Anal. Chem. 2002, 74, 3160-3167
Quantitative Analysis of Mitoxantrone by Surface-Enhanced Resonance Raman Scattering Clare McLaughlin,† Donald MacMillan,‡ Colin McCardle,‡ and W. Ewen Smith*,†
Department of Pure and Applied Chemistry, Strathclyde University, Glasgow G1 1XL, Scotland, and Department of Surgery, Glasgow Royal Infirmary, Glasgow G4, Scotland
Mitoxantrone is an anticancer agent for which it is important to know the concentration in blood during therapy. Current methods of analysis are cumbersome, requiring a pretreatment stage. A method based on surface-enhanced resonance Raman scattering (SERRS) has been developed using a flow cell and silver colloid as the SERRS substrate. It is simple, sensitive, fast, and reliable. Both blood plasma and serum can be analyzed directly, but fresh serum is preferred here due to reduced fluorescence in the clinical samples available. Fluorescence is reduced further by the dilution of the serum in the flow cell and by quenching by the silver of surfaceadsorbed material. The effectiveness of the latter process is dependent on the contact time between the serum and the silver. The linear range encompasses the range of concentrations detected previously in patient samples using HPLC methods. In a comparative study of a series of samples taken from a patient at different times, there is good agreement between the results obtained by HPLC and SERRS with no significant difference between them at the 95% limit. The limit of detection in serum using the final optimized procedure for SERRS was 4.0 × 10-11 M (0.02 ng/mL) mitoxantrone. The ease with which the SERRS analysis can be carried out makes it the preferred choice of technique for mitoxantrone analysis Mitoxantrone (MX) (1,4-dihydroxy-5,8-bis[[2-(2-hydroxyethyl)amino]ethylamino-9,10-anthracenedione dihydrochloride] is a synthetic chemotherapeutic agent with proven antitumor activity against breast cancer,1 ovarian cancer,2 non-Hodgkin’s lymphoma,3 and acute leukemia.4 The structure of mitoxantrone is shown in Figure 1. To increase the local concentration of mitoxantrone at the tumor site, intraarterial and intraperitoneal infusion regimes rather than systemic administration are often used since they reduce systemic overkill. In these therapies there is a need for a sensitive assay to monitor drug concentrations in blood plasma or serum. †
Strathclyde University. Glasgow Royal Infirmary. (1) Neidhart, J. A.; Gochnour, D.; Roach, R. A. J. Clin. Oncol. 1986, 4, 672677. (2) Lawton, F.; Blackledge, G.; Mould, J. Cancer Treat. Rep. 1987, 71, 627629. (3) Coltman, C. A.; McDaniel, T. M.; Balcerzak, S. P. Invest. New Drugs 1983, 1, 65. (4) Birot-Barbapulle, F.; Catovsky, D.; Slocumbe, G. Cancer Treat. Rep. 1987, 71, 161-163. ‡
3160 Analytical Chemistry, Vol. 74, No. 13, July 1, 2002
Figure 1. Structure of mitoxantrone.
Determination of mitoxantrone in body fluids and tissues has been reported using electroanalytical5-11 and immunoassay methods,12,13 but reversed-phase HPLC is the most widely used method.14-36 Detection is usually by a spectrophotometric method (5) Golabi, S. M.; Hassan-Zadeh, V. Talanta 1996, 43, 397. (6) Wirth, M.; Gabor, F.; Schalkhammer, T.; Pittner, F. Mikrochim. Acta 1995, 121, 87. (7) Del Pozo, J. A.; Costa-Garcia, A.; Tunon-Blanco, P. Anal. Chim. Acta 1994, 289, 169. (8) Villar, J. C. C.; Garcia, A. C.; Blanco, P. T. Talanta 1993, 40 (3), 325. (9) Villar, J. C. C.; Garcia, A. C.; Blanco, P. T. Anal. Chim. Acta 1992, 256, 231. (10) Villar, J. C. C.; Garcia, A. C.; Blanco, P. T. J. Pharm. Biol. Anal. 1992, 10 (4), 263. (11) Wang, H.; Hua, E.; Yang, P. Talanta 1995, 42, 1519. (12) Flavell, S. U.; Flavell, D. J. J. Immunol. Methods 1988, 115, 179. (13) Nicolau, G.; Szucs-Myers, V.; Morrison, J.; Lanzoilotti, A. Invest. New Drugs 1985, 51, 3. (14) Greidanus, J.; De Vries, E. G. E.; Mulder, N. H.; Sleijfer, D. Th.; Uges, D. R. A.; Oosterhuis, B.; Willemese, J. Clin. Oncol. 1989, 7, 790, (15) Payet, B.; Arnoux, J.; Catalin, J.; Cano, J. P.; J. Chromatogr. 1988, 424, 337-345. (16) Choi, K. E.; Sinkule, D. S.; Hans, D. S.; McGrath, S. C.; Daly, K. M.; Larson, R. A. J. Chromatogr. 1987, 420, 81. (17) Ehninger, G.; Proksch, B.; Schiller, E. J. Chromatogr. 1985, 342, 119. (18) Van Belle, S. J. P.; de Planque, M. M.; Smith, I. E.; Van Oosteron, A. T.; Schoemaker, T. J.; Denuve, W.; McVie, J. G. Cancer Chemother. Pharmacol. 1980, 18, 27. (19) Ostroy, F.; Gams, R. A. J. Liq. Chromatogr. 1980, 3, 637. (20) Launay, M. C.; Iliais, A.; Richard, B. J. Pharm. Sci. 1989, 78, 877. (21) Lin, K. T.; Rivard, G. E.; LeClerc, J. M. J. Chromatogr. 