because it does not adsorb to the stationary phase as strongly as HFBA and retention is clearly through an ion-pair mechanism. The best reagent combination is TFA and PFPA (70). Simultaneous separation of 5 aminoglycosides has been achieved by IP with PFPA as counterion (17). The aminoglycosides are streptomycin, dihydrostreptomycin, tobramycin, gentamicin as its 4 components, and neomycin. Detection and identification are by MS. Gradient elution is needed for satisfactory separation within reasonable time. With an ordinary C18 column, both gentamicin and neomycin peaks tail. YMC basic-column improves the peak shapes. Triethylamine (TEA) is the only positively charged counterion used for IP separation of aminoglycosides in gentamicin (71), tobramycin, and netilmicin (33) analyses. Gentamicin has been analyzed with a mobile phase containing phosphate buffer at pH 6.2 (71). EDTA and TEA in the mobile phase are compared. Alteration of the reagent does not change retention times, but separation and column stability are better with TEA. Basic tris-buffer–methanol (pH 7.9) mobile phase in tobramycin and netilmicin analysis causes netilmicin to elute first and the compounds separate well (33). Removal of tris-buffer from the mobile phase results in a very broad netilmicin peak. The chromatographic pattern is improved by TEA, eliminating peaks of the blank samples (33). Attempts to analyze gentamicin with this method have been unsuccesful because of the lack of retention (33). The effects of analyte charges and of pH on ion-pair formation and retention are not discussed. Ion-Exchange Chromatography Ion-exchange chromatography (IE) should be a popular method for analysis of polar aminoglycosides that are ionized in solutions. However, IE requires careful regulation of pH, temperature, and ionic strength, and it has not become widely used in aminoglycoside analysis. One of the earliest chromatographic methods for aminoglycoside analysis is anion-exchange chromatography (72–74). Kanamycin (72), paromomycin (72), and neomycin (72–74) have been analyzed with a column containing a strongly basic Dowex 1×2 anion exchanger and water elution. After separation, aminoglycosides have been detected as their ninhydrin derivatives (72, 73) or polarometrically (74). The authors assume that separation is based on adsorption of analytes to the ion-exchange resin according to their molecular weight/amino group content ratio (72). Neomycins B and C; kanamycins A, B, and C; and paromomycins I and II can be separated (72). Tobramycin also has been analyzed with a polymeric CarboPac PA1 anion exchanger and NaOH gradi- ent for elution (75). Pulsed amperometry is used for detection. The aminoglycosides are like other aminosugars, anionic at high pH and retained to anion exchangers (75). Cation exchange has been used to separate tobramycin (76), gentamicin (77), and kanamycin (78). With tobramycin, cation exchange serves primarily as a concentrating method before the main separation in a polystyrene column (76). Tobramycin is adsorbed to the column from phosphate buffer (pH 5.2) and elutes with 0.25M NaOH. Gentamicin has been analyzed with a strong-cation exchanger (Partisil SCX) and elution with acetonitrile–phosphate buffer after precolumn derivatization with fluorescamine, but the different components are not separated (77). The fluorescamine derivatives have a free carboxyl group. Therefore, an attempt has been made to analyze gentamicin by anion exchange at pH 6–8. A wide asymmetric peak indicates the presence of unreacted amine groups. These amino groups are positively charged and can be adsorbed to cation exchangers. Fluorescamine and OPA have been used as alternative postcolumn derivatization reagents for analysis of kanamycin with an SCX (78). A cellulose phosphate PII column has been used to separate gentamicin components and analyze the purity of the drug (10). Tobramycin, neomycin, sisomicin, kanamycin, and amikacin have been separated with an Amberlite IRC 50 column (79). Aminoglycosides have been eluted with sulfuric acid and postcolumn derivatized with dihydrolutidine (79). Normal-Phase Chromatography Normal-phase chromatography (adsorption chromatography) is usually used for simple compounds that are not ionized and that dissolve readily in organic solvents. The hydrophilicity and poor solubility in organic solvents of aminoglycosides makes normal-phase chromatographic analysis impossible without derivatization. However, normal-phase chromatography has been used for analysis of aminoglycoside drug preparations when separation of stereoisomers was of interest. The aminoglycosides are derivatized prior to separation. Isocratic methods for simultaneous analysis of neomycins B and C and neamine have been developed (80–82). Simultaneous analysis was achieved with neomycin derivatized with nephthalene sulfonyl chloride (NSCl; 81) or FDNB (80, 82). Helboe and Kryger (80) stated that in analysis of dinitrophenyl derivatives of neomycin 1,2-dichloroethane–methanol–water–diethylamine elution gives better and faster separation than chloroform–THF–water elution. Three gentamicin components and kanamycin can be analyzed by normal-phase chromatography with NSCl precolumn derivatization (81), and 3 gentamicin components, paromomycin, and kanamycin can be analyzed as their dinitrophenyl derivatives (82). When normal- and reversed-phase separations of dimethyl–phenyl derivatized amikacin are compared, the normal-phase method is more sensitive, although retention time variations are higher and repeatability is poor (39). 1030 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 Detection Methods in Aminoglycoside Analysis In GC, aminoglycosides have been determinated with either FID or ECD. ECD is used mainly for fluorinated aminoglycoside derivatives. Because GC methods are few, HPLC detection methods are the primary focus of this discussion. UV–Vis and refractive index detectors are popular HPLC detectors. Refractive index detectors are generally applicable, but they are sensitive to pressure and temperature changes. They also lack selectivity and sensitivity. However, refractive index detection is good for preparatory methods when sample derivatization is impossible and concentrations are high (15, 70). Refractive index detectors have been used mainly for preparatory separations of tobramycin, kanamycin, amikacin, sisomicin, streptomycin, neomycin, and gentamicin (15, 70). However, refractive index detection is too sensitive to changes in the environment for kanamycin analysis (78). Aminoglycosides lack UV-absorbing chromophores and have to be derivatized to gain UV detectability (12). In process control, when fast and specific chemical assay is needed, UV detection at 195 nm has been used for streptomycin and dihydrostreptomycin (58, 61). It has been assumed that the method would be applicable also to neomycin and paromomycin (58). Optical rotation has been used to detect nonderivatized aminoglycosides (10, 74, 83). Neomycin components A, B, and C are all optically active (74), and gentamicin includes 7 optically active components, of which C1, C1a, C2a, and C2 are biologically active (10). The optical activity allows their detection with polarimetry. The advantages of polarimetry over other direct detection methods are selectivity and sensitivity. Polarimetry has been considered a promising method for aminoglycoside applications (83). Derivatization in Detection Enhancement Derivatization facilitates extraction, analysis, and identification of poorly soluble and ionized compounds and improves sensitivity. In LC, derivatization can be performed pre- or postcolumn and on- or off-line. Most derivatizations improve UV, fluorescence, or electrochemical detectability. Fluorescing derivatives usually are more desirable because fluorescence detection generally provides better sensitivity and selectivity than UV absorption. Aminoglycosides lack chromophores and fluorescence, making derivatization essential for their UV detection (84). Aminoglycosides have several primary amino groups and, therefore, are easily derivatized (14). Partial derivatization is possible (44). The reactive hydroxyl groups further increase the possibility to form different derivatives (37). The reagents for aminoglycoside derivatization are presented in Table 6. In addition, a reagent called dihydrolutidine has been used to derivatize neomycin, sisomicin, kanamycin, tobramycin, and amikacin (79). Dihydrolutidine derivatives are formed when the amino group of the aminoglycoside reacts with the keto and enol forms of acetylacetone. The derivatives are fluorescent and can be detected with an excitation wavelength of 421 nm and an emission wavelength of 488 nm. The most widely used reagent is OPA, which reacts in the presence of mercaptan or other strong reducing agents in basic conditions with primary amines to form fluorescent derivatives (85). Its popularity is due to the fast rate of the reaction at room temperature and the possibility to perform the reaction in water solution (31, 78). In addition, OPA is stable in different buffers (78). Borate buffer (pH 9.5–10.5) has been used predominantly with OPA derivatization of aminoglycosides. In fluorescence detection, the excitation wavelength generally is 340 nm, and the emission wavelengths, 450 (86, 87), 430 (52), 440 (5), 418 (50, 56, 57, 67), 448 (68), or 455 nm (25, 51, 53, 54). Other excitation wavelengths used are 350, 360, and 365 nm, and the corresponding emission wavelengths are 450 (13, 44), 430 (32), and 440 nm (62). Background noise is lowest with an excitation wavelength of 260 nm and an emission wavelength of 340 nm (71). UV absorption at 330 (59, 63–66), 350 (11, 82), or 254 nm (82) is also used. The first reported OPA derivatization for aminoglycosides is postcolumn derivatization of kanamycin (78). The reducing agent is mercaptoethanol and the basic buffer is borate. In postcolumn derivatization of aminoglycosides, OPA usually has been used with mercaptoethanol (5, 13, 50, 54, 56, 57, 62, 78) in borate buffer at pH 9.5–10.5. Brij-35 has been used to prevent polysulfide precipitation in the detector (13). Triton-X-100 in lieu of Brij-35 enhances fluorescence (5, 56). However, Triton-X-100 is more temperature sensitive than Brij-35 (5). Derivatization