Rapid Determination of Terpene Lactones in Ginkgo Biloba Commercial Products by HPLC with Evaporative Light-Scattering Detection

456 LCGC NORTH AMERICA VOLUME 22 NUMBER 5 MAY 2004 www.chromatographyonline.com Rapid Determination of Terpene Lactones in Ginkgo Biloba Commercial Products by HPLC with Evaporative Light-Scattering Detection Tim Herring Alltech Associates, Inc., 2701 Carolean Industrial Drive, State College, Pennsylvania 16801, e-mail therring@alltechemail.com A rapid, sensitive, and reproducible high performance liquid chromatography gradient method has been developed for the measurement of ginkgolides A, B, C, and J, along with bilobalide, in a Ginkgo biloba commercial product. The separation was achieved in less than 14 min, employing a water methanol trifluoroacetic acid mobile phase and an evaporative light-scattering detector. No sample clean-up procedures were used with the methanol extraction of the Ginkgo biloba dietary supplement. The detection limit (signal-to-noise ratio. 5) is less than 125 ng oncolumn for each terpene lactone on a reversed-phase C18 column. Both intra- and interday reproducibility were evaluated. Four brands of standardized Ginkgo biloba herbal supplements were assessed for their terpene lactone content. This method is applicable for analyzing a Ginkgo biloba dietary supplement in capsule, tablet, or liquid form. The world s most ancient surviving tree, Ginkgo biloba, originated more than 200 million years ago. Ginkgo, sometimes known as the maidenhair tree, has an average life span of 1000 years and can grow to 20 40 m in height. A monotypic tree with broad fanshaped leaves, Ginkgo biloba has been valued for its medicinal properties for many centuries in Asia, where it is used to treat diseases such as asthma, tuberculosis, and arteriosclerosis. In western medicine, its leaves and extractions are used primarily to treat demential disorders such as concentration difficulties and memory impairment (1). Ginkgo biloba leaf extract has been shown to possess antioxidant, anti-ischemic, and neuroprotective properties (2), and has been shown to improve mental capacities in Alzheimer s patients (3). The pharmacologically active terpene lactones selectively inhibit the platelet-activating factor, preventing thrombus formation and bronchoconstriction. Bilobalide is reported to possess neuroprotective properties (4). Figure 1 shows the structures of the terpene lactones in Ginkgo biloba. In the U.S. market, Ginkgo biloba preparations were the top-selling nutraceutical in 1998 and 1999, comprising more than 20% of the total sales in the herbal nutraceutical industry (5). Today, it is still the best-selling herbal supplement worldwide. Therefore, a rapid and reproducible method for extracting and determining the active constituents in Ginkgo biloba is very desirable in industry. Experimental Reagents and chemicals: Methanol (high performance liquid chromatography [HPLC] grade) was purchased from Fisher Scientific Co. (Fair Lawn, New Jersey). Deionized water was obtained with an inhouse U.S. Filter Service Deionization water system (Pittsburgh, Pennsylvania). Trifluoroacetic acid (99%) was acquired from Aldrich Chemicals Co. (St. Louis, Missouri). Bilobalide, ginkgolide J, ginkgolide C, ginkgolide A, and ginkgolide B standards were obtained from Alltech Associates, Inc. (Deerfield, Illinois). Several brands of Ginkgo biloba supplements were bought from a local department store. Supplies for the preparation of standards and samples included 5-mL Luer tip plastic syringes; nylon 13-mm diameter, 0.45-mm porosity polypropylene-encased syringe filters; 0.45- mm porosity, 47-mm nylon filter membranes; and 4-mL HPLC vials with caps (Alltech Associates). Chromatographic Conditions: The HPLC system was assembled from an online degassing system (Alltech Associates), a model L-7100 quaternary gradient HPLC