1989, 465, 75. (22) Chiccarelli, F. S.; Morrison, J. A.; Cosulich, D. B.; Perkinson, N. A.; Ridge, D. N.; Sum, F. W.; Murdock, K. C.; Woodward, D. L.; Arnold, E. T. Cancer Res. 1986, 46, 4858. (23) Reynolds, D. L.; Ulrich, K. K.; Patton, T. F. Int. J. Pharm. 1981, 9, 67. (24) Savaraj, N.; Lu, K.; Valdivieso, M.; Burgess, T. Clin. Pharm. Ther. 1982, 31, 312. (25) Roboz, J.; Paciucci, P. A.; Silides, D. Cancer Chemother. Pharmacol. 1984, 13, 67. (26) Alberts, D. S.; Peng, Y. M.; Leigh, S.; Davies, T. P.; Woodward, D. L. Cancer Res. 1985, 45, 1879. (27) Smythe, J. F.; Macpherson, J. S.; Warrington, C. F.; Wolf, C. R. Cancer Chemother. Pharmacol. 1986, 17, 149. (28) Reynolds, D. L.; Sternson, L.; Repta, A. J. J. Chromatogr. 1981, 222, 225. (29) Van Belle, S. J. P.; Schoemaker, T. J.; Verwey, S. L.; Paalman, A. C. A.; McVie, J. G. J. Chromatogr. 1985, 337, 73. 10.1021/ac010067k CCC: $22.00
© 2002 American Chemical Society Published on Web 05/30/2002
at 254 nm or at 658 nm, although electrochemical14,16 and fluorescence methods31 have been reported. Extensive sample pretreatment is required, and techniques such as solid-phase extraction,14,21,22,26,27,34 liquid-liquid extraction,16,18,19,24,25,29,33 and adsorption onto glass wool22 or polymeric resin17,23,28 are employed. Despite the need for sample pretreatment, only some of the published methods detail the use of an internal standard. An optimum detection limit of 0.1 ng/mL and a linear range of 0.11000 ng/mL were reported using electrochemical detection.16 Typically, however, sensitivities are of the order of 1-10 ng/mL with linear ranges 1-3 orders of magnitude at best. There are difficulties in the determination of mitoxantrone in body fluids using HPLC, including the need for sample pretreatment prior to analysis and therefore the need to use internal standards. In addition, there are problems with adsorption of the drug onto glass,19,21 nonspecific drug binding, reactivity of the drug with solvents, and instability of the drug in plasma.27,28 Overall, although mitoxantrone can be determined in clinical samples, the methods are cumbersome and prone to error if incorrectly applied. Surface-enhanced resonance Raman scattering (SERRS) has unique advantages for the development of a simple, fast, and reliable assay for mitoxantrone. In this method, an analyte containing a chromophore is adsorbed on a roughened surface of a suitable metal such as silver and a laser frequency is chosen so that it is in resonance or preresonance with the chromophore.37,38 The Raman scattering efficiency is increased by surface enhancement due to interaction with the plasmons on the silver surface and by resonance from the chromophore. The advantages of SERRS are as follows: (1) It uses standard Raman equipment, which recently has become reliable and simple to operate. (2) It is very sensitive with single molecule detection claimed. (3) Sharp molecularly characteristic signals are obtained from aqueous solution/suspension. (4) Fluorescence, which interferes with the detection of Raman scattering, is quenched from molecules directly adsorbed on the surface. One difficulty with using SERRS in that efficient surface adsorption of the analyte is necessary, but where this occurs naturally, as with mitoxantrone, there is an advantage, since it improves the selectivity in the assay and provides a method of concentration of the analyte if required. Further, mitoxantrone absorbs in the visible region so that resonance enhancement as well as surface enhancement is easily obtained with the correct choice of laser frequency. Therefore, by use of a flow cell to give reproducibility, it has been possible to develop an assay for mitoxantrone in serum or plasma using SERRS detection, which is simple and for which there is with little interference from the biological matrix. This avoids sample pretreatment and therefore overcomes some of the problems associated with HPLC analysis. (30) Czejka, M. J.; Georgapoulas, A. J. Chromatogr. 1988, 425, 429. (31) Bell, D. H. Biochim. Biophys. Acta 1988, 949, 132. (32) Priston, M. J.; Sewell, G. J. J. Pharma. Biomed. Anal. 1994, 12, 1153. (33) Rentsch, K. M.; Schewender, R. A.; Hanseler, E. J. Chromatogr., B 1996, 679, 185. (34) Law, S. L.; Jang, T. F. J. Chromatogr., A 1994, 670, 234. (35) Catalin, J.; Peloux, A. F.; Coloma, F.; Payet, B.; Lacarelle, B.; Cano, J. P. Biomed. Chromatogr. 1994, 8, 37. (36) Micelli, G.; Lozupone, A.; Quaranta, M.; Donadeo, A.; Coviello, M.; Lorusso, V. Biomed. Chromatogr. 1992, 6, 168. (37) Cotton, T. M.; Kim, J. H.; Chumanov, G. D. J. Raman Spectrosc. 1991, 22, 729. (38) Rodger, C.; Smith, W. E.; Dent, G.; Edmondson, M. J. Chem. Soc., Dalton Trans. 1996, 791-799.