temperature, reactor geometry, and flow rate of the derivatization reagent have been optimized (5, 54, 56, 57). The results are similar for the different aminoglycosides tested. Reactor geometry affects both sensitivity and separation (56). The reactor coil length is usually 1–2 m, but to separate gentamicin components properly, the coil must be less than 2 m (57). Resolution of amikacin suffers when the coil is more than 1 m (56). Reducing the internal diameter of the coil increases sensitivity (13, 57). The most common reaction temperatures for OPA postcolumn derivatization are 45E and 50EC. The effect of reaction temperature has been investigated in the range 20E to 75EC (5, 56). Best results are obtained for gentamicin (5) and tobramycin (56) at 45EC, for amikacin at 50EC (56), and for neomycin at 33EC (51). Partial derivatization occurs with amikacin at temperatures below 50EC, and fluorescence yield decreases at reaction temperatures over 60EC (56). This result has led to the assumption that fluorescence is temperature dependent. Reagent flow affects the observed fluorescence (5, 56, 57). However, the optimum reagent flow rate depends on other chromatographic parameters as well, and no absolute values can be determined. Some authors have stated that the optimum flow rate is half the flow rate of the mobile phase (54). Optimal flow rates ranging from 0.30 to 0.55 mL/min have been reported (5, 56, 57). If the flow is too high, the eluate is diluted to such an extent that the signal becomes too weak (56). Gentamicin components react differently to changes of reagent flow (57), which does not affect the fluorescence of C1a ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 1031 Table 6. Reagents used for aminoglycoside derivatizationa Aminoglycoside ami Derivatization reagent Detection method References OPA fl 26, 44 OPA pc fl 56, 57, 102 FDNB UV 38, 39 TNBS UV 40, 41 dib OPA fl 26 gnt OPA fl 26, 31–33, 59, 67, 68, 71 kan neo ntl par sis OPA UV 11, 59, 63–66 OPA pc fl 5, 13, 50, 52, 57, 62 FDNB UV 29, 30, 82 FMOC–Cl fl 34 BSC UV 27 Fluorescamine fl 77 NSCl UV 81 Dansyl chloride fl 28 DNBCl UV 35 TNBS UV 40 OPA pc fl 5, 13, 78 FDNB UV 38, 82 NSCl UV 81 TNBS UV 41 Fluorescamine fl 78 Ninhydrine UV 72 OPA pc fl 5, 13, 25, 51, 53, 54 OPA fl 45 FDNB UV 80, 82 NSCl UV 81 Ninhydrine UV 72, 73 OPA pc fl 13, 14, 50 OPA fl 26, 33, 86 OPA UV 65 BSC UV 27 Dansyl chloride fl 43 OPA pc fl 54 FDNB UV 82 Ninhydrine UV 72 OPA pc fl 13, 14, 50 26, 86, 87 OPA fl FDNB UV 30 TNBS UV 40, 42 str OPA pc fl 54 NQS fl 60, 88, 89 dhs OPA pc fl 54 NQS fl 60, 89 OPA pc fl 56 tob a OPA fl 26, 33 FDNB UV 30, 36, 37 TNBS UV 40, 42 For definitions, see footnotes to Table 4. 1032 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 but increases the fluorescence of C1 and C2 when flow rate is increased. This result suggests that the reaction of C1a is faster than the reactions of the other 2 components. A weakness of postcolumn derivatization is the increase of baseline noise as a consequence of reagent pumping (26, 71). Reagent consumption is also greater than with precolumn derivatization, and an extra pump is needed for the derivatization reagent. Improving the sensitivity in postcolumn derivatization is difficult because derivatization time can be increased only by increasing the dead volume of the system, which, in turn, results in broadening of bands. Precolumn OPA derivatization of aminoglycosides also has been reported (31–33, 44, 45, 64, 66–68, 86, 87). Advantages are simplicity of equipment and effective separation of interfering compounds (32). Different mercaptans, such as mercaptoethanol (31–33, 44, 45, 67, 68), mercaptoacetic acid (11, 59, 63–66), and mercaptopropionic acid (86, 87) are used in precolumn OPA derivatizations. However, information concerning their applicability is contradictory. Mercaptoethanol has been widely used and other reagents have been compared to it. Nonetheless, mercaptoethanol–OPA derivatives are not stable and, therefore, unsuitable for precolumn derivatization (87). OPA derivatives of aminoglycosides are stable for 8 h at –20EC (33), and OPA derivatives of neomycin are formed even during the first 5 min of freezing (45). When mercaptoethanol–OPA derivatives are kept at –4EC, the fluorescence yield decreases 5% in 1 h (33). At room temperature, the derivatives decompose exponentially, causing a 30% decrease in fluorescence in 10 min (78). Mercaptopropionic acid is used for OPA derivatization of sisomicin and netilmicin when better stability of derivatives is needed (86, 87). The steric group in mercaptopropionic acid is bigger than that in mercaptoethanol, and this is believed to improve the stability of the derivatives (87). The solution where derivatives are kept also affect stability (86). Unlike what happens with mercaptoethanol, the extra mercaptan does not destabilize the derivatives when mercaptopropionic acid is used (87). A reaction time of 10 min is satisfactory at 60EC, but at 20EC, 1 h is needed. Partial derivatization of sisomicin results in formation of 3 different derivatives. The reaction is accompanied by interfering side reactions. Mercaptoacetic acid–OPA derivatization of gentamicin (11, 63–66) and netilmicin (63) is performed either at 90EC (64) or at 60EC (66) for 15 min. The derivatives are more stable than those formed with mercaptoethanol (65), but they decompose under acidic conditions (63). A 16–43% decomposition occurs during the first hour after derivatization if derivatives are kept in acidic buffer or in the mobile phase (63). This decomposition does not occur in basic solutions. Gentamicin yields 4 derivatives with mercaptoacetic acid, compared with 3 with mercaptoethanol. Detector responses of components are different. The relationship between response and number of derivatized groups is not discussed. Only mercaptoethanol has been used as reducing agent in OPA derivatization of aminoglycosides within the precolumn. Silica (32, 44, 67, 68) and Amberlite CG-50 ion exchange (45) precolumns have been used. The technique has been used for amikacin (44), gentamicin (32, 67, 68), and neomycin (45). The stability of derivatives is not discussed, except that neomycin needs to be frozen for complete derivatization (45). Detector responses of OPA derivatives of the different gentamicin components have been compared and results differ. On the basis of the number of derivatized groups, the responses of C1a and C2 should be equal and that of C1 smaller. Same responses indeed are reported for C1a and C2, but the response is much smaller than that recorded for C1 (71). Another study finds equal C2 and C1 responses but a smaller C1a response (32). Moreover, different detector responses are reported for all the components (26, 66). Reaction kinetics and reasons for the observed differences are not discussed. Furthermore, most authors use a gentamicin standard that contains an unknown mixture of the components. Consequently, identification of components is not based on individual standards. The variation in detector responses indicates that different derivatives are formed with the various methods, and they probably also affect the elution order. However, the same elution order is suggested in different reports. In partial OPA derivatization of amikacin (44) and neomycin (45), 2 peaks are observed instead of one for neomycin, and several peaks are produced for amikacin if the derivatization solution is not heated. Discussion of partial derivatization has been limited, and most authors do not take this possibility into account. FDNB reacts with primary and secondary amines in basic conditions (36), producing highly UV absorbing derivatives (37). The UV absorption maximum is at 365 nm. FDNB has been used for derivatization of almost all aminoglycosides (29, 30, 36–39, 80, 82), and a comprehensive study of its reaction kinetics with aminoglycosides has been published with tobramycin as model compound (37). Derivatization conditions have been developed with borate (39, 80, 82) or tris (29, 30, 36, 37) buffers. Phosphate, phthalate, acetate, and bicarbonate buffers also have been investigated (37, 39). FDNB reacts with the amine or hydroxyl groups of most buffers, producing compounds that interfere with separation and detection (39). However, FDNB does not react with tertiary amines. Thus use of such buffers seems the preferred approach. The main problem of FDNB derivatization of aminoglycosides is the different solubilities of the reagent and the analytes (82). Organic solvents must be used in the derivatization solution to dissolve FDNB. Derivatization buffers, organic solvents, temperatures, times, and pH conditions have been varied to optimize derivatization conditions. Also a big problem with FDNB is its toxicity (37). FDNB irritates the skin and may cause blistering dermatitis and severe allergic reactions. Absorption of FDNB by inhalation, ingestion, or through the skin can be lethal. The different methods used for aminoglycoside derivatization with FDNB are presented in Table 7. Tobramycin analysis has revealed that yield and speed of the FDNB reaction, as well as occurrence of several side reactions, are strongly pH dependent (37). Derivatization of tobramycin amine groups and of hydroxyl groups thereafter follow second-order kinetics. Hydrolysis of FDNB in basic ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 1033 Table 7. Reaction conditions for FDNB derivatization of aminoglycosidesa Aminoglycoside gnt, tob gnt, sis, tob Solvent Buffer Temperature, EC Time, min References ACN Tris pH 9.0 80 45 29, 36 ACN Tris pH 9.0 80 45 30 tob Ethylisopropylamine Tris pH 8 70 ? 