Quantitative Analysis of Lycopene Isomers in Human Plasma Using

Anal. Chem. 2003, 75, 812-817

Quantitative Analysis of Lycopene Isomers in Human Plasma Using High-Performance Liquid Chromatography-Tandem Mass Spectrometry Liqiong Fang, Natasa Pajkovic, Yan Wang, Chungang Gu, and Richard B. van Breemen*

Department of Medicinal Chemistry and Pharmacognosy, University of Illinois at Chicago, 833 South Wood Street, Chicago, Illinois 60612

An analytical method for the determination of the concentrations of total lycopene and its cis and all-trans isomers in human plasma has been developed using highperformance liquid chromatography-tandem mass spectrometry (LC-MS-MS). This method was based on the observation that, during negative ion atmospheric pressure chemical ionization with collision-induced dissociation, a unique fragment of m/z 467 was formed from the molecular ion of m/z 536 by elimination of a terminal isoprene group. The use of multiple reaction monitoring facilitated the selective detection of lycopene isomers and an internal standard without interference from the isobaric carotenoids r-carotene and β-carotene, which are also abundant in human plasma. Measurement of total lycopene was carried out using a C18 high-performance liquid chromatography (HPLC) column and an isocratic mobile phase consisting of acetonitrile/methyl tert-butyl ether (95:5) so that all lycopene isomers eluted as a single chromatographic peak. all-trans-Lycopene was separated from its various cis isomers by using a C30 carotenoid column and a gradient solvent system from methanol to methyl tert-butyl ether. The effects of sample preparation and handling parameters on the stability of lycopene were evaluated such as the stability of lycopene in the HPLC autosampler and the effect of saponification upon lycopene isomerization. For example, the half-life of all-translycopene in the HPLC mobile phase in the autosampler at 4 °C was determined to be ∼16 h. Also, saponification of plasma samples was determined to cause lycopene degradation and isomerization so that lycopene recovery was reduced. The accuracy and interassay precision of this LC-MS-MS assay for lycopene showed a standard deviation of less than 10% over the range of 5-500 pmol injected on-column. The limit of detection was 11.2 fmol injected on-column, and the limit of quantitation was 22.8 fmol. Lycopene is a nonprovitamin A carotenoid responsible for the red color of the fruit of the tomato, Solanum lycopersicum. The extended system of 11 conjugated double bonds and 2 nonconjugated double bonds makes this molecule the most effective * Corresponding author. Telephone: (312) 996-9353. Fax: (312) 996-7107. E-mail: [email protected]..