Figure 2. Diagram of flow cell.
EXPERIMENTAL SECTION Chemicals and Reagents. Mitoxantrone solution (15 mg/ mL) was supplied by Lederle Laboratories. Silver nitrate (>99.999% w/w, Sigma), trisodium citrate (Aldrich), sodium chloride (BDH), methanol (BDH), and n-pentane (Fisons) were all analytical grade. Water for colloid preparation and preparation of all solutions was HPLC grade. Sigmacote (Sigma) was used for silanization of the glass capillaries. Silver Colloid. Silver colloid was prepared by citrate reduction of silver nitrate according to the Lee and Meisel method.39 Unless otherwise stated, the colloid was prepared in 500-mL volumes and diluted to 1 L with distilled water prior to use. This gives a suspension of silver particles with 45 mg of silver present per liter. Several batches of colloid were used. The electronic spectrum and pH of each was recorded. All spectra had an adsorption maximum of 400-402 nm and a full width half-height of 50-60 nm. Flow Cell. A schematic diagram of the flow cell used is shown in Figure 2. A multichannel peristaltic pump (Watson Marlow 3000) pumped silver colloid (A), sodium chloride (B), and mitoxantrone solution (C) through three separate channels. Manifold pump tubing of varying diameters (Elkay, Co., Galway, Eire) was used to give flow rates of 3.4 or 1.6 (A), 0.23 (B), and 0.32 or 0.23 mL/min (C). The mixing coils were made from coiled lengths of silicone tubing (21 cm in length, 1.5-mm i.d., Omnifit). A glass capillary was attached at the point of detection to enable the Raman scattering to be detected. These capillaries were silanized by being left overnight in Sigmacote (Sigma) and rinsed thoroughly with n-pentane and methanol prior to use. Instrumentation. Raman scattering was collected using a Renishaw 1000 microprobe spectrometer (Renishaw Ltd.) with a Spectra Physics argon ion laser (25 mW, 514.5 nm) and a Renishaw 2000 microprobe spectrometer with a Spectra Physics helium-neon laser (40 mW, 632.8 nm). A macrosampler (Ventacon) was used to excite the sample and collect the scattered light. For quantitative estimation of mitoxantrone, spectra were collected using a fixed grating centered at 1300 cm-1. All other spectra were collected as extended scans over the range 2002000 cm-1. Procedure for Concentration Studies. Methanol was placed in a silanized glass capillary tube and the intensity of the main band at 1036 cm-1 recorded. The flow cell was connected to this capillary tube, and colloid, aggregating agent, and water were pumped through the channels. Ten SERS spectra of this ag(39) Lee, P. C.; Meisel, D. J. Phys. Chem. 1982, 86, 3391-3395.
Analytical Chemistry, Vol. 74, No. 13, July 1, 2002
3161
gregated colloid “blank” were recorded. The water solution was then replaced with a solution of mitoxantrone. The flow was started, and 2 min after the total stream first passed through the capillary, the first spectrum was recorded. Unless otherwise stated, five replicate measurements were made at each mitoxantrone concentration. A wash stage in which all channels were set to pump water for 1 min was introduced and a methanol spectrum from a second tube recorded. The peak height of the main band from mitoxantrone at 1305 cm-1 was standardized against the methanol peak. The calibration graphs are shown on a logarithmic scale, but the statistical analysis was carried out on the original data, using a regression analysis. To optimize aggregant concentration for flow cell analysis, solutions of 2.5, 5, and 7.5% w/v NaCl and 5 × 10-7 M mitoxantrone in distilled water were prepared. The flow cell was set to pump 1.6 mL/min colloid, 0.23 mL/min 5 × 10-7 M mitoxantrone, and 0.23 mL/min 1% w/v NaCl solution. Flow was started, and a SERRS spectrum was collected every 10 s for 5 min. This was repeated using the other NaCl concentrations. SERRS Studies on Spiked Plasma and Serum. Blood samples from healthy volunteers were supplied by the Department of Surgery at Glasgow Royal Infirmary in tubes containing lithium heparin-coated beads. The samples were centrifuged at 2000 rpm for 15 min and the plasma removed. Initial spectra were obtained in cuvettes. The samples were spiked with mitoxantrone to give a concentration of 5 × 10-8 M. A total of 0.8 mL of spiked plasma, 4 mL of colloid, 4 mL of distilled water, and 0.8 mL of 5% w/v sodium chloride solution were added and mixed, and an aliquot was transferred to the cuvette. After 5 min, SERS spectra at 514 and 632 nm was recorded. Accumulation times of 20 s using a 200- 2000 cm-1 extended scan were used. To obtain a calibration with the flow cell, 240 mL of fresh frozen plasma and serum were thawed at 37 °C and divided to give 13 volumes of 18 mL each. These samples were spiked with increasing concentrations of a fixed volume of mitoxantrone solution and mixed thoroughly for 30 s to give stock concentrations of 2.5 × 10-9-1 × 10-6 M. SERRS was recorded as before from the