37 neo MeOH Borate 0.02M pH 9 60 60 80 ami MeOH NaOH 100 45 38 neo MeOH Borate 0.02M pH 9 100 45 82 ami MeOH Borate 0.1M pH 9.3 80 30 39 a For definitions, see footnotes to Table 4. solutions follows pseudo first-order kinetics and is faster the more basic the derivatization solution is. Hydrolysis of the tobramycin derivative decreases with increasing derivatization temperature, but increasing the solution pH results in faster dissociation. The optimum pH range of the reaction is very narrow because of the high reactivity of the aminoglycoside hydroxyl groups with FDNB and the pH-dependent dissociation of the derivatives (37). Best results for gentamicin, sisomicin, and tobramycin derivatization with FDNB have been achieved with tris buffer at pH > 9.5 and 80EC (29). Higher reaction temperature decreases the repeatability, and reaction time of 90 min results in incomplete derivatization (29). Molar absorptivity and elemental analysis have confirmed derivatization of all amino groups of tobramycin (36). Gentamicin derivatization is more sensitive to derivatization conditions than tobramycin. Proton nuclear magnetic resonance spectrometry of the amikacin–FDNB derivative indicates that amino groups are derivatized (38, 39). Derivatization yields of 92% for standards and 77% for serum have been achieved with NaOH (39). NaOH gives slightly higher yields than borate buffer (38). Inorganic salts such as Na2SO4 interfere with derivatization. Side products of aminoglycoside derivatization with FDNB in tris buffer include 1-hydroxy-2,4-dinitrobenzene (the reaction product of FDNB and water), 2-[N-(2,4-dinitrobenzene)amino]-2-hydroxymethyl-1,3-prop anediol (the reaction product of tris and FDNB), and 2,4-dinitrophenyl, the hydrolysis product FDNB (36). Generally, these side products can be seen in chromatograms, and they may interfere with analysis. FMOC–Cl is a very reactive derivatization reagent, reacting with both primary and secondary amines (34). The reaction is very fast in basic solutions at room temperature. The derivatives are very stable and fluorescent. However, during derivatization FMOC–OH is formed, which can interfere with chromatography. For gentamicin derivatization (34), the best derivatizating solution is borate–acetonitrile (1 + 1) at pH 7.5–9.0. Acetonitrile is needed to dissolve FMOC–Cl. Derivatization is complete in 10 min. The detector responses of the FMOC derivatives of gentamicin components are equal. Naphthoquinone sulfonate (NQS) reacts with guanidine compounds in basic conditions to produce fluorescent derivatives (88). It is suitable for derivatization of streptomycin and dihydrostreptomycin. NQS decomposes rapidly in basic solutions. Therefore, it is not suitable for postcolumn derivatization in a basic environment. Postcolumn derivatization with NQS would require 2 extra pumps, one for NQS and one for the base. NQS has been added as a solution to the acidic mobile phase before separation (88). Optimal conditions for streptomycin analysis include a 5 m postcolumn reactor coil at 65EC, NQS concentration of 0.4 mM, and NaOH concentration of 0.5M (88, 89). A high-quality postcolumn system is essential for maintaining baseline stability and constant reactor temperature (89). For fluorescence detection, excitation wavelengths of 347, 351, and 365 nm have been used with emission wavelengths of 418, 420, and 418 nm, respectively (60, 88, 89). TNBS reacts with primary amines in basic aqueous solution at room temperature without interfering side reactions (41, 42). The molar absorptivity of the TNBS derivatives is highest at 340 nm. TNBS has been used to derivatize gentamicin (40), tobramycin (42), and amikacin (41) with kanamicin (41) and sisomicin (42) as internal standards. The aminoglycosides are dissolved in acetonitrile to speed up the derivatization and to dissolve the nonpolar derivatives (41). At room temperature, the reaction requires 19–20 h (42). Increasing the temperature improves the results, and 70EC is best temperature for derivatization (41, 42). At lower temperatures, several derivatives are formed when derivatization time is less than 30 min. At temperatures higher than 80EC, tobramycin derivatives decompose (42). TNBS derivatization of gentamicin is complete in 15 min at 70EC (40). The trinitrophenyl derivatives have absorption maxima at 350 and 420 nm and give different detector responses. TNBS derivatization of amikacin and tobramycin is affected by the pH and the buffer capacity of the solution (41, 42). At pH < 9, the reaction is not complete, and at pH > 10, the reagent decomposes. Derivatization, therefore, is performed within pH 9.5–10.0. Carbonate, borate, and phosphate buffers are not satisfactory because of their low buffering capacities in this pH range and solubility problems resulting in 1034 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 complex formation with aminoglycosides. Therefore, tris buffer produces the best results (41, 42). Fluorescamine reacts rapidly in a basic aqueous solution at room temperature with primary amines to produce strongly fluorescent derivatives (90). The reagent itself and its degradation products are not fluorescent. Kanamycin (78) and gentamicin (77) have been derivatized with fluorescamine. For kanamycin, postcolumn derivatization in acetone has been used (78). The derivatives are stable, and the sensitivity of the method is good. For precolumn derivatization of gentamicin, best results are achieved with 0.03M or stronger phosphate buffer at pH 7.2–8.0 (77). The fluorescamine concentration must exceed the gentamicin concentration by 10 times. The fluorescence responses of the different gentamicin components are equal, and the method is accurate and repeatable. Dansyl chloride has been used for precolumn derivatization of netilmicin (43) and gentamicin (28, 91). Dansyl chloride reacts with amines to form fluorescent derivatives (92). The derivatization solution for aminoglycosides is acetonitrile–phosphate buffer at pH 11 (28, 43, 91). Reaction temperature and time are 75EC and 5 min, respectively. Buffer volume, pH, reaction temperature, and time affect results (28). At 100EC, the derivatives decompose rapidly, and at low temperatures, reaction is not complete. The possibility of partial derivatization of gentamicin has been investigated but could not be detected (28). Furthermore, dansyl chloride reagents from different producers differ in solubility, and a too low dansyl chloride amount gives incorrect results (91). DNBCl has been used to derivatize gentamicin, tobramycin, kanamycin, and amikacin (35). Derivatives are detected with UV at 254 nm. The reaction is repeatable, and the derivatives are stable. However, an unidentified derivative is formed with gentamicin. BSC has been used to derivatize netilmicin and gentamicin (27). Derivatization must be at pH 7–9 and 75EC for 10 min. Too high a temperature or too long a reaction time destroys the derivatives. NSCl reacts by nucleophilic substitution with hydroxyl and amine groups, forming UV-absorbing (254 nm) derivatives (93). Neomycin, gentamicin, and kanamycin have been derivatized with NSCl at pH 8.0–9.0 and 100EC for 10 min (81, 93). Phosphate buffer is best for NSCl derivatization of aminoglycosides. It is difficult to compare the different derivatization reagents used in aminoglycoside analysis because most reports do not give derivatization yields and method sensitivities. Fluorescent derivatives do not seem to be detectable at lower concentrations than UV-absorbing derivatives, as would be expected. Postcolumn derivatization with OPA appears preferable when analysis speed is critical because the derivatization time is negligible. However, because of the limitations of postcolumn derivatizations, use of an OPA precolumn derivatization could be a better option when reaction time is 0–15 min. Other fast-reacting reagents are FMOC–Cl, dansyl chloride, and BSC, which all react in <10 min. The disadvan- tage of dansyl chloride is that the derivative has to be extracted from the derivatization solution before chromatography. Dansyl chloride has been applied to gentamicin (28) and netilmicin (43); FMOC–Cl, to gentamicin (34); and BSC, to gentamicin and netilmicin (27). No studies about the applicability of these reagents to other aminoglycosides have been reported. In terms of derivatization time, FDNB is the most problematic reagent because the reaction requires 30–60 min. Long incubations may be a problem also for analyte stability, and some losses may occur. Buffer selection is not critical because all reagents require a basic environment. FDNB reagent is the most limited in terms of buffer selection because of its side reactions with buffer ions and its narrow pH optimum for the reaction. The limited solubility of aminoglycosides demands use of reagents that are soluble in water or aqueous solutions of methanol or acetonitrile. The most problematic reagents in terms of solubility are dansyl chloride and FDNB. The dinitrophenyl derivatives of aminoglycosides also are poorly water soluble. OPA has the best solubility characteristics among derivatization reagents. The stability of the aminoglycoside derivatives has been subject to limited investigation. Because most reagents are used for precolumn derivatization, the stability of derivatives is very important in consideration of the best derivatization option especially when automatic equipment is not available and quantitative results are needed. Furthermore, the structures of aminoglycoside derivatives formed with different reagents have been inadequately investigated. The number of derivatized groups is rarely specified. When several different derivatives are formed, attention should be paid to the derivatization yield and to the number of different derivatives formed. Of the reagents used for aminoglycoside derivatization, only FDNB has been investigated with respect to derivatization kinetics and the structures of the aminoglycoside derivatives formed. Electrochemical Detection Amines, thiols, and hydroxyl groups oxidize readily at electrodes such as the carbon electrodes used for electrochemical detection (55). The multiple amino and hydroxyl groups in the aminoglycoside structures are apparently the source of their electrochemical activity (55). The advantage of electrochemical detection in aminoglycoside analysis is that it does not require derivatization. However, the structure of the detected compound cannot be identified, and confirmation of a specific chromatographic peak is not possible (16). The inability to do these is a particular disadvantage when retention times and elution orders change. All gentamicin components have been detected electrochemically with a carbon electrode (16, 55). A voltammogram measured manually for gentamicin and a detection voltage of +1.3 V was selected. When the detection potential was changed to a more positive potential, the signal increased