812 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

singlet oxygen quencher among ∼600 naturally occurring carotenoids.1 Interest in lycopene research has been growing rapidly since epidemiological studies indicated an association between a tomato-rich diet and a lower risk of prostate cancer.2 The various dietary sources, absorption, tissue distribution, metabolism, biological functions, and potential effects on disease prevention of lycopene have been reviewed recently.3-5 The predominant geometric isomer of lycopene from plant sources is the all-trans form. However, various cis isomers are present in human serum and tissues and constitute more than 50% of total lycopene.6,7 Most methods for the determination of lycopene concentration in biological samples use high-performance liquid chromatography (HPLC) for separation from other carotenoids (β-carotene, R-carotene, lutein, cryptoxanthin, etc.) and then absorbance detection at 450-472 nm.6-13 Since many carotenoids have overlapping UV/visible absorbances, careful separation procedures are necessary to avoid interference during quantitative analysis. However, the occurrence of multiple cislycopene isomers makes their HPLC separation challenging even when using C30 chromatography, which was developed specifically for carotenoid separations.14,15 Therefore, the use of more selective detection methods would help reduce the possibility of interference during the quantitative analysis of lycopene isomers. (1) DiMascio, P.; Kaiser, S.; Sies, H. Arch. Biochem. Biophys. 1989, 274, 532538. (2) Giovannucci, E. J. Natl. Cancer Inst. 1999, 91, 317-331. (3) Stahl, W.; Sies, H. Arch. Biochem. Biophys. 1996, 336, 1-9. (4) Clinton, S. K. Nutr. Rev. 1998, 56, 35-51. (5) Rao, A. V.; Agarwal, S. Nutr. Res. 1999, 19, 305-323. (6) Stahl, W.; Schwarz, W.; Sundquist, A. R.; Sies, H. Arch. Biochem. Biophys. 1992, 294, 173-177. (7) Clinton, S. K.; Emenhiser, C.; Schwartz, S. J.; Bostwick, D. G.; Williams, A. W.; Moore, B. J.; Erdman, J. W., Jr. Cancer Epidemiol. Biomarkers Prev. 1996, 5, 823-833. (8) Lyan, B.; Azaı¨s-Braesco, V.; Cardinault, N.; Tyssandier, V.; Borel, P.; Alexandre-Gouabau, M.; Grolier, P. J. Chromatogr., B 2001, 751, 297303. (9) Nierenberg, D. W.; Nann, S. L. Am. J. Clin. Nutr. 1992, 56, 417-426. (10) Froescheis, O.; Moalli, S.; Liechti, H.; Bausch, J. J. Chromatogr., B 2000, 739, 291-299. (11) Ferreira, A. L. A.; Yeum, K.-J.; Liu, C.; Smith, D.; Krinsky, N. I.; Wang, X.D.; Russell, R. M. J. Nutr. 2000, 130, 1256-1260. (12) Oshima, S.; Sakamoto, H.; Ishiguro, Y.; Terao, J. J. Nutr. 1997, 127, 14751479. (13) Ishida, B. K.; Ma, J.; Chan, B. Phytochem. Anal. 2001, 12, 194-198. (14) Emenhiser, C.; Sander, L. C.; Schwartz, S. J. J. Chromatogr., A 1995, 707, 205-216. (15) Emenhiser, C.; Simunovic, N.; Sander, L. C.; Schwartz, S. J. J. Agric. Food Chem. 1996, 44, 3887-3893. 10.1021/ac026118a CCC: $25.00

© 2003 American Chemical Society Published on Web 01/21/2003

The use of liquid chromatography-mass spectrometry (LCMS) helps improve the selectivity of carotenoid analysis,16,17 and LC-MS with positive ion atmospheric pressure chemical ionization (APCI) has been used for the quantitative analysis of lycopene in human serum and prostate tissue.18,19 During positive ion APCI, lycopene forms abundant protonated molecules with no significant fragmentation. Since R-carotene, β-carotene, and their cis and alltrans isomers also form abundant protonated molecules at m/z 537 during APCI, mass spectrometric detection alone is insufficient to distinguish these compounds from isomeric lycopene. Using positive ion continuous-flow fast atom bombardment with LCMS-MS, we have shown that lycopene can be distinguished from isomeric R-carotene and β-carotene.20 However, this technique was unsuitable for quantitative analysis, for unattended operation, or for the analysis of large numbers of samples. Therefore, we report here a quantitative and robust LC-MS-MS assay that distinguishes lycopene and its geometrical isomers from isomeric R-carotene and β-carotene while providing high sensitivity and high throughput. EXPERIMENTAL SECTION Chemicals and Reagents. Lycopene (from tomato), R-carotene, and β-carotene standards were purchased from Sigma (St. Louis, MO) and stored at -80 °C. Human plasma and the antioxidant butylhydroxytoluene (BHT, 2,6-di-tert-butyl-4-methylphenol) were also purchased from Sigma. The tomato extract (LycO-Mato) was a gift from LycoRed Natural Products (Beer-Sheva, Israel) and was stored at -20 °C. Methanol, acetonitrile, methyltert-butyl ether (MTBE), and hexane were HPLC-grade or better and were obtained from Fisher Scientific (Fair Lawn, NJ). Absolute ethanol (ACS reagent grade) was purchased from Aldrich (Milwaukee, WI). The internal standard, [13C6]-β-carotene, was a gift from Johan Lugtenburg of Leiden University (Leiden, The Netherlands).21 Deionized water was prepared using a Barnstead (Dubuque, IA) Nanopure purification system. Standard Solutions. A stock solution of lycopene was prepared by dissolving ∼1 mg in 20 mL of chloroform followed by storage at -80 °C. Fresh calibration solutions were prepared each day from the stock solution after measurement of the concentration of lycopene using a Shimadzu (Columbia, MD) UV2401PC UV-visible recording spectrophotometer. Lycopene concentrations were determined based on the absorbance at 502 nm using a molar absorptivity of 1.72 × 10-5 L mol-1 cm-1 in hexane/ CH2Cl2 (98:2; v/v).22 The purity of the lycopene stock solution was verified using LC-MS. all-trans-Lycopene but no other carotenoids was detected in fresh aliquots of the stock solution. LC-MS-MS. High-performance liquid chromatography was carried out using either a Waters (Milford, MA) 2690 HPLC (16) van Breemen, R. B.; Huang, C.-R.; Tan, Y.; Sander, L. C.; Schilling, A. B. J. Mass Spectrom. 1996, 31, 975-981. (17) van Breemen, R. B. Pure Appl. Chem. 1997, 69, 2061-2066. (18) van Breemen, R. B.; Xu, X.; Viana, M. A.; Chen, L.; Stacewicz-Sapuntzakis, M.; Duncan, C.; Bowen, P. E.; Sharifi, R. J. Agric. Food Chem. 2002, 50, 2214-2219. (19) Hagiwara, T.; Yasuno, T.; Funayama, K.; Suzuki, S. J. Chromatogr., B 1998, 708, 67-73. (20) van Breemen, R. B.; Schmitz, H. H.; Schwartz, S. J. Anal. Chem. 1993, 65, 965-969. (21) Lugtenburg, J.; Creemers, A. F. L.; Verhoeven, M. A.; van Wijk, A. A. C.; Verdegem, P. J. E.; Monnee, M. C. F.; Jansen, F. J. H. M. Pure Appl. Chem. 1999, 71, 2245-2251.