flow cell, which pumped 1.6, 0.23, and 0.23 mL/min colloid, 2.5% w/v NaCl, and sample, respectively. Patient Samples. Optimization of the Flow Cell for Clinical Measurements. Serum was collected on the day of the study from patients who had been treated with mitoxantrone. To optimize conditions, two samples were spiked with mitoxantrone to give a 2.5 × 10-6 M solution. SERRS was recorded at 632.8 nm. Following these experiments, the second coil in the flow cell was doubled in length to increase contact time between the colloid and the sample. The optimum concentration of sodium chloride was reinvestigated and the final flow rates chosen as 0.5, 0.08, and 0.08 mL/min colloid, 7.5% sodium chloride, and sample, respectively. These conditions were chosen to construct a final calibration graph using serum, and the results from a series of samples taken at different times after injection were recorded and compared to previous results from an HPLC analysis. Safety. All blood, plasma, and serum samples were taken with ethical permission by clinicians at Glasgow Royal Infirmary. Samples were prepared and disposed of in a designated area using approved protocols. 3162
Analytical Chemistry, Vol. 74, No. 13, July 1, 2002
RESULTS AND DISCUSSION Preliminary SERRS Studies. The metal substrate preferred to give surface enhancement was citrate-reduced silver colloid. This colloid has a long lifetime and with care can be made relatively reproducibly.40 To achieve efficient enhancement, the colloid was aggregated with sodium chloride to form an aggregated suspension, which is stable for a sufficient time to enable the spectrum to be obtained. The enhancement mechanism from such a suspension is little understood, but related work on immobilized colloid would suggest that most of the signal arises from those aggregates having plasmon frequencies in resonance with the excitation frequency40 with quantitation possible due to the averaging effect of collecting scattering from many aggregates. Brownian motion of the aggregates improves this effect. Initial measurements were carried out by adding aqueous solutions of mitoxantrone to the silver colloid in a quartz cuvette. Sodium chloride was used as aggregating agent. The reproducibility of the signals was poor, and there were changes in relative intensity with concentration. One problem with sodium chloride aggregation which will affect the reproducibility is that it is a dynamic process, which leads to the formation of large aggregates, which drop out of suspension with time. However, the use of an organic agent (poly-(L-lysine)) in place of sodium chloride to provide more control over aggregation kinetics gave poor SERRS possibly because there was competition between the polymer and mitoxantrone at the colloid surface. Further, mitoxantrone signals could be detected from used cuvettes washed and refilled with colloid, indicating adsorption on the quartz of the cell. The change in relative intensity of the peaks suggests either changes in orientation or different methods of complexing of mitoxantrone to the surface occur under subtly different conditions. To obtain quantitative measurements, control over the kinetics of aggregation and surface adsorption was obtained by using a flow cell.41-52 Silicone tubing was used wherever practical, and all quartz surfaces were silanized before use to prevent adsorption. SERRS Intensity and Mitoxantrone Concentrations. Two concentration studies of aqueous solutions of mitoxantrone were carried out using the flow cell. The 632.8-nm excitation was used as it was the closest available frequency to the resonance wavelength of the chromophore of mitoxantrone.53 The two studies used different batches of silver colloid, and the aggregating agent (40) Rodger, C. Ph.D. Thesis, University of Strathclyde. (41) Weibenbacher, N.; Lendl, B.; Frank, J.; Wansenbock, H. D.; Mizaikoff, B.; Kellner, R. J. Mol. Struct. 1997, 410, 539-542. (42) Pothier, N. J.; Force, R. K. Anal. Chem. 1990, 62, 678-680. (43) Chen, H. Y.; Long, Y. T. Anal. Chim. Acta 1999, 382, 171-177. (44) Pothier, N. J.; Force, R. K. Appl. Spectrosc. 1994, 48, 421-425. (45) Berthol, A.; Laserna, J. J.; Winefordner, J. D. Appl. Spectrosc. 1987, 41, 1137-1141. (46) Laserna, J. J.; Berthod, A.; Winefordner, J. D. Microchem. J. 1988, 38, 125136. (47) Cabalin, L. M.; Ruperez, A.; Laserna, J. J. Anal. Chim. Acta 1996, 318, 203-210. (48) Force, R. K.,; Anal. Chem. 1988, 60, 1987-1989. (49) Freeman, R. D.; Hammaker, R. M.; Meloan, C. E.; Fateley, W. G. Appl. Spectrosc. 1988, 42, 456-460. (50) Gouveia, V. J. P.; Gutz, I. G.; Rubim, J. C. J. Electroanal. Chem. 1994, 371, 37-42. (51) Ni, F.; Sheng, R. S.; Cotton, T. M. Anal. Chem. 1990, 62, 1958-1963. (52) Taylor, G. T.; Sharma, S. K.; Mohanan, K. Appl. Spectrosc. 1990, 44, 635-640. (53) Nabiev, I.; Baranov, A.; Chourpa, I.; Beljebbar, A.; Sockalingum, G. D.; Manfait, M. J. Phys. Chem. 1995, 99, 1608.