but the baseline deteriorated. The detector response for gentamicin was linear from 16 to 30 µg with r = 0.999. Substantial changes in the detector response were observed ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 1035 Table 8. Pulsed electrochemical methods for aminoglycoside detectiona Aminoglycoside Working electrode PAD tob Pt, rotating disc Ag/AgCl 0.55 0.7 –0.9 250 125 425 0.25 76 PAD str, dhs Au Not defined 0.1 0.6 –0.8 480 120 300 0.3 49 PAD tob Au Not defined 0.1 0.6 –0.8 300 120 300 0.5 75 PED neo Au Ag/AgCl 0.05 0.75 –0.15 400 190 390 0.5 8 PED tob Pt, rotating disc Ag/AgCl 0.7 –0.2 125 125 400 0.25 76 Method a b Counter electrode E1, V E2, V –1.3 E3, V t1, ms t2, ms t3, ms NaOH, Mb Ref. For definitions, see footnotes to Table 4. Concentration after column. with changes of Na2SO4 concentration in the mobile phase. A change from 0.2 to 0.3M reduced peak areas by 40%. Pulsed methods for aminoglycoside detection are presented in Table 8. All aminoglycosides react via a common mechanism in pulsed amperometry (76). The sensitivity of detection in neomycin analysis with pulsed electrochemical detection (PED) improves by adding NaOH to the eluent to increase the pH to 13 (8). In tobramycin analysis, a high pH is required to oxidize tobramycin, and a gold electrode is used (75). The sensitivity of the tobramycin method (75) is similar to or better than that reported for different derivatization methods. With an injection volume of 10 µL, the limit of detection of the method is 2 ng/injection, and the detector response is linear from 10 ng to 1 µg/injection with r = 0.999 (75). Stability problems have been encountered when neomycin was analyzed with PED, and specialization and experience are required from the analyst to achieve accurate and repeatable results (8). Mass Spectrometry In theory, all drug compounds can be detected by MS but the use is limited by the complexity of the equipment and high costs (83). Regulatory methods require confirmatory identification at the action level (94). Because of the sensitivity and possibility for confirmatory identification, MS is often the only method that offers the required sensitivity and selectivity for confirmation of different analytes. No GC/MS application for aminoglycosides has been published. Available methods have been limited to LC/MS instruments and to the MS analysis of pure standards. LC/MS interfaces used in aminoglycoside analysis have been thermospray (16), moving belt (46), ion spray (17, 49), and heated pneumatic nebulization with atomspheric pressure chemical ionization (APCI; 47, 48). For aminoglycoside ionization in MS analysis, plasma (95) and field (69) desorption, chemical ionization (CI; 46, 96), and APCI (47, 48) have been used. The ion spray interface (IS) is well adapted to aminoglycosides that are ionized in solutions (49). Desorption of ions from the liquid phase to the gas phase is caused by an electric field without heating or other severe conditions. The advantage of IS–MS in quantitative work is that the ion flow concentrates one specific species, improving sensitivity. A disadvantage of IS is that the ionic strength of the mobile phase has to be minimized while the organic solvent content is maximized. This is a limitation especially for aminoglycosides, which are usually analyzed by IP. For this reason, the concentration of the ion-pair reagent must be carefully optimized. However, the counterion concentration does not have significant effects on mass spectra (49). Gentamicin, neomycin, tobramycin, streptomycin, and dihydrostreptomycin have been analyzed with LC/IS–MS (17, 49). Better sensitivity has been obtained with positive-ion than with negative-ion monitoring when IS–MS is used in aminoglycoside analysis. Thus, positive-ion spectra have been investigated in more detail (17). Selected-ion monitoring does not provide sufficient selectivity for complex matrixes, but selectivity and sensitivity can be improved with MS/MS. For these reasons, MS/MS appears especially suitable for confirmative analysis. With limited heating in IS, only a few fragments can be observed. Collision-induced activation (CAD) allows formation of daughter ions from various ions of the spectra, enabling confirmation of the structure (17). In MS/MS, use of singly charged parent ions of aminoglycosides provides a lower signal-to-noise (S/N) ratio in the chromatograms than use of doubly charged parent ions, partly because of the smaller absolute intensity of singly charged species and partly because of the mutual repulsion of protonated groups in doubly charged species. In the mass spectra of aminoglycosides, doubly charged species dominate, and single or triple charging rarely occurs (17). However, when the ionization potential is increased, the quantity of singly charged ions increases (17, 49). With heated pneumatic nebulization systems, APCI is usually preferred (47, 48). LC–APCI is generally applicable for nonvolatile compounds that are unsuitable for GC analysis. The main advantage of APCI–MS is that it can be performed under usual HPLC separation conditions with flow rates of 0.3–2.0 mL/min (47). The most important parameters to be optimized in analysis of nonvolatile compounds are the temperatures of the nebulizer and vaporizer (48). Kanamycin (47, 48) and gentamicin (48) have been detected by LC–APCI. The mass spectra of gentamicin components are different from each other in APCI–MS (48). The kanamycin isomers A, B, and C can be separated by LC–APCI, and the mass spectra of the different components are also different. 1036 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 Thermal decomposition of several aminoglycosides has been observed in LC–APCI interfaces (17). APCI spectra resemble CI spectra but with less fragmentation. The limited fragmentation frequently offers insufficient structural information (48) and, therefore, CAD is used with APCI–MS to obtain structure-confirming fragment ions. Most ions obtained with CAD also can be seen in the original APCI spectra. The thermospray interface (TS; 75) has been used for gentamicin analysis (16). Three gentamicin components (C1, C1a, and C2) and another component assumed to be gentamicin C2a have been detected. Instead of strong molecular ions [M + H], gentamicin components produce high intensity fragments with TS–MS. Identification of the individual components is possible on the basis of the mass spectral information, and changes in elution order can be detected immediately. The greatest advantage of the MS detection apart from identification is the lack of need for derivatization. In addition, the sensitivity of LC–TS for gentamicin is much better than the sensitivity obtained with an electrochemical or UV–Vis detector (16). The suitability of moving-belt interfaces for aminoglycoside analysis has been investigated (46). Kanamycin, tobramycin, and neamine have been studied with chemical ionization with ammonia. Different belt types have an effect on the spectra. The molecular weight and the order of amino sugars in the molecule can be determined for all aminoglycosides studied with this method (46). Plasma desorption MS (PD–MS) yields useful information about the structure of nonvolatile or thermolabile aminoglycosides (95). The mass spectra of neomycin, kanamycin, paromomycin, tobramycin, streptomycin, dihydrostreptomycin, amikacin, netilmicin, sisomicin, and gentamicin have been determined with time-of-flight (TOF) PD–MS and positive-ion monitoring (95). This method facilitates confirmative analysis of aminoglycosides presumably because the resulting spectra are very repeatable. Very strong molecular or quasimolecular ions are obtained for aminoglycosides studied with positive-ion monitoring, but much more fragmentation occurs with negative-ion monitoring. Extraction of strongly polar aminoglycosides from biological matrixes can be difficult. Therefore, more easily extractable dinitrophenyl derivatives have been prepared, and their ionization and detectability have been studied. Except for derivatives of streptomycin and dihydrostreptomycin, the molecular weights and mass spectra of all the dinitrophenyl derivatives can be recorded with positive- and negative-ion monitoring. The best spectra are obtained with positive-ion monitoring (95). Negative ionization produces only little fragmentation for the derivatized aminoglycosides and [M–H]– or M− C is the main peak formed. The molecular ion is either very weak or does not exist at all in aminoglycoside mass spectra when electron impact ionization (EI) is used (95). Field desorption MS (FD–MS) has been used to determine gentamicin from LC fractions (69). However FD–MS is usually applicable only to molecular weight determination because practically no fragmentation occurs (96). Thus, emitter chemical ionization (CI), a tech- nique characterized by use of the activated FD emitter as a solid probe, has been used for aminoglycosides to obtain both molecular ions and characteristic fragments (96). Gentamicin, kanamycin, and dibekacin have been analyzed with ammonia or isobutane as reagent gas. The molecular peak and 9 characteristic ions resulting from the dissociation of the glycoside bonds can be detected with both reagent gases. The molecular weights of kanamycin and dibekacin cannot be determined from MS using conventional CI. Thus, emitter CI can be considered a better method. Generally, aminoglycosides fragment between the glycosidic oxygen and the anomeric carbon when the hydrogen is transferred to oxygen (96). Extraction and Analysis of Aminoglycosides from Different Matrixes Pure Drug Preparations The quality of drug preparations is continuously monitored (1) during preparation by the producer and later in quality control according to requirements of pharmacopoeias. Confirmative methods of analysis are preferred. When the drug is formed from several components, limits for component ratios are given (63). Fast, specific chromatographic methods are needed for quality control of aminoglycoside preparations and injectable solutions (64). Accurate methods for measuring gentamicin components are required because the different components apparently have different toxicity (40). Quality control chromatographic methods for gentamicin have to be in agreement with the official