system interfaced to a Micromass (Manchester, U.K.) Quattro II triple quadrupole mass spectrometer or a ThermoFinnigan (San Jose, CA) Surveyor HPLC system with a Quantum triple quadrupole mass spectrometer. For the determination of total lycopene, all-trans-lycopene and lycopene cis isomers were eluted as a single chromatographic peak and separated from the internal standard [13C6]-β-carotene using a Waters Xterra MS C18 column (3.5 µm, 2.1 × 100 mm) with isocratic acetonitrile/MTBE (95:5; v/v) as the mobile phase at a flow rate of 0.4 mL/min. Separation of alltrans-lycopene from its various cis isomers was carried out using a reversed-phase YMC (Wilmington, NC) C30 carotenoid column (3 µm, 4.6 × 250 mm) with a 40-min linear gradient from 50:50 to 40:60 methanol/MTBE at a flow rate of 0.65 mL/min. Both positive and negative ion APCI were evaluated for the LC-MS-MS analysis of lycopene. During LC-MS-MS using acetonitrile and MTBE, the APCI parameters were optimized to facilitate the formation of lycopene molecular ions, M-‚, in negative ion mode and protonated molecules in positive ion mode. Following collision-induced dissociation (CID) in negative ion mode, the product ion of m/z 467 (formed by elimination of a terminal isoprene group from the molecular ion) was recorded selectively using multiple reaction monitoring (MRM). The molecular ion of the internal standard, which did not fragment under these conditions, was also recorded at m/z 542. The MRM dwell time was 2 s/ion. The corona voltage was optimized to 3.85 kV, and the cone voltage was 50 V. Argon was used as collision gas with a collision energy of 20 eV. The source block temperature was set to 110 °C, and the probe temperature was 350-400 °C. The probe position was optimized by tuning the instrument during the infusion of a standard lycopene solution under the appropriate mobile-phase conditions. Calibration curves were obtained using freshly prepared lycopene standard solutions in the range of 0.1-50 µM in acetonitrile/MTBE (1:1, v/v). Plasma was not used for the preparation of standard solutions since plasma already contains lycopene. The internal standard (∼20 µM) was added to each standard and sample to correct for variation in instrument response. Aliquots of 10 µL of each standard solution were injected onto the column for LC-MS-MS analysis, and calibration curves were constructed by linear regression analysis of the area ratios of lycopene/internal standard versus the amount of lycopene (pmol) injected. Concentrations of all-trans- and cis-lycopenes or total lycopene in each sample were calculated by comparing their peak areas to the standard curve. Sample Preparation. Human plasma (300 µL) was mixed with internal standard (10 µL, ∼20 µM) and ethanol (300 µL). The mixture was extracted twice with 2 mL of hexane containing 100 mg/L BHT. The hexane extracts were collected after centrifugation at 500g for 5 min (4 °C), combined, and evaporated to dryness under vacuum. The extract was reconstituted in 100 µL of acetonitrile/MTBE (1:1, v/v) or methanol/MTBE (1:1, v/v). An aliquot of 10 µL was injected onto the LC-MS-MS for total lycopene measurement, and 20 µL was injected for measurement of lycopene geometrical isomers. To evaluate the effects of saponification on total lycopene concentration and the percentage of trans and cis isomers, human plasma (300 µL) was mixed with internal standard, ethanol (300 µL) containing 2% BHT, and 200 µL of 60% KOH. The mixture Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