www.chromatographyonline.com pump (Hitachi Instruments, San Jose, California), a model 7125 manual sample injector (Rheodyne, Cotati, California) with a 5- ml PEEK flex-connect sample loop (Alltech Associates), a model ELSD 2000 evaporative light-scattering detector (Alltech Associates), and a PeakSimple chromatography data system (SRI, Torrance, California). The detector was operated in the impactor off mode (no aerosol splitting). The drift tube temperature was 110 C, and the nitrogen gas flow was 3.1 L/min. A 100 mm 3 4.6 mm, 3-mm d p Alltima C18 column (Alltech Associates) was used for all separations, and they were done at ambient temperature. A binary mobile phase was used with mobile phase A being 0.1% trifluoroacetic acid in deionized water and mobile phase B consisting of 0.1% trifluoroacetic acid in methanol (6). The gradient was 10 55% B over 15 min. All runs were baseline-equilibrated before the initiation of the next run. The flow rate was 1.3 ml/min. All HPLC injections of standards and samples had an injection volume of 5 ml. All mobile phases were mixed, degassed, and filtered before use. Standard Solution Preparation: 1.017 mg of bilobalide, 1.011 mg of ginkgolide J, 1.025 mg of ginkgolide C, 1.031 mg of ginkgolide A, and 1.004 mg of ginkgolide B were weighed accurately on a microscale and quantitatively transferred into a clean, dry MAY 2004 LCGC NORTH AMERICA VOLUME 22 NUMBER 5 457 Figure 1: Structures of the terpene lactones in Ginkgo biloba. (Courtesy of the Institute for Natural Products Research.) 10-mL volumetric flask. Methanol was added to volume, and the compounds were dissolved. From this, a subsample was

458 LCGC NORTH AMERICA VOLUME 22 NUMBER 5 MAY 2004 www.chromatographyonline.com obtained from the Brand A (capsule) over three days (n 5 18, each injected in triplicate) for the reproducibility study. The terpene lactone content was also evaluated in brands B (tablet) and C (capsule) (n 5 6, each injected in triplicate). The standardized liquid extract (Brand D) was injected six times to obtain Ā. A solution of standards was injected seven times to obtain Ā for the terpene lactones. 1. The average concentration (micrograms per capsule or tablet) of each terpene lactone (for example, BB) is: g (BB in STD) A (BB in sample) 10 ml ml A (BB in STD) capsule (or tablet) where STD = standard Note: The last conversion factor (10 ml/capsule or tablet) is not used for the liquid extract because it is unaltered. 2. Converting into milligrams per gram: Figure 2: Separation of terpene lactones in (a) Brand A capsule, (b) Brand B tablet, (c) Brand C capsule, and (d) Brand D liquid extract. See the text for conditions. Peaks: 1 5 bilobalide, 2 5 ginkgolide J, 3 5 ginkgolide C, 4 5 ginkgolide A, 5 5 ginkgolide B. syringe-filtered through a 13-mm, 0.45-mm nylon-syringe filter into a 4-mL HPLC vial that was capped tightly and stored at 4 C. Sample Solutions Preparation: Four Ginkgo biloba herbal supplements were analyzed. Samples of the supplements in capsule form were prepared by emptying the content of single capsules into clean, dry, and tared 4-mL vials. Each vial was weighed accurately to determine the mass of the sample. Samples of the supplements in tablet form were prepared by thoroughly pulverizing single tablets with a mortar and pestle. The powders were transferred carefully into individual clean, dry, and tared 4- ml vials. Each vial was accurately weighed to determine the mass of the sample. All solid samples were extracted as follows: 3 ml of methanol was added to a vial containing a solid sample. The vial was tightly capped and sonicated for 10 min. Table I: Terpene lactones in standard solution (n = 7) The content of the vial was allowed to settle for 10 min, and the supernatant was transferred by pipette into a 5-mL syringe fitted with a 13-mm diameter, 0.45-mm porosity nylon syringe filter. The liquid extract was filtered into a clean, dry, 10-mL volumetric flask. For each solid sample, the extraction was repeated two more times, and the combined extracts were brought to volume with methanol (7). The samples were capped tightly and stored at 4 C. The commercially prepared liquid Ginkgo biloba extract with 2 g of leaf extracted into 1 ml of deionized water and ethanol (1:1) was analyzed directly by HPLC with no sample preparation. Quantitation: Determination of the ginkgolides (A, B, C, and J) and bilobalide (BB) present in the standardized Ginkgo biloba supplement was as follows: Average area counts (Ā) for these constituents were Bilobalide Ginkgolide J Ginkgolide C Ginkgolide A Ginkgolide B Area count 55.843 55.933 55.722 97.761 78.198 Concentration 101.7 101.1 102.5 103.1 100.4 (mg/ml) RSD 4.22% 5.08% 6.66% 2.86% 6.73% mg/capsule or tablet (avg. conc. from eq. 1) g/capsule or tablet (avg. total mass per capsule Reproducibility and Sensitivity: The precision of the method was assessed by examining sample-to-sample and day-today reproducibility. Run-to-run reproducibility also was evaluated for the standards. The limit of detection for the evaporative light-scattering detector was measured for this HPLC method by diluting the standard sample until an S/N greater than 5 was reached. Results and Discussion Evaporative light-scattering detection (ELSD) was chosen over other methods of detection for this analysis for several reasons. Independent of functional group properties, it produces a response that gives a closer estimate of the true sample mass when compared with UV detection, which depends upon the optical characteristics (extinction coefficients) of the analytes. The terpene lactones in Ginkgo biloba are poor chromophores with weak absorption in the 200 220 nm range. Refractive-index detection was considered and rejected because a stable baseline is more easily attained with ELSD, resulting in faster equilibration. In addition, ELSD is gradient compatible, whereas the refractive-index detection is not. Gas chromatography with flame ionization detection is a reliable and sensitive way to perform this analysis, but the process requires time-consuming sample prepara-