Table 1. Analysis of Data for Studies 1 and 2 study laser wavelength colloid flow cell LOCb estimated LODc linear range correlation coefficient r2 equation gradient of logarithmic plot % RSD (n ) 5) accumulation times (s) a
1 632.8 nm colloid 1 absorption maximum 404 nm fwhha 60 nm, pH 8.45 3.4 mL/min colloid 0.23 mL/min 5% NaCl 0.32 mL/min mitoxantrone 1.6 × 10-10 M (0.08 ng/mL) 0.03 ng/mL 1.6 × 10-10-8.1 × 10-8 M (0.08-42 ng/mL) 0.9971 y ) (5.86 ( 0.81) × 1010x + 144 1.02 ( 0.07 0.7-8.0 60-5
2 632.8 nm colloid 2 absorption maximum 399 nm fwhh 50 nm, pH 7.80 3.4 mL/min colloid 0.23 mL/min 5% NaCl 0.32 mL/min mitoxantrone 6.4 × 10-10 M (0.33 ng/mL) 0.06 ng/mL 6.4 × 10-10-8.5 × 10-8 M (0.33-45 ng/mL) 0.9986 y ) (8.10 ( 0.38) × 1010x - 130 1.02 ( 0.03 0.3-4.4 60-1
fwhh, full width half-height of the stated absorbance band. b LOC, lowest observable concentration. c LOD, limit of detection.
and aqueous mitoxantrone solutions were prepared fresh for each. The spectra are qualitatively similar to published SERS spectra of mitoxantrone obtained at 514.5 nm.53 A detailed assignment of the SERS spectrum of mitoxantrone on aggregated silver colloid is available elsewhere.53 The relative peak intensities and peak positions at all concentrations in both studies are constant within experimental error. For quantitative studies, the spectral range was limited to the region from 1150 to 1487 cm-1 as a fixed grating centered at 1300 cm-1 was used. This simplifies the collection of spectra and improves stability of signals for quantitative estimation. Concentrations were measured from the peak height of the band at 1305 cm-1. A linear response was found within the stream concentration ranges of 1.6 × 10-10 (0.08 ng/mL) to 8.1 × 10-8 M (42 ng/mL) r ) 0.9971 for study 1 and 6.4 × 10-10 (0.3 ng/ mL) to 8.5 × 10-8 M (44 ng/mL) r ) 0.9986 for study 2. The data and conditions are summarized in Table 1. The correlation coefficients over this range and the y residuals showed a random distribution around zero. Limits of detection (LODs) were 0.03 ng in study 1 and 0.06 ng/mL in study 2. The reason for a higher LOD in study 2 is that the colloid used had a SERS absorption band from citrate at 1400 cm-1 which obscured the main mitoxantrone peak at very low concentrations. Little is known about the degree of surface coverage of citrate on silver colloid, but previously we observed differences in citrate SERS intensity depending on sample batch. Other authors have also reported interference from citrate bands using silver colloids of this type.54 Thus, the LOD is colloid dependent but more extensive studies of colloid batches are required and it is recommended that a standard be applied daily in any case. The precision at each concentration was measured as the relative standard deviation (RSD) of the mean from five replicates. Values of 0.7-8.0% and 0.3-4.4% were calculated for studies 1 and 2, respectively. In both cases, the highest errors were obtained at the LOD. This could be improved by increasing the accumulation time. A decrease in scattering intensity was observed when the stream concentration of mitoxantrone exceeded ∼1 × 10-7 M probably due to analyte-induced aggregation causing growth and precipitation of the aggregates. Thus, SERRS can be applied to the quantitative determination of mitoxantrone in aqueous solution. The detection limits achieved (54) Sanchez-Cortes. S.; Garcia-Romos, J. V. J. Raman Spectrosc. 1998, 29, 365.