813

Figure 1. Negative ion APCI tandem mass spectrum with CID of the molecular ion of lycopene at m/z 536.

was sealed in a vial and incubated in a 60 °C water bath for 1 h before extraction as described above. Since lycopene is unstable to light, extraction was carried out in dim light or in the dark, and all samples were stored in the dark. RESULTS AND DISCUSSION Although lycopene formed abundant protonated molecules with minimal fragmentation during positive ion APCI, subsequent CID in the collision region of the triple quadrupole mass spectrometer did not produce any abundant fragment ions. However, a molecular ion of m/z 536 was formed during negative ion APCI, which fragmented during CID to form an abundant and unique fragment ion of m/z 467 (Figure 1). Therefore, negative ion APCI was used for all subsequent studies. The m/z 467 ion corresponds to the elimination of a terminal isoprene group, [M - C5H9]-. Since lycopene is a linear and symmetrical molecule, loss of an isoprene group from either end of the molecular ion forms the same even-electron fragment ion. A proposed structure for the ion of m/z 467 and a possible mechanism for its formation are shown in Figure 1. No other abundant fragment ions were observed. During LC-MS with negative ion APCI, lycopene, R-carotene, and β-carotene formed molecular ions of m/z 536 (see LC-MS chromatograms in Figure 2 and Figure 3). Although isomeric with lycopene, R-carotene and β-carotene contain terminal rings instead of acyclic isoprene groups and consequently do not form fragment ions of m/z 467 during negative ion APCI and CID (see LCMS-MS chromatogram in Figure 2C and Figure 3B). Therefore, selected reaction monitoring was used during LC-MS-MS to measure the lycopene signal corresponding to the unique reaction pathway m/z 536 f 467. Another consequence of the high selectivity of the MS-MS detection used in this assay is that the analysis is less dependent on the chromatographic separation. By simplifying the chromatography step, the overall analysis time may be decreased. For example, the separation of lycopene from R-carotene and β-carotene shown in the LC-MS-MS chromatogram in Figure 2 was completed in less than 8 min. In the LC-MS-MS chromatogram 814 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

Figure 2. Comparison of negative ion APCI LC-MS and LC-MSMS analyses of a commercial tomato extract (Lyc-O-Mato). (A) LCMS with selected ion monitoring of m/z 536 showing peaks for the molecular ions of lycopene and β-carotene at 3.7 and 6.5 min, respectively; (B) LC-MS of the tomato extract spiked with R-carotene and β-carotene standards showing a new peak at 6.1 min corresponding to R-carotene; and (C) LC-MS-MS analysis of the spiked tomato extract using selected reaction monitoring of m/z 536 f 467. Note that R- and β-carotene were not detected using these LC-MSMS parameters. A trace of γ-carotene was detected at 4.8 min.

of a human plasma extract shown in Figure 3B, numerous lycopene isomers were resolved or partially resolved from each other and detected in less than 40 min. In contrast, some LCUV/visible analyses with C30 columns of comparable dimensions require 60-70 min.7,15 In general, the efficiency of the HPLC separation and level of selectivity for detection that are required must be determined for a specific application on a case-by-case basis. Although the carotenoids ζ-carotene, neurosporine, phytoene, and phytofluene contain terminal isoprene groups, they cannot interfere with this assay since they have different molecular ions of m/z 540, 538, 548, and 546, respectively. However, there are two carotenoids, δ-carotene and γ-carotene, that have the potential