microbiological method (59). Compatible results between bioassay and chromatographic methods can be achieved only when gentamicin components C2a and C2 elute together (59). Remarkable differences (up to 100%) in the ratios of different components have been found among gentamicins from different producers (11). Gentamicin component ratio (10, 11, 13, 16, 55, 59, 63, 64) and stability (66) have been extensively studied. Occasionally, when gentamicin preparations are analyzed, unsubstantiated assumptions have been made. These include the presumptions that equimolar amounts of the different components give the same detector response as OPA derivatives (59, 63, 64) or that the microbiological activities of the different components are equal (64). Some authors have used individual gentamicin components for quantitation (10, 11, 13, 55). The area response factors of the 4 OPA-derivatized gentamicin components differ (11). Consequently, a quantitative method based on the assumption that the area response factors of OPA-derivatized gentamicins are similar gives results that are not representative of the actual component ratio (11). In quantitation of the 4 gentamicin components, it is necessary to know their specific responses in the chromatographic procedure (11). Different references give divergent analytical results, and the component C2a has been determined differently in different papers. Some authors consider it as an individual component (16, 63–66), while others include it in the C2 component (29–33, 48, 55, 67, 68, 71), making comparison of methods difficult. ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 1037 Neomycin in ointments has been analyzed by dissolving samples in chloroform and separating neomycin by centrifugation (19). The neomycin is then dissolved in water and freeze dried for GC analysis. Recovery of the method is 98–100%. Dissolution of ointments in chloroform also has been used for gentamicin analysis (66). Gentamicin is extracted from the chloroform solution with phosphate buffer. The gentamicin in the phosphate buffer is derivatized with OPA and analyzed by HPLC. The recovery is 94–97%. Plasma and Serum The response of a patient to a specific drug depends on various factors, such as age, sex, renal and liver function, and simultaneous use of other drugs (97). Concentration in serum after drug administration is often difficult to predict (33). Correlation between the achieved concentration in serum and pharmacological and toxicological effects is substantial, and knowledge of the concentration in serum is of great help in treatment, especially when the therapeutic range is narrow. Therapeutic drug monitoring is used frequently when aminoglycosides or certain other drugs are used for treatment (14, 25, 29, 40, 57, 97). The methods used for monitoring aminoglycoside concentrations in plasma and serum are summarized in Table 9. Because of their solubility characteristics, direct extraction of aminoglycosides from plasma is difficult and impractical (27, 28). Protein precipitation and separation of amino acids from aminoglycosides prior to chromatography is necessary because of interference both in chromatography and in derivatization (14, 28). Interferences have been reported in gentamicin, amikacin, tobramycin (57), and streptomycin (61) analyses. For protein precipitation, trichloroacetic acid (TCA; 25), acidic or basic phosphate buffer with acetonitrile (27, 28, 33, 43, 77), basic tris buffer with acetonitrile (29–31, 36), phosphate buffer (44), or tris buffer (41, 42) alone and acid incubation (23) have been used. Contradictory information is available about the applicability of different buffers and organic solvents for aminoglycoside extraction, and different aminoglycosides are often extracted with different buffers or solvents. TCA and tungstic acid precipitate gentamicin with the proteins (28). However, >90% recovery of neomycin from serum has been achieved with TCA to precipitate proteins (25). However, neomycin recoveries from plasma are lower than from water when TCA is used, perhaps because of endogenic substances in the plasma. The pH of the precipitation buffer affects streptomycin recovery (61). Recovery is maximum at pH 2 and decreases with increasing pH. In gentamicin analysis, ultrafiltration is useless because the compound is adsorbed to the membrane (28). Tris buffer is well suited for extraction of tobramycin and amikacin from serum and for protein precipitation (41, 42). However, protein precipitation is completed with TNBS−acetonitrile derivatization solution. Tris buffer together with acetonitrile results in considerably lower tobramycin recovery than has been obtained for gentamicin or sisomicin (36) and also less than observed for tobramycin with tris alone. Phosphate buffer with acetonitrile provides better recovery for netilmicin than for gentamicin, while tobramycin recovery decreases with decreasing phosphate dilution volume (33). However, recoveries with phosphate buffer and acetonitrile are higher than those with tris and acetonitrile. Preference has been given to tris buffer in precipitation because inorganic buffers tend to precipitate from acetonitrile–water solutions (29). Acetonitrile is the best organic solvent for extracting gentamicin from serum and for precipitating serum proteins (28, 77). Other water-soluble organic solvents such as ethanol, methanol, and acetone do not provide satisfactory results (28, 77). Use of methanol for protein precipitation and gentamicin extraction from plasma has been reported (62). Diethyl ether, ethyl acetate, hexane, and methylene chloride have been investigated for back extraction of the precipitation mixture (28, 77). Best results are achieved with methylene chloride (28, 77). Gentamicin recovery is poor (77), but it could be improved by adding phosphate buffer to acetonitrile. Tobramycin recovery is also low after acetonitrile precipitation and cannot be improved (30). Solid-phase extraction (SPE) has been used to extract TNBS and FDNB derivatives of aminoglycosides (39, 41, 42). This extraction prolongs column life and removes the large solvent front and interfering peaks. Silica SPE has been used to extract and clean up dinitrophenyl derivatives of amikacin (39). The derivatives are eluted from silica with acetonitrile–water (68 + 32). Trinitrobenzoyl derivatives of amikacin and tobramycin (with kanamycin and sisomicin as internal standards) have been extracted with a Bond-Elut C18 solid-phase column (41, 42). SPE with C18 (61) and silica (32, 44, 71) columns and CM-Sephadex C25 ion-exchange resins (14, 38, 50, 56, 57, 87) has been used to extract aminoglycosides from plasma. SPE has been used without precipitation of serum proteins, and aminoglycosides have been separated from the matrix according to partition differences (32). Among organic solvents, SPE and CM-Sephadex, CM-Sephadex gives the highest recoveries for sisomicin, tobramycin, and netilmicin (14). Aminoglycosides are strongly basic compounds, and ion exchange can be expected to be a good sample preparation method. CM-Sephadex is applicable to all aminoglycosides except streptomycin (14, 38, 50, 56, 57, 87). In extraction of streptomycin from plasma, CM-Sephadex ion exchange has been considered too complicated, and interfering compounds in plasma cannot be separated from streptomycin (61). Silica gel has been used for sample cleanup and for extraction of gentamicin (32, 71) and amikacin (44). Gentamicin is charged to the silica column in water-diluted serum, and amikacin, in phosphate–buffer-diluted serum. These aminoglycosides have been derivatized with OPA in the silica column before elution with ethanol (32, 44) or methanol (71). Ion exchangers with carboxylic acid functional groups have worked well in extraction of gentamicin (34). Gentamicin is retained on a carboxypropylsilica ion-exchange column at pH 6–8 and eluted at pH > 9.5. The ion-exchange method is simple and reliable. Table 9. Methods used to monitor aminoglycoside levels in plasma and serum Aminoglycosidea sis, ntl, tob* Extraction and protein precipitation Cleanup Chromatography derivatizationb Intraday CV, % Interday CV, % r Determination Slope of Linear range, limit, µg/mL calibration curve µg/mL Recovery, % Ref. — CM-Sephade x IP, OPA pc — 1.9–3.1 0.999 — 0.034 (sis) 0.101 (ntl) 0.3–22 (sis) 0.2–11 (ntl) — 14 0.5% H2SO4 incubation 80EC Hexane GC, TMSI/HFBI — — 0.979 <0.6 — 2.5–20 — 23 gnt, tob, ntl, 0.5% H2SO4 incubation 83EC ami, par*, kan* Hexane GC, TMSI/HFBI — — — <0.6 — 2.5–20 — 23 gnt, tob, kan* gnt — CM-Sephade x IP, OPA pc — — 0.997 — 4.72 1.0–10 95 50 neo 20% TCA — IP, OPA 4.4 9.8 0.981 <0.25 — 0.25–1.0 90–116 25 — Silica RP, OPA 5.9 — 0.999 (gnt) — 16.6 (gnt C1) 9.5 (gnt C1a) 7.2 (gnt C2) 0–16 — 26 gnt, ntl* ACN + fos, NaOH — RP, BSC 2.0–5.1 — — 0.2 — 2.5–10 — 27 gnt ACN + fos, NaOH DCM RP, DansCl 0.84–3.00 6.11–7.61 0.999 1.2 1.516 (gnt C1) 0.832 (gnt C1a and gnt C2) 0–40 — 28 gnt ACN + tris — RP FDNB — 1.7–5.9 — 0.33 — 1–16 83–84 29 gnt, sis, tob* ACN + tris — RP FDNB — 1–10 0.999 — 0.267 (sis) 0.091 (C1a) 0.204 (gnt C1 and gnt C2) 0.5–16 84 (sis) 83–84 (gnt) 65 (tob) 30 ami, dib, gnt, ntl, sis, tob gnt ACN + tris chl RP, OPA 2.2–14.4 ≤8 0.99 0.5 — 0.5–40 ≥85 31 gnt Incubation 100EC Silica RP, OPA 6 3.9–7.5 — — — 0–20.0 80–105 32 ACN + fos DCM IP, OPA 1.7–7.9 (n) 3.6–7.4 (n) — 0.5 — 0–24 110–96.7 (ntl) 94.1–98.3 (tob) 91.5–91.8 (gnt) 33 — Carboxy propylsilica RP, FMOC 4.3–8.6 5.0–8.9 0.999 <0.05 72.7 (gnt C2a) 120.7 (gnt C2) 118.2 (gnt C1) 92.9 (gnt C1a) 0–5 93.9 (gnt C2a) 97.8 (gnt C2) 96.8 (gnt C1) 99.0 (gnt C1a) 34 tob, gnt* ACN + tris — RP FDNB — ≤3 0.998 0.25 0.398 0.5–16 75 (tob) 36 ami, kan* — CM-Sephade x RP FDNB 1.5–5.3 9 0.993 1 — 1–64 95 (ami) 92 (kan) 38 MeOH + borate Silica RP, FDNB — 0.8–9.9 0.996 — 3.21 2–64 — 39 ntl, gnt, tob gnt ami Table 9. (continued) Aminoglycosidea gnt ami, kan* Extraction and protein precipitation Cleanup Chromatography derivatizationb Intraday CV, % Interday CV, % r Determination Slope of Linear range, limit, µg/mL calibration curve µg/mL Recovery, % — Silica IP, OPA 5.13–11.66 6.49– 19.76 0.991– 0.996 0.02 (gnt C1) 0.08 (gnt C1a and gnt C2) 0.311 (gnt C1) 0.131 (gnt C1a) 0.129 (gnt C2) 1–20 — 71 2M tris Bond Elut C18 RP, TNBS 3.5–6.0 2.8–3.1 — <0.5 — 2.5–50 