were better than those reported for HPLC using electrochemical detection.16 The flow cell procedure is rapid and simple and is ideal for automated analysis, allowing high sample throughput. The nature of the colloid is critical, and in order to maximize sensitivity and linear range, interfering signals from citrate must be minimized. This is possible by careful attention to the preparation of the colloid. A limitation of the use of the flow cell setup as described was the large sample size and the high-volume consumption of reagents. In subsequent studies using human plasma and serum, conditions were adjusted so that the sample size was reduced. SERRS Studies on Plasma and Serum Samples Spiked with Mitoxantrone. The low LOD values indicated that, even with the dilution effects of the flow cell, the detection of mitoxantrone at clinically relevant concentrations is possible. In fact, the dilution of the sample is an advantage since it will reduce interference from fluorescence due to nonadsorbed components from more complex sample matrixes such as plasma or serum. Further, when sufficiently low amounts of sample are used and the contact time is sufficiently long, a significant fraction of the protein can adhere to the surface and fluorescence quenching from the surface may reduce fluorescence. To evaluate the extent of fluorescence interference, three plasma samples were compared before and after spiking with mitoxantrone. One sample was collected on the day of the study and two had been frozen 15 and 78 days previously. Cuvettes rather than the flow cell were used for this experiment. The samples excited with 632.8-nm excitation showed intense fluorescence which was significantly quenched by adding aggregated silver colloid (Figure 3). No Raman scattering was observed except for a low-frequency SERS band from the silver chloride layer that forms on the surface. Each sample was spiked with mitoxantrone to give a concentration in solution of 5 × 10-8 M. The SERRS recorded with 632.8-nm excitation was similar in each case and comparable to the SERRS of mitoxantrone in aqueous solution at the same concentration (Figure 4). There is a fluorescence contribution from the plasma matrix. However, this does not mask the main band at 1305 cm-1. Thus, it would be possible to determine mitoxantrone from plasma samples containing the drug without pretreatment. This study was repeated with Analytical Chemistry, Vol. 74, No. 13, July 1, 2002
3163
Figure 5. Measured intensity of the 1305-cm-1 peak with time at three NaCl concentrations (514.5 nm).
Figure 3. Spectra obtained from blood plasma (632.8 nm) before (top) and after (bottom) addition of silver colloid indicating the change in fluorescence. The peak at low frequency is due to SERS from silver chloride/oxide on the silver surface.
Figure 4. SERRS of mitoxantrone from (a) spiked plasma and (b) aqueous solution.
514.5-nm excitation, and the fluorescence was lower at this wavelength. To compare the relative advantages of using plasma or serum, a single blood sample was collected into three sets of tubes containing no anticoagulant, lithium heparin anticoagulant, and EDTA anticoagulant. Immediately after collection, the samples were centrifuged for 15 min at 2000 rpm, and the supernatant was removed, giving one serum and two plasma samples, respectively. A small volume of each was transferred into a glass cuvette containing aggregated colloid, and after 5 min, the spectra at 632.8 and 514.5 nm were collected. The two plasma samples were similar and were more fluorescent than the serum sample. Fluorescence was more intense at 632.8 nm and was not significantly altered by the addition of recommended stabilizing agentss5% w/v ascorbic acid14,17,18,20,2126,28-30 and 0.001% w/v sodium metabisulfite.16 The serum sample excited at 514.5 nm showed the least fluorescence. 3164 Analytical Chemistry, Vol. 74, No. 13, July 1, 2002
A concentration-dependent study on both plasma and serum was carried out using the flow cell with 514.5-nm excitation. To minimize colloid consumption, the flow cell was modified to pump 1.6, 0.23, and 0.23 mL/min colloid, NaCl aggregating agent, and mitoxantrone solution, respectively. The concentration of aggregating agent to provide the most intense SERRS signals from mitoxantrone was optimized. Figure 5 shows that the most intense SERRS signals were detected using a 2.5% w/w solution of NaCl to aggregate the colloid. The signal was relatively stable with 5.7% variability in the mean intensity over 5 min. Use of more concentrated solutions resulted in a decrease in the observed intensity, as the colloid precipitates out of solution more quickly. This optimized aggregant concentration was used for the concentration studies. The extent of fluorescence from plasma was very sample dependent. When a fresh batch of frozen plasma used, the fluorescent background from the plasma was more intense than in the preliminary study. As a result, with 514.5-nm excitation, the 1305-cm-1 band could not be detected as a distinct band until a stock concentration of 2.5 × 10-7 M was injected. This corresponds to a stream concentration of 2.8 × 10-8 M or 14.4 ng/mL. Fluorescence was greater at 632.8 nm. A fresh serum sample was collected on the day of the study and spiked with mitoxantrone. The background fluorescence was more intense than in the preliminary study, and three bands at 1000, 1152, and 1514 cm-1 were detected from unspiked serum. However, it was possible to discriminate the main mitoxantrone peak from the background at a concentration in the stream of 2.2 × 10-9 M (1.1 ng/mL). Table 2 summarizes the experimental details and results for the serum concentration study. The data over the stream concentration range from 2.2 × 10-9 (1.1 ng/ mL) to 1.62 × 10-7 M (83 ng/mL) showed a good straight-line response (r ) 0.9954, y residuals randomly distributed around zero). This corresponds to injected stock solution concentrations of between 1.0 × 10-8 (10 ng/mL) and 7.5 × 10-7 M (741 ng/ mL). The gradient of the logarithmic plot was 1.08, indicating a linear response. The precision of SERS intensities was within the range 1.2-12.2% and was dependent on concentration. Accumulation times were 90 s at most. The repeatability of the SERRS intensities was checked with 514.5-nm excitation using multiple additions of mitoxantronespiked serum. The flow cell was set up to pump 1.6, 0.23, and 0.23 mL/min colloid, 2.5% NaCl, and distilled water, respectively,