Figure 3. C30 reversed-phase HPLC separations of carotenes with on-line negative ion APCI MS-MS or MS detection. (A) LC-MS-MS chromatogram of a lycopene standard solution using CID and MRM; (B) LC-MS-MS analysis of a human plasma extract obtained using CID and MRM; (C) LC-MS chromatogram of a human plasma extract showing lower signal-to-noise and lower selectivity; note the detection of additional peaks corresponding to R-carotene and β-carotene.

to produce peaks in this LC-MS-MS assay since they are isomeric with lycopene and contain one terminal acyclic isoprene group each. In the first case, δ-carotene is not widely distributed in the human diet and is only abundant in the food plant called the peach palm.23 Since the peach palm is uncommon in the diets of developed countries, it is not surprising (to the best of our knowledge) that δ-carotene has not been detected in humans. Therefore, δ-carotene is unlikely to be detected during this assay. In the second case, γ-carotene is known to be a constituent of the tomato24 as well as of some less common Latin American foods such as buriti and pitanga.23 The LC-MS chromatogram of the carotenoid-rich tomato extract Lyc-O-Mato in Figure 2A shows a small peak for γ-carotene (∼5% relative to lycopene) eluting at 4.8 min. Note that, in the corresponding LC-MS-MS chromatogram shown in Figure 2C, the abundance of γ-carotene relative to lycopene is diminished by ∼50% since γ-carotene has only one terminal isoprene group and is less likely than lycopene to eliminate this group during CID. Therefore, γ-carotene will be detected using this LC-MS-MS assay but will produce a diminished response relative to lycopene. Furthermore, since γ-carotene elutes more than 1 min later than lycopene under these chromatographic conditions, these two carotenoids may still be distinguished. Another advantage of LC-MS-MS analyses of lycopene compared to LC-MS, LC-UV/visible, or HPLC with electrochemical detection25 is an enhancement of signal-to-noise ratio and therefore greater sensitivity. The limit of detection (LOD) and limit of quantitation (LOQ) for the LC-MS-MS analysis of lycopene were determined by analyzing serial dilutions of standard solutions. When 11.2 fmol of lycopene was injected, a response (22) Hengartner, U.; Bernhard, K.; Meyer, K.; Englert G.; Glinz, E. Helv. Chim. Acta 1992, 75, 1848-1865. (23) Rodriguez-Amaya, D. B. Arch. Latinoam Nutr. 1999, 49, 74S-84S. (24) Tonucci, L. H.; Holden, J. M.; Beecher, G. R.; Khachik, F.; Davis, C. S.; Mulokozi, G. J. Agric. Food Chem. 1995, 43, 579-586. (25) Ferruzzi, M. G.; Nguyen, M. L.; Sander, L. C.; Rock, C. L.; Schwartz, S. J. J. Chromatogr., B 2001, 760, 289-299.

Table 1. Accuracy and Interday Precision of LC-MS-MS Analysis of Lycopene Standard Solutionsa concn of std soln (µM)

nominal amt injected on-column (pmol)

measd amt (mean ( SD, pmol)

accuracy (% deviatn)

Interday precision (RSD, %)

0.5 1 5 10 50

5 10 50 100 500

5.47 ( 0.37 9.22 ( 0.75 50.73 ( 2.77 97.78 ( 3.81 498.69 ( 3.06

9.36 -7.8 1.46 -2.22 -0.26

6.80 8.09 5.46 3.90 0.61

a

n ) 5 on five different days.