92.8–98.4 (ami) 41 Ref. 2M tris Bond Elut C18 RP, TNBS 4.0–4.9 4.6–5.1 — — — 1–25 94–98.6 (tob) 42 ntl ACN + fos, NaOH DCM RP, DansCl 1.49–2.02 — 0.999 0.5 0.563 0–20 — 43 ami tob, sis* Incubation Silica RP, OPA 5 5–6 — 1.0 — 1–15 — 44 ami, tob* — CM-Sephade x IP, OPA pc 3.2–25 (RSD) 3.9–28 (RSD) — — — 0.025–2 — 56 gnt, ami, tob nag* — CM-Sephade x IP, OPA pc 2.3 (tob)–4.0 (gnt) 2.8 (ami)–3.6 (gnt) 0.999 — 0.136 (ami) 0.278 (tob) 0.395 (gnt) 1.0–12 (gnt) 2.0–32 (ami) 20–15 (tob) 89 57 str, dhs* fos Sep-Pak C18 IP, — 2.07–5.46 — 0.999 2 0.0476 5–50 80 61 Drying and extraction on paper IP, OPA 7.5–9.0 (sis) 3.5–13.7 (ntl) 10.5 (sis) 16.4 (ntl) 0.999 (sis) 0.998 (ntl) 0.053 (sis) 0.5 (ntl) 0.334 (sis) 0.306 (ntl) 0.1–7.4 (sis) 1.0–10 (ntl) 93–105 (sis) 97–106 (ntl) 86 sis, ntl str 3.5% perchloric acid — IP, NQS 2.67–3.02 3.01–3.50 0.999 0.5 0.329 5–50 100 88 gnt MeOH — IP, OPA pc 2–2.5 2.3–3.2 0.999 0.5 0.792 2.7–16.5 97–103 62 gnt ACN + fos, NaOH DCM:fos IE, fluorescamine 2 3.5 0.999 1.0 2.4–3.8 0–40 93 77 tob, neo, sis, kan, ami — Amberlite IRC 50 IE, dihydrolutidine — — — 0.05 — 0–5 — 79 sis — CM-Sephade x RP, OPA 5.5–6.8 7.2–8.0 0.995 0.08 1.571 — 111 87 a b Compounds marked with * are internal standards. IP = ion-pair chromatography; TMSI = trimethylsilylimidazole; HFBl = heptafluorobutyric imidazole; RP = reversed-phase chromatography; BSC = benzenesulfonyl chloride; DansCl = dansyl chloride; FDNB = 1-fluoro-2,4-dinitrobenzene; FMOC = 9-fluorenylmethoxycarbonyl chloride; TNBS = 2,4,6-trinitrobenzenesulfonic acid; NQS = $-naphthoquinone-4-sulfonate; OPA = o-phthalaldehyde; pc = postcolumn derivatization. 1040 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 No single method can be identified as superior on the basis of method recoveries. Methods using acetonitrile and tris buffer for extraction and protein precipitation generally give 10% lower recoveries than other methods. Comparisons are difficult, however, because recoveries are not reported in about half of the methods reviewed here. Most of the reported methods cover the therapeutic concentration range of the aminoglycosides within their linear range. Limits of determination are generally below therapeutic concentrations. Urine Analysis of aminoglycosides in urine generally does not require complex sample preparation such as protein precipitation. Several methods used for aminoglycoside analysis in plasma could also be applied to urine as such as has been done with gentamicin extraction using acetonitrile and phosphate buffer (28). Two methods concerning analysis of gentamicin in urine have been reported. One uses only derivatization for sample preparation (77), and the other is based on SPE (31). The SPE column is Sep-Pak C18, and gentamicin is eluted with methanol containing ammonia. The eluent is evaporated to dryness to remove ammonia, and gentamicin is derivatized with OPA. The method is linear from 0.5 to 5 mg/L, and recovery is >85%. The advantages of SPE over liquid–liquid extractions are cleaner chromatograms and prolonged analytical column life (31). The sensitivity of the method can be improved easily by increasing the sample amount (31). In the direct method, phosphate buffer (pH 7.35) and acetone are added to the urine, and the solution is derivatized with fluorescamine (77). The calibration curve is linear from 0 to 71 mg/L, and the method is repeatable. Analysis of neomycin in urine has been done by centrifugation of samples and addition of counterion concentrate to the supernatant for chromatography. The supernatant is then analyzed by IP with OPA postcolumn derivatization (25). Recovery was 75% for a neomycin concentration of 1 µg/mL and 104% for a neomycin concentration of 5 µg/mL. Muscle Tissue, Kidneys, and Liver Neomycin, gentamicin, streptomycin, and dihydrostreptomycin are widely used as veterinary therapeutics (95). The illegal or irresponsible use of these drugs can cause residues in edible tissues (17). Characteristic to aminoglycoside pharmacokinetics is their tissue binding, especially to renal cortex (65). Therefore, aminoglycoside residues in food products of animal origin are not uncommon, and continuous monitoring of these residues is necessary (54). Monitoring of aminoglycoside residues requires confirmative methods (95). Sample preparation for residue analysis usually includes 2 stages: (1) mixing tissue in a buffer or in a protein-precipitating agent using organic solvent, precipitator, or alkaline or enzymatic digestion and (2) extraction of analytes by liquid–liquid, liquid–solid, or immunoaffinity methods (5). The methods used for sample preparation and extration of aminoglycosides in tissues are presented in Table 10. Matrix solid-phase dispersion (MSPD) has been used to extract aminoglycosides from kidneys (17). The tissue homogenate is mixed with cyanopropyl packing material, and the mixture is poured into a cartridge and compacted to a desired volume. During analysis of aminoglycosides, the stationary phase is washed with hexane, water, and methanol, and then the analytes are eluted with water. The method has been validated with fortified tissue samples because matrix components affect the shape and retention times of chromatographic peaks. Coeluting matrix components affect the MS performance. MS–PD also has been used for streptomycin and dihydrostreptomycin (49). Neomycin has been extracted from kidneys, liver, and muscle tissue with phosphate buffer for homogenization and protein precipitation (53, 54). Recovery and the effect of protein precipitation on recovery have been investigated with radioactively labeled neomycin, and recovery has been compared with that achieved from extractions with saline (54). Neomycin recovery from kidneys is 60% with saline, and drops to 30% after incubation and protein precipitation. With phosphate buffer, incubation does not affect recovery. However, neomycin becomes adsorbed to glass, and incubation and extraction must be performed in plastic. Gentamicin recovery has been investigated with different protein-precipitating buffers and CM-Sephadex extraction (5). With the precipitation agents HClO4, TCA, and acidic phosphate, recoveries are 78.5, 85.9, and 39.2%, respectively (5). TCA has been chosen as precipitation buffer, with addition of EDTA, because gentamicin has been reported to chelate positively charged metal cations (98). No gentamicin precipitation is observed with this technique. The effect of phosphate buffers on gentamicin recovery from edible tissues also has been studied (67). Results are contradictory to ones reported earlier because only basic buffers are found to extract gentamicin. When the pH of the phosphate buffer is 4.5, gentamicin is not extracted at all. When the pH is 8–10, recoveries vary between 70 and 98%. Addition of Na2SO4 to the precipitation buffer increases gentamicin recovery. However, the effect decreases as pH increases. Simultaneously, too high a sulfate concentration decreases recovery. Best results are achieved with 0.1M phosphate buffer when sulfate concentration is 0.1M and pH is 8.8. Recovery is then 98%. Separation of gentamicin from matrix components is difficult because of the similarity of chemical properties (65). Most of the methods used for sample preparation use only one retention mechanism, and that usually is not satisfactory for cleanup of aminoglycosides from tissue samples. For example, when phosphate buffer precipitation and Sephadex extraction are used, some interfering elements elute with gentamicin, making it impossible to quantitate all 3 components (67). The applicability for gentamicin extraction of SPE materials with TCA as precipitation buffer has been studied (5). Different nonpolar groups, such as C8, phenyl, and C18, bound to silica have been studied. Good recoveries are achieved with all of these (55–81%), but specificity to gentamicin is not sat- ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 1041 Table 10. Sample preparation methods for aminoglycosides from edible tissues Analyte neo Chromatography and derivatizationa Extraction and precipitationb Cleanupc Recovery, % Ref. IP, OPA pc fos 0.2M pH 8, incubation 100EC 10 min — 80 53 neo IP, OPA pc fos incubation 100EC 5 min — 90 54 gnt IP, OPA pc 5% TCA, EDTA 1 mM IE, CM-Sephadex 68–98 5 IP, — MSPD, cyanopropyl — 46–75 17 d gnt, neo, str, dhs e IP, OPA 0.1M fos, 0.1M Na2SO4, pH 8.8 IE, C18COOH 90 65 str, dhs IP, — MSPD, cyanopropyl 0.1M Na2SO4 — — 49 gnt IP, OPA fos incubation 100EC 5 min IE, CM-Sephadex 83–101 67 str, dhs IP, NQS 3.6% perchloric acid IE, (Bakerbond) aromatic sulfonic acid 53–66 89 GC, HFBI/TMSI 10% TCA Amberlite CG 50 74–83 24 gnt kan a b c d e For definitions, see footnotes to Table 4. MSPD = matrix solid-phase dispersion. IE = ion-exchange chromatography. Paromomycin as internal standard. Netilmicin as internal standard. isfactory. When silica, weak cation exchanger (WCX), or polymeric resins are used, recovery is not sufficient. Thus CM-Sephadex, which is specific to gentamicin and gives recoveries of 65% or more, has been chosen for analysis (5). Others have reported good results for cleanup of gentamicin samples with WCXs (65). The method based on WCX–SPE can be used for all tissues including liver, kidneys, and lungs (65). Precipitation of proteins and extraction of streptomycin and dihydrostreptomycin from kidney and muscle tissues with 3.6% perchloric acid has been done (89). Recoveries of the drugs vary depending on tissues. For swine muscle, the lowest dihydrostreptomycin recovery is obtained when recovery of streptomycin is highest. Adding perchloric acid during extraction does not increase recovery, but it increases interferences in chromatography. Milk Milk and milk product matrixes contain several compounds that might interfere in analysis (83). Analytical methods for antibiotics in milk include cleanups to remove proteins and lipophilic compounds. Gentamicin (65, 68), neomycin (45, 51), streptomycin, and dihydrostreptomycin (60) in milk have been analyzed. An Amberlite CG-50 ion-exchange column has been used to extract neomycin from milk when no protein precipitation is used (45). Neomycin is derivatized on column with OPA, and the derivatives are eluted with basic methanol. Recovery of the method is 94–102% at concentrations of 0.1–5.0 µg/mL. The