Table 2. Experimental Details and Results of Mitoxantrone Concentrations in Spiked Serum with 514.5-nm Excitation Compared to Those from the Final Optimized Procedure with 632.8-nm Excitation Using Serum from a Patient laser wavelength colloid flow cell LOCb estimated LODc range of correlation correlation coefficient r equation gradient of logarithmic plot % RSD (n ) 5) accumulation times (s) a
514.5 nm colloid 3 absorption maximum 402 nm fwhha 65 nm, pH 6.92 1.6 mL/min colloid 0.23 mL/min 2.5% NaCl 0.23 mL/min mitoxantrone 2.2 × 10-9 M (1 ng/mL) 3.0 × 10-10 M (0.2 ng/mL) 2.2 × 10-9-1.6 × 10-7 M (1-84 ng/mL) 0.9954 y ) (1.66 ( 0.16) × 1010x - 61 1.08 ( 0.05 1.2-6.7 90-15
632.8 nm colloid 4 absorption maximum 404 nm fwhh 55 nm, pH 8.02 1.6 mL/min colloid 0.23 mL/min 7.5% NaCl 0.23 mL/min serum 1.12 × 10-10 M (0.06 ng/mL) 4.0 × 10-11 M (0.02 ng/mL) 1.12 × 10-10-8.37 × 10-8 M (1-43 ng/mL) 0.9990 y ) (1.14 ( 0.01) × 1011x + 50 1.01 ( 0.02 0.6-19.2 120-3s
fwhh, full width half-height of the stated absorbance band. b LOC, lowest observable concentration. c LOD, limit of detection.
Figure 6. Intensity at 1305 cm-1 with time for five injections of serum spiked with mitoxantrone into flow cell (514.5 nm).
through the three channels. Spectra from the flowing stream were acquired automatically every 10 s. One minute after the first spectrum was collected, the water stream was replaced by a stream of serum containing 5 × 10-7 M mitoxantrone. This was allowed to flow for 1 min 45 s, and then water was flowed through this channel for 2 min. This procedure was repeated four times. Figure 6 shows the plot of the measured intensity at 1305 cm-1 with time. The variability of the peak maximum for five injections was 0.9%. After five injections, the carry-over was 3.5% of the maximum signal due to adsorption of aggregated colloid on the internal wall of the capillary. This could be completely eliminated by replacing the glass capillary used to enable scattering to be detected after each measurement. The variation in SERRS intensity of mitoxantrone from spiked serum samples over time was checked with 632.8-nm excitation at times 0, 1, 6, 12, 18, and 24 h after spiking. A methanol spectrum was run at each time point. The intensity of the main band at 1305 cm-1 was found to remain constant within 5% over 24 h. This procedure was repeated with a spiked plasma sample, and the variability of the main band intensity was 6.5%. Thus, quantitative analysis of mitoxantrone directly from serum or plasma is possible. However, the treatment and storage of the blood or blood fractions used for analysis is important, particularly since the intensity of fluorescence appears to be affected. Consequently, the choice of excitation frequency and blood
fraction for a clinical study is dependent on the sample pretreatment. The sensitivity achieved at 514.5 nm is less by ∼1 order of magnitude than that at 632.8 nm probably due to decreased resonance enhancement from mitoxantrone, but fluorescence is lower. For the analysis of patient samples in this study, fresh serum was available, fluorescence interference proved to be low, and the use of 632.8-nm excitation was preferred. The 514.5-nm excitation may be preferable with some samples treated differently. If it is used, longer accumulation times may be required for samples containing low levels of the drug. Determination of Mitoxantrone in Patient Samples. Serum samples from two patients who had been treated with mitoxantrone, methotrexate, and mitomycin C were analyzed to compare SERRS and HPLC detection. Methotrexate and mitomycin C were present in the patient samples at levels up to 13 µmol/L and 1523 ng/mL respectively. An attempt to obtain SERRS of these drugs from serum with either 514.5- or 632.8-nm excitation produced no peaks other than those from aggregated citrate silver colloid. Levels of mitoxantrone determined by HPLC40 in a previous study ranged from 0 to 375 ng/mL. Optimization experiments to minimize the fluorescence were carried out using spiked aliquots of this serum. Since the observed fluorescence from plasma was greater with flow cell detection than with cuvette detection, the contact time between the colloid and the serum was increased by reducing the pump speed and increasing the length of the coil in the flow cell in which the serum and aggregated colloid mix. Further, low pump speeds favor low consumption of reagents, reducing the patient serum volume required. Contact times between the colloid and serum were estimated from the time a signal could be obtained and recorded at 55, 26, 12, and 7 s, respectively. Five SERRS spectra were recorded at each pump speed, 2 min after the stream had first passed through the capillary. Although the lowest flow rate gave the best signalto-noise ratio, the time taken for the SERRS intensity to reach a maximum was ∼14 min (Figure 7). A pump speed of 10 rpm in combination with an increase in the length of the second coil in the flow cell provided a contact time of ∼60 s. There was a decrease in the maximum signal but the time to reach a steady signal was reduced to ∼2 min. There was little background Analytical Chemistry, Vol. 74, No. 13, July 1, 2002
3165
Table 3. Summary of HPLC and SERRS Analysis Results for Patient Samples concn MXa (ng/mL)
Figure 7. Intensity (1305 cm-1) with time from serum spiked with mitoxantrone recorded from a flowing stream at two pump speeds.