was observed with a signal-to-noise (peak-to-peak) ratio of 3:1, which was defined as the LOD. The LOQ was determined to be 22.8 fmol on-column and was defined as the lowest amount of lycopene that could repeatedly produce a response with a signalto-noise (peak-to-peak) ratio of greater than 10:1. For comparison, the LOD of our LC-MS method18 was 0.93 pmol on-column, and the LOD of a recent method using HPLC with electrochemical detection was 50 fmol, which was reported to be 10-100-fold more sensitive than UV/visible absorbance methods.25 Ten calibration curves were constructed on five different days over the ranges 0.5-5 (5-50 pmol on-column) and 5-50 µM (50500 pmol on-column). Typical equations for these calibration curves were as follows: y ) 0.0583x + 0.0006 (R2 ) 0.998, 5-50 pmol), and y ) 0.0872x - 1.90 (R2 ) 0.9994, 50-500 pmol). The upper limit of the calibration curves was 50 µM, since this was the limit of solubility of lycopene in acetonitrile/MTBE (1:1, v/v) Since human plasma samples without any lycopene were unavailable, the recovery of lycopene during extraction was determined based on the measured lycopene levels in spiked samples after subtracting the amount of lycopene measured in unspiked plasma. Lycopene recovery was determined by extraction of human plasma spiked with known concentrations of lycopene at 0.2 and 0.5 µM. Three unspiked and three spiked Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

815

Table 2. Effects of Saponification on Total Lycopene Concentration and Lycopene Isomerization

no saponification with saponification a

total lycopene (µM ( SD)

all-translycopene (% ( SD)

total cislycopene (% ( SD)

cis-lycopenes rta 13-21 min (% ( SD)

cis-lycopenes rt 23-27 min (% ( SD)

5-cis-lycopene rt 37.5 min (% ( SD)

0.80 ( 0.03 0.69 ( 0.06

36.0 ( 1.4 25.6 ( 2.0

64.0 ( 1.4 74.4 ( 2.1

19.7 ( 1.1 37.5 ( 2.8

13.5 ( 0.8 14.1 ( 1.4

30.8 ( 0.4 22.8 ( 1.7

rt, retention time during C30 LC-MS-MS analysis.

human plasma samples were extracted for each concentration, and total lycopene concentrations were measured. The recovery values were 91.0 ( 3.6% (RSD 4.0%) at 0.2 µM and 94.7 ( 9.1% (RSD 9.6%) at 0.5 µM. Accuracy and interday precision for the analysis of total lycopene in human plasma were assessed using five data sets obtained on five different days. Standard solutions containing five concentrations of lycopene at 0.5, 1, 5, 10, and 50 µM were measured. The mean concentrations calculated from linear equations based on peak area ratios (lycopene/internal standard) were compared with nominal concentrations. Accuracy was expressed as the percentage deviation of the measured concentration from the nominal concentration. In the range of 0.5-50 µM, the deviations between measured and nominal concentrations of lycopene standard solutions were less than 10% (see Table 1). The interday precision expressed by the relative standard deviation (RSD) was within 9% at all concentration levels (Table 1). Intraday precision (repeatability) for the measurement of total lycopene was assessed by LC-MS-MS analysis of extracts of pooled human plasma samples. The average concentration of five replicate analyses of human plasma samples was 0.37 µM with a standard deviation (SD) of 0.038 µM. The intraday precision or repeatability, expressed as the RSD, was 10.3%. Although all-trans-lycopene is the only geometrical isomer formed in the tomato, various cis isomers of lycopene are found in human plasma.6,7 For example, Figure 3 shows the cis- and alltrans-lycopene distribution in a lycopene standard and in a human plasma extract by separation using a C30 reversed-phase HPLC column with MS-MS detection. Under these conditions, all-translycopene eluted at ∼36.5 min and constituted >95% of the lycopene in the standard solution. In the human plasma extract, the ratio of cis-/all-trans-lycopene was ∼60:40. The most abundant cis isomer, 5-cis-lycopene, eluted at ∼37.5 min immediately after the all-trans isomer. The percentages of all-trans- and cis-lycopenes can be determined based on the respective peak areas in the LCMS-MS chromatogram. Since the negative ion tandem mass spectrum of all-trans-lycopene is identical to those of the various cis isomers, the MS-MS response for each of the isomers is identical. Furthermore, the standard curve for the all-trans isomer may be used for the quantitative analysis of the various cis isomers. These results are consistent with the identical mass spectrometric responses that we have observed for the geometric isomers of retinoic acid.26 It is important to note that UV/visible absorbance spectroscopy does not produce identical responses for various cis and all-trans isomers of lycopene.27