limit of determination is 0.05 µg/mL. In another study, neomycin is extracted from milk by protein precipitation with 20% TCA (51). OPA postcolumn derivatization and fluorescence detection are used. Recovery of neomycin is 94–122%, and the method is linear in range 0.15–12 µg/mL. CM-Sephadex C-25 ion exchanger has been used to extract gentamicin from milk. Milk is added to the column in NaOH solution (68). After elution from the column, gentamicin is derivatized in a silica Sep-Pak SPE column with OPA. Attempts to precipitate the milk proteins before Sephadex extraction have been unsuccessful. However, the method is quantitative and selective, and recoveries are 94–106% at gentamicin concentrations of 0.2–10 µg/mL. A C18 COOH ion-exchange column also has been used to extract gentamicin from milk at concentrations of 0.625–80 µg/mL. The limit of determination is 0.6 µg/mL (65). In another study, gentamicin is extracted from milk with 30% TCA and a C18 SPE column (52). Gentamicin is eluted from the SPE cartridge with ammonia–methanol. Recovery is 72–88%, and the limit of determination for individual components is 0.4 ng/mL. Streptomycin and dihydrostreptomycin have been extracted from milk with 3.6% perchloric acid, followed by IP and postcolumn derivatization with NQS (60). Recoveries were 32.6–65.0%, depending on milk fat content. Comparison and Evaluation of Methods Chromatographic, Microbiological, and Immunological Methods Microbiological and immunological methods are still popular for aminoglycoside analysis and are used in both therapeutic monitoring (99) and screening for aminoglycoside residues (9). The reliability of results obtained with these methods is, however, poorer than can be achieved with chromatographic methods. The slowness of microbiological methods is a major disadvantage, particularly in therapeutic monitoring where the speed of analysis is important. Immunoassays provide high throughput of samples but often give erroneous results (99). The advantages of chromatographic methods over microbiological or immunological methods include greater 1042 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 accuracy, repeatability, and specificity of results. However, when antimicrobial activity is of concern, only bioassays can directly measure this property. Immunoassays based on enzyme-linked immunosorbent assay (ELISA) or RIA techniques generally measure cut-off points ranking result as positive or negative with reference to a concentration limit. Maximum residue limits (MRLs) for aminoglycoside vary widely among different aminoglycosides and different tissues. Therefore, it appears extensively difficult to produce a general assay that would monitor these drugs as a group relative to a regulatory tolerance level. The structural similarity of the different aminoglycosides is likely to cause cross-reactivity between compounds, complicating quantitative determination. Aminoglycoside bioassays and immunological methods usually use the aqueous extract of most matrixes without need for further concentration. This lack of need for further concentration is an advantage for compounds that are difficult to extract with organic solvents. However, bioassays do not detect compounds bound to proteins, which makes results difficult to quantitate. For example, in determination of neomycin concentrations in incurred tissue samples, the microbial assay shows wider variation in results than HPLC as a function of standards prepared in buffer or tissue homogenate (99). Purity or ratio of drug components in drug preparations cannot be determined microbiologically or immunologically. Thus, chromatographic methods are necessary. Furthermore, when the importance of judicial and economic consequences of results increases, the reliability and accuracy of results become more significant. When linearity, accuracy, precision, reliability, or diversity of the method are considered, HPLC is at present the best method for aminoglycoside analysis. Separation capacity and low costs also contribute to HPLC as the method of choice. An additional advantage of chromatographic methods is the possibility to use internal standards, which often increases reliability. Usually, microbiological methods are compared with chromatography by the so-called blind-duplicate method. The purpose is to evaluate the correlation of results obtained by the 2 methods. Usually chromatographic results are presented as a function of microbiological results, and the slope and r of the plot are indicative of the correspondence. Both of the mentioned values should be as close as possible to 1. The results from the use of this method are presented in Table 11. Correlations of chromatographic methods with enzymological, immunological, or microbiological methods differ depending on the aminoglycosides and the chromatographic methods (27, 30–33, 36, 38, 41–44, 61, 62, 88, 100). For gentamicin anal ysis, for example, the best correlation is between reversed-phase LC method using FDNB derivatization (31) and RIA. IP with OPA postcolumn derivatization (62) correlates better with microbiological assays than a precolumn derivatization method (32) with RP. Interestingly, both methods separate the 3 gentamicin components and the quantitative differences could not be explained by different amounts of the separated components. It appears that quantitation suffers when OPA postcolumn derivatization is changed to precolumn derivatization. Table 11. Comparison of the microbiological, immunological, and chromatographic methods used for aminoglycoside analysis by blind duplicate method Aminoglycosidea ami gnt ntl Derivatization and chromatographya Reference methodb Slope r Reference TNBS, RP RIA 1.047 0.999 41 FDNB, RP MB 0.92 0.993 38 OPA, RP MB 0.94 0.94 44 BSC, RP EMIT 0.86 0.995 27 OPA, RP MB 0.87 0.99 32 OPA pc, IP MB 1.02 0.934 62 FDNB, RP RIA 1.002 0.9996 31 OPA, IP RIA 1.08 0.98 33 OPA, IP MB 1.13 0.99 33 DansCl, RP MB 1.048 0.949 43 sis FDNB, RP MB 0.91 0.99 30 str —, IP FPIA 0.931 0.969 61 NQS, IP FPIA 1.003 0.99 88 tob TNBS, RP RIA 0.971 0.968 42 TNBS, RP EMIT 0.981 0.971 42 FDNB, RP MB 1.04 0.997 36 a b For definitions, see footnotes to Table 4. RIA = radioimmunoassay; MB = microbiological assay; EMIT = enzyme immunoassay technique; FPIA = fluorescence polarization immunoassay. ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 1043 Between the microbiological method and chromatographic methods using OPA precolumn derivatization for analysis of amikacin (44), gentamicin (32), and netilmicin (33), the best agreement is for amikacin (Table 11). For netilmicin, the agreement between RIA and the same chromatographic method (33) is only marginally better than between the microbiological and chromatographic methods. The microbiological and RIA methods appear to correlate well. Results of RP with FDNB derivatization are in good agreement with reference methods for gentamicin (30) and tobramycin (36) and slightly less for amikacin (38) and sisomicin (30; Table 11). The best correlations for tobramycin and amikacin are obtained with reversed-phase methods with TNBS derivatization (41, 42). For streptomycin, NQS postcolumn derivatization after IP (88) has better agreement with fluorescence polarization immunoassay (FPIA) than does the reversed-phase method with low-wavelength UV detection (61). Chromatographic Methods Most chromatographic determinations of aminoglycosides are done by HPLC, and the aminoglycosides are derivatized to improve detection. The choice of derivatization reagent has a remarkable effect on quantitative and qualitative results. In general, IP is used with OPA derivatization, MS, or electrochemical detection. When MS, electrochemical detection, or postcolumn derivatization with OPA is used, IP as the method of choice is justified because the analyzed aminoglycosides are charged. However, when OPA is used for derivatization prior to chromatography, it is not clear why IP is the method of choice for separation. It is unclear if retention occurs through an ion-pairing mechanism. Furthermore, when mercaptoacetic acid (11, 63–66) or mercaptopropionic (86, 87) acid is used with OPA, the derivatives include negatively charged groups that could complicate ion-pair formation with sulfonates and facilitate ion-pair formation with positively charged counterions. Unfortunately, this possibility is not discussed. A good example of the lack of clarity associated with IP and OPA precolumn derivatization is the analysis of sisomicin by IP with heptane sulfonate as counterion and precolumn derivatization with OPA (86). Sisomicin is separated with the original method but when heptane sulfonate is replaced with EDTA, the separation improves (87). This result suggests that separation by the original method does not occur by an ion-pair mechanism. On the basis of promising results so far, FMOC–Cl will find more applications in aminoglycoside analysis. The methods based on TNBS and FDNB, which work for most aminoglycosides except for streptomycin and dihydrostreptomycin, will probably get more attention, too. Comparison of FDNB and TNBS with OPA as derivatization agents does not reveal marked differences in limits of determination even though OPA derivatization allows fluorescence detection. Incomplete derivatization of aminoglycosides with OPA may, at least partly, explain this phenomenon. In therapeutic monitoring and in the analysis of drug preparations, the chromatographic methods using derivatization and IP or RP chromatographic separation will remain dominant be- cause of their simplicity, short analysis time, and low cost. In laboratories where confirmatory methods are needed, LC/MS applications can be expected to replace other methods. In these cases, IP with the use of fluorinated carboxylic acids such as TFA, PFPA, and HFBA as counterions will probably become the methods of choice. The advantage of these methods is that the retention of aminoglycosides with these counterions has been thoroughly studied (69, 70), and separation of several aminoglycosides can be achieved in one run (15, 69, 70). Conclusions Aminoglycosides are a heterogenic group of antibiotics characterized by a wide spectrum of activity and toxicity. Because of their toxicity, the monitoring of their concentrations in plasma