a
Figure 8. Calibration plot of the intensity of the SERRS of the mitoxanthrone band at 1305 cm-1 against serum concentration in the flow cell stream.
fluorescence. At higher pump speeds, the time was reduced but the increase in the consumption of the reagents and decrease in effective SERRS intensities mitigated against use of these higher speeds. As the pump speed and tubing lengths within the flow cell had been changed, the concentration of NaCl was reoptimized. A 7.5% w/v NaCl solution for aggregation gave the most intense SERRS signals. The precision from five measurements at this concentration was 3.2%. In addition, the amount of colloid added could affect the degree of fluorescence quenching obtained by increasing the surface area for adsorption. Silver colloid diluted 50% with distilled water is used throughout this study (see Experimental Section). It was compared with undiluted colloid and colloid 50% concentrated by centrifugation. However, the fluorescence observed from serum at either 514.5 or 632.8 nm was little affected by altering colloid concentration, and therefore, there was little advantage in preconcentration of the colloid. A final calibration plot using 632.8-nm excitation was generated from fresh serum using the optimized conditions (Figure 8). The results are given in Table 2 along with those from the study using 514.5-nm excitation and, for comparison, the previous conditions before changing the flow cell dimensions and the NaCl concentration. As can be seen, a range of excitation wavelengths and conditions can give satisfactory results. The stability of the measurements with time was estimated using a fresh batch of frozen serum spiked with 5 × 10-7 M mitoxantrone. The estimated concentrations were within 5% of the mean over 24 h. 3166 Analytical Chemistry, Vol. 74, No. 13, July 1, 2002
time
HPLC
SERRS
0 5 10 15 20 30 45 60 90 120 180 240 360 720
0 3.1 1.9 252.3 186.9 53.7 23.7 11.6 9.4 6.2 5.7 1.8 3.2 1.8
ndb 2.9 1.0 247.3 183.2 54.1 20.3 9.2 8.9 6.0 4.8 1.0 3.0 1.3
MX, mitoxantrone. b nd, not detected.
The limit of detection achieved by SERRS in this study was 4.0 × 10-11 M (0.02 ng/mL). This is the lowest reported for any analytical method to date, and flow cell sampling means the method is reliable. In addition, mitoxantrone was determined directly from serum samples without a preconcentration or extraction stage, and the method is fast and could be easily automated. The linear range encompasses the range of mitoxantrone concentrations detected in the patient samples by HPLC. The precision results (n ) 5) ranged from 0.6 to 19.2% with the greatest variability in signals (19.2-9.4%) at the three lowest concentrations. Serum samples collected from one patient over a 12-h time period after the patient had been treated with mitoxantrone were analyzed. The concentration of mitoxantrone in the samples had previously been determined by HPLC. The samples were supplied frozen and were thawed at 37 °C in a water bath immediately before SERRS analysis.. Ten SERRS spectra were recorded from the flowing stream for each individual sample. Very little fluorescence was observed. Table 3 summarizes the results obtained for the SERRS analysis of the three patient samples along with those obtained by the HPLC assay. A paired t-test was used to compare the results obtained by the two methods to check for significant differences. At the 95% confidence limit, there was no significant difference in the results between the two methods. CONCLUSIONS Determination of mitoxantrone in the serum of patients down to subnanogram levels without prior need for sample manipulation has been achieved. The analysis time from addition of the serum to the flow cell was of the order of 2 min, and only microliters of sample were consumed. A typical precision of less than 5% (n ) 3) was achieved. For the clinical samples used in this study, if there was sufficient contact time between the silver colloid and the biological sample, any fluorescence from the matrix was suppressed efficiently. Serum is preferred to plasma for the analysis, but both can be used. Other laser excitation frequencies will be effective, but of the two used here, 632.8 nm is preferred. The main conclusion is that SERRS could be the technique of choice for quantitative analysis of mitoxantrone. It is simple, fast,
sensitive, selective, and reliable. This example provides a real application of SERRS, which uses the well-known potential advantages of selectivity and sensitivity with a flow cell to provide control over the kinetics of the process. The absence of interference from the other drugs in the serum indicates that it is important to select targets in which the analyte adheres well to the surface and gives strong SERRS. However, in selected cases, SERRS has considerable promise for quantitative analysis of clinical samples.
ACKNOWLEDGMENT We thank Dr. P. C. White for bringing this problem to our attention.
Received for review January 16, 2001. Accepted October 10, 2001. AC010067K
Analytical Chemistry, Vol. 74, No. 13, July 1, 2002
3167