Saponification is applied routinely during lycopene extraction from tissue samples in order to enhance recovery.7,18 For example, Ferreira et al.11 investigated the effect of saponification during the extraction of lycopene from several tissues and found that saponification resulted in greater total lycopene recovery. Although some groups also saponify serum or plasma when measuring lycopene,7 it is unclear whether this practice is necessary in this case. To resolve this issue, we used LC-MS-MS to evaluate the effect of saponification on the composition of lycopene isomers and on the recovery of lycopene from human plasma. Identical plasma samples were either extracted directly or saponified and then extracted. LC-MS-MS analysis was then carried out using C18 HPLC for total lycopene measurement or C30 HPLC for the determination of lycopene isomer ratios. The results are shown in Table 2. Since saponification decreased the total lycopene concentration by 13.8%, this processing step actually caused significant lycopene decomposition instead of enhancing recovery. Furthermore, saponification produced changes in the isomeric composition of lycopene in the extract. The relative amount of all-trans-lycopene decreased from 36.0% to 25.6% while there was a proportional increase in the relative composition of total cislycopenes. However, the increase in cis-lycopenes was not distributed uniformly across all isomers (Table 2). For example, the relative amount of 5-cis-lycopene decreased from 30.8% to 22.8% while the cis-lycopene isomers eluting between 13 and 21 min increased from 19.7% to 37.5%. The relative amounts of the cislycopenes eluting between 23 and 27 min were more constant at 13.5% and 14.1% without and with saponification, respectively. The degradation of lycopene caused by saponification for 1 h prompted us to investigate its stability during routine handling and storage. Since lycopene extracts in acetonitrile/MTBE (1:1, v/v) were held in a temperature-controlled autosampler at 4 °C during LC-MS-MS analysis, the stability of lycopene under these conditions was determined by analyzing aliquots of the same solution at 4 °C every 2 h for 16 h. These results (data not shown) indicate that the half-life of lycopene in the refrigerated autosampler is ∼16 h. In a follow-up analysis, total lycopene in a sample solution at 4 °C was measured every 12 min for 2 h and found to be relatively stable during the first 60 min (data not shown). Therefore, lycopene solutions were placed in the autosampler no more than 30 min before injection onto the LC-MS-MS. Although samples should be stored in the autosampler for less than 60 min prior to analysis, lycopene stock solutions in chloroform were found to be stable for at least one month at -80 °C.

(26) Wang, Y.; Chang, W. Y.; Prins, G. S.; van Breemen, R. B. J. Mass Spectrom. 2001, 36, 882-888. (27) Holloway, D. E.; Yang, M.; Paganga, G.; Rice-Evans, C. A.; Bramley, P. M. Free Radical Res. 2000, 32, 93-102.

CONCLUSIONS This LC-MS-MS assay is the most selective method currently available for the quantitative analysis of total lycopene and its

816 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

geometrical isomers in human plasma. Compared to previous methods,7,18 the selectivity of this new approach has been improved by incorporating selected reaction monitoring of the transition m/z 536 f 467. This method not only selects precursor ions corresponding to the appropriate molecular ion but then filters these signals so that only those ions are recorded that fragment to eliminate a terminal isoprene group. As a result, signals for R-carotene or β-carotene and their cis and all-trans isomers are not detected and cannot be confused with lycopene isomers. Because of the high selectivity of the LC-MS-MS method, the chromatographic analysis time may be reduced compared to previous methods.7,15 The application of this LC-MS-MS method to the study of lycopene stability emphasizes the need to analyze samples as quickly as possible after extraction and to store them as cold as possible until analysis (preferably at -80 °C). Saponification of

plasma samples should be avoided in order to prevent unnecessary lycopene degradation. Although the effects of light on lycopene stability and isomerization were not evaluated in this investigation, all samples should be extracted and stored in the dark. Together, these precautions will help minimize lycopene degradation and isomerization during analysis. ACKNOWLEDGMENT This research was supported by Grants R01 CA70771 from the National Cancer Institute and RR10485 from the National Center for Research Resources.

Received for review December 19, 2002.

September

9,

2002.

Accepted

AC026118A

Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

817