during therapy is necessary. The wide use of aminoglycosides in veterinary medicine also requires suitable routine and confirmatory methods for their detection in edible tissues. In both cases, the methods must be accurate, sensitive, and robust against interferences. However, the chemical characteristics of aminoglycosides such as polarity, water solubility, lack of volatility, and lack of chromophore make development of applicable methods difficult. Chromatographic analysis almost always requires derivatization to improve either detectability or separation. However, derivatization is difficult when the aminoglycoside structure incorporates several groups of different reactivity, leading to possible partial derivatization. The most popular method for aminoglycoside analysis is IP chromatography with OPA derivatization. The interpretation of the results is, however, unclear and contradictory, and the reliability and reproducibility of results is questionable. The most promising method seems to be LC/MS in combination with IP. The combination permits both quantitative and confirmatory analysis for most aminoglycosides. For the analysis and extraction of aminoglycosides from different tissues, no generally applicable method has been reported. Extraction from edible tissues appears extremely difficult. However, promising results have been achieved with methods using SPE. More work is needed to develop generally applicable and validated regulatory methods. Acknowledgment We thank James D. MacNeil for his constructive criticism during preparation of this review. References (1) Ullman’s Encyclopedia of Industrial Chemistry (1985) 5th Ed., Graphischer Betrieb Konrad Triltsch, Wurtzburg, Germany, Vol. A2, p. 467 (2) Zaske, D.E. (1992) in Applied Pharmacokinetics, Principles of Therapeutic Drug Monitoring, 3rd Ed., W.E. Evans, J.J. Schentag, W.J. Jusko, & M.V. Relling (Eds), Edwards Brothers, Ann Arbor, MI, Chapter 14 (3) Shaikh, B., & Allen, E. (1985) J. Assoc. Off. Anal. Chem. 68, 1007 1044 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 (4) Chambers, H.F., & Sande, M.A. (1996) in Goodman and Gilman’s The Pharmacological Basis of Therapeutics, 9th Ed., J.G. Herdman, L.E. Limbird, P.B. Molinoff, & R.W. Ruddon (Eds), McGraw-Hill, New York, NY, p. 1103 (5) Sar, F., Leroy, P., Nicolas, A., & Archimbault, P. (1993) Anal. Chim. Acta 275, 285 (6) Kawaguchi, H. (1976) J. Infect. Dis. 134, 242 (7) Kabins, S.A. (1976) Antimicrob. Agents Chemother. 10, 139 (8) Adams, E., Schepers, R., Roets, E., & Hoogmartens, J. (1996) J. Chromatogr. 741, 233 (9) MacNeil, J.D., & Cuerpo, L. (1995) in Residues of Some Veterinary Drugs in Animals and Foods, FAO, Food and Nutrition Paper 41/7, Rome, Italy, p. 45 (10) Thomas, A.H., & Tappin, S.D. (1974) J. Chromatogr. 97, 280 (11) Claes, P.J., Busson, R., & Vanderhaeghe, H. (1984) J. Chromatogr. 298, 445 (12) Maitra, S.K., Yoshikawa, T.T., Guze, L.B., & Schotz, M.C. (1979) Clin. Chem. 25, 1361 (13) Anhalt, J.P., Sanclio, F.D., & McCorcle, T. (1978) J. Chromatogr. 153, 489 (14) Kawamoto, T., Mashimo, I., Yamauchi, S., & Watanabe, M. (1984) J. Chromatogr. 305, 373 (15) Inchauspé, G., & Samain, D. (1984) J. Chromatogr. 303, 277 (16) Getek, T.A., Vestal, M.L., & Alexander, T.G. (1991) J. Chromatogr. 554, 191 (17) McLaughlin, L.G., Henion, J.P., & Kijak, P.J. (1994) Biol. Mass Spectrom. 23, 417 (18) Margosis, M., & Tsuji, K. (1973) J. Pharm. Sci. 62, 1837 (19) Van Giessen, B., & Tsuji, K. (1971) J. Pharm. Sci. 60, 1068 (20) Tsuji, K., & Robertson, J.H. (1969) Anal. Chem. 41, 1332 (21) Tsuji, K., & Robertson, J.H. (1970) Anal. Chem. 42, 1661 (22) Mayhew, J.W., & Gorbach, S.L. (1978) Antimicrob. Agents Chemother. 14, 851 (23) Mayhew, J.W., & Gorbach, S.L. (1978) J. Chromatogr. 151, 133 (24) Nakaya, K., Sugitani, A., & Yamada, F. (1985) Shokuhin Eiseigaku Zasshi 26, 443; Chem. Abstr. (1996) 104, 147272q (25) Shaikh, B., Jackson, J., Guyer, G., & Ravis, W.R. (1991) J. Chromatogr. 571, 189 (26) Essers, L. (1984) J. Chromatogr. 305, 345 (27) Larsen, N.-E., Marinelli, K., & Heilesen, A.M. (1980) J. Chromatogr. 221, 182 (28) Peng, G.W., Gadalla, M.A.F., Peng, A., Smith, V., & Chiou, W.L. (1977) Clin. Chem. 23, 1838 (29) Barends, D.M., Van Der Sandt, J.S.F., & Hulshoff, A. (1980) J. Chromatogr. 182, 201 (30) Barends, D.M., Zwaan, C.L., & Hulshoff, A. (1981) J. Chromatogr. 222, 316 (31) D’Souza, J., & Ogilvie, R.I. (1982) J. Chromatogr. 232, 212 (32) Maitra, S.K., Yoshikawa, T.T., Hansen, J.L., Nilsson-Ehle, I., Palin, W.J., Schotz, M.C., & Guze, L.B. (1977) Clin. Chem. 23, 2275 (33) Bäck, S.E., Nilsson-Ehle, I., & Nilsson-Ehle, P. (1979) Clin. Chem. 25, 1222 (34) Stead., D.A., & Richards, R.M.E. (1996) J. Chromatogr. 675, 295 (35) Elrod, L., Jr, White, L.B., Spanton, S.G., Stroz, D.G., Cugier, P.J., & Luka, L.A. (1984) Anal. Chem. 56, 1786 (36) Barends, D.M., Zwaan, C.L., & Hulshoff, A. (1981) J. Chromatogr. 225, 417 (37) Barends, D.M., Blauw, J.S., Mijnsbergen, C.W., Govers, L.R., & Hulshoff, A. (1985) J. Chromatogr. 322, 321 (38) Barends, D.M., Blauw, J.S., Smits, M.H., & Hulshoff, A. (1983) J. Chromatogr. 276, 385 (39) Wong, L.T., Beaubien, A.R., & Pakuts, A.P. (1982) J. Chromatogr. 231, 145 (40) Gambardella, P., Punziano, R., Gionti, M., Guadalupi, C., Mancini, G., & Mangia, A. (1985) J. Chromatogr. 348, 229 (41) Kabra, P.M., Bhatnager, P.K., & Nelson, M.A. (1984) J. Chromatogr. 307, 224 (42) Kabra, P.M., Bhatnager, P.K., Nelson, M.A., Wall, J.H., & Marton, L.J. (1983) Clin. Chem. 29, 672 (43) Peng, G.W., Jackson, G.G., & Chiou, W.L. (1977) Antimicrob. Agents Chemother. 12, 707 (44) Maitra, S.K., Yoshikawa, T.T., Steyn, C.M., Guze, L.B., & Schotz, M.C. (1978) Antimicrob. Agents Chemother. 14, 880 (45) Agarwal, V.K. (1990) J. Liquid Chromatogr. 13, 2475 (46) Games, D.E., McDowall, M.A., Levsen, K., Schafer, K.H., Dobberstein, P., & Gower, J.L. (1984) Biomed. Mass Spectrom. 11, 87 (47) Kato, Y., Tkahashi, S., Hirose, H., Sakairi, M., & Kambara, H. (1988) Biomed. Environmental Mass Spectrom. 16, 331 (48) Sakairi, M., & Kambara, H. (1988) Anal. Chem. 60, 774 (49) McLaughlin, L.G., & Henion, J.D. (1992) J. Chromatogr. 591, 195 (50) Anhalt, J.P. (1977) Antimicrob. Agents Chemother. 11, 651 (51) Shaikh, B., & Jackson, J. (1989) J. Liq. Chromatogr. 12, 1497 (52) Kijak, P.J., Jackson, J., & Shaikh, B. (1997) J. Chromatogr. 691, 377 (53) Shaikh, B., & Jackson, J. (1993) J. AOAC Int. 76, 543 (54) Shaikh, B., Allen, E.H., & Gridley, J.C. (1985) J. Assoc. Off. Anal. Chem. 68, 25 (55) Getek, T.A., Haneke, C., & Selzer, G.B. (1983) J. Assoc. Off. Anal. Chem. 66, 172 (56) Wichert, B., Schereier, H., & Derendorf, H. (1991) J. Pharm. Biomed. Anal. 9, 251 (57) Anhalt, J.P., & Brown, S.D. (1978) Clin. Chem. 24, 1940 (58) Whall, T.J. (1981) J. Chromatogr. 219, 89 (59) Weigand, R., & Coombes, R.J. (1983) J. Chromatogr. 281, 381 (60) Gerhardt, G.C., Salisbury, C.D.C., & MacNeil, J.D. (1994) J. AOAC Int. 77, 765 (61) Kurosawa, N., Kuribayashi, S., Owada, E., Ito, K., Nioka, M., Arakawa, M., & Fukuda, R. (1985) J. Chromatogr. 343, 379 (62) Kubo, H., Kinoshita, T., Kobayashi, Y., & Tokunaga, K. (1982) J. Chromatogr. 227, 244 (63) Chissell, J.F., Freeman, M., Loran, J.S., Sewell, G.J., & Stead, J.A. (1986) J. Chromatogr. 369, 213 (64) Albracht, J.H., & De Wit, M.S. (1987) J. Chromatogr. 389, 306 (65) Fennell, M.A., Uboh, C.E., Sweeney, R.W., & Soma, L.R. (1995) J. Agric. Food Chem. 43, 1849 (66) Freeman, M., Hawkins, P.A., Loran, J.S., & Stead, J.A. (1979) J. Liquid Chromatogr. 2, 1305 (67) Agarwal, V.K. (1989) J. Liquid Chromatogr. 12, 613 (68) Agarwal, V.K. (1989) J. Liquid Chromatogr. 12, 3265 ISOHERRANEN & SOBACK: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 5, 1999 1045 (69) Inchauspé, G., Delrieu, P., Dupin, P., Laurent, M., & Samain, S. (1987) J. Chromatogr. 404, 53 (70) DeMiguel, I., Puech-Costes, E., & Samain, D. (1987) J. Chromatogr. 407, 109 (71) Rumble, R.H., & Roberts, M.S. (1987) J. Chromatogr. 419, 408 (72) Inouye, S., & Ogawa, H. (1964) J. Chromatogr. 13, 536 (73) Maehr, H., & Schaffner, C.P. (1965) Anal. Chem. 36, 104 (74) DeRossi, P. (1975) Analyst 100, 25 (75) Statler, J.A. (1990) J. Chromatogr. 527, 244 (76) Polta, J.A., Johnson, D.C., & Merkel, K.E. (1985) J. Chromatogr. 324, 407 (77) Walker, S.E., & Coates, P.E. (1981) J. Chromatogr. 223, 131 (78) Mays, D.L., Van Apeldoorn, R.J., & Lauback, R.G. (1976) J. Chromatogr. 120, 93 (79) Csiba, A. (1979) J. Pharm. Pharmacol. 31, 115 (80) Helboe, P., & Kryger, S. (1982) J. Chromatogr. 235, 215 (81) Tsuji, K., & Jenkins, K.M. (1986) J. Chromatogr. 369, 105 (82) Tsuji, K.T., Goetz, J.F., VanMeter, W., & Gusciora, K.A. (1979) J. Chromatogr. 175, 141 (83) Bobbitt, D.R., & Ng, K.W. (1992) J. Chromatogr. 624, 153 (84) Rouan, M.C. (1985) J. Chromatogr. 340, 361 (85) Roth, M. (1971) Anal. Chem. 43, 880 (86) Tawa, R., Hirose, S., & Fujimoto, T. (1989) J. Chromatogr. 490, 125 (87) Tawa, R., Koshide, K., Hirose, S., & Fujimoto, T. (1988) J. Chromatogr. 425, 143 (88) Kubo, H., Kobayashi, Y., & Kinoshita, T. (1986) Anal. Chem. 58, 2653 (89) Gerhardt, G.C., Salisbury, C.D.C., & MacNeil, J.D. (1994) J. AOAC Int. 77, 334 (90) Udenfriend, S., Stein, S., Böhlen, P., Dairman, W., Leimgruber, W., & Weigele, M. (1972) Science 178, 871 (91) Chiou, W.L., Nation, R.L., Peng, G.W., & Huang, S.-M. (1978) Clin. Chem. 24, 1846 (92) Snyder, L.R., & Kirkland, J.J. (1979) Introduction to Modern Liquid Chromatography, 2nd Ed., John Wiley, New York, NY (93) Tsuji, K., & Jenkins, K. (1985) J. Chromatogr. 333, 365 (94) Brumley, W.C., & Sphon, J.A. (1981) Biomed. Mass Spectrom. 8, 390 (95) Khan, A.H., Shaikh, B., Allen, E.H., & Sokolski, E.A. (1988) Biomed. Environmental Mass Spectrom. 17, 329 (96) Takeda, N., Harada, K., Suzuki, M., Tatematsu, A., & Kubodera, T. (1982) Org. Mass Spectrom. 17, 247 (97) Marshall, W.J. (1988) Illustrated Textbook of Clinical Chemistry, 1st Ed., Gower Medical Publishing, London, UK, p. 285 (98) Reynolds, J.E.F. (Ed.) (1989) Martindale The Extra Pharmacopoeia, 29th Ed., Pharmaceutical Press, London, UK, p. 237 (99) White, L.O. (1998) Ther. Drug Monit. 20, 464 (100) Shaikh, B., Jackson, J., & Thaker, N. (1995) High J. Vet. Pharmacol. Therap. 18, 150 (101) The Merck Index (1983) 10th Ed., Merck, Rahway, NJ (102) Rosano, T.G., Brown, H.H., Meola, J.M., & McDermontt, C. (1979) Clin. Chem. 25, 1064