www.chromatographyonline.com dependent upon the gas flow rate used in the analysis. The optimum gas flow rate will produce the highest S/N. In this study, the gas flow rate of 3.1 L/min gave the highest S/N. Then the mobile phase evaporates in a heated stainless steel drift tube, leaving a fine mist of dried sample particles in the solvent vapor. The optimum tube temperature will produce the most stable baseline without compromising sensitivity (S/N), which MAY 2004 LCGC NORTH AMERICA VOLUME 22 NUMBER 5 459 Figure 3: Separation of terpene lactone standards. See the text for conditions. Peaks: 1 5 bilobalide, 2 5 ginkgolide J, 3 5 ginkgolide C, 4 5 ginkgolide A, 5 5 ginkgolide B. tion and derivatization, which potentially introduces more errors into the method (8). The unique detection principle of ELSD involves nebulization, evaporation, and detection of the remaining nonvolatile solute particles. Inside the nebulizer, the column effluent passes through a needle and mixes with nitrogen gas to form an aerosol consisting of a uniform distribution of droplets. The size of each droplet is The HPLC method was applied to Ginkgo biloba standardized supplements in capsule, tablet, and liquid form. in this application was 110.0 C. Detection occurs inside an optical cell, where nonvolatile sample particles interrupt a laser light beam. The scattered light is detected by a silicon photodiode, producing a signal that is sent to the analog output for data Figure 4: Limit of detection of terpene lactones. See the text for conditions. Peaks: 1 5 bilobalide, 2 5 ginkgolide J, 3 5 ginkgolide C, 4 5 ginkgolide A, 5 5 ginkgolide B. collection (9). The HPLC method was applied to Ginkgo biloba standardized supplements in capsule, tablet, and liquid form. The terpene lactone peaks were identified by direct comparison of the retention times of the peaks in the standard chromatogram with the peaks resolved in the sample chromatogram. The order of elution is as follows: bilobalide, ginkgolide J, ginkgolide C,

460 LCGC NORTH AMERICA VOLUME 22 NUMBER 5 MAY 2004 www.chromatographyonline.com Figure 5: Comparison of terpene lactones in commercial Ginkgo biloba standardized supplements. ginkgolide A, and ginkgolide B. Figure 2 shows the chromatograms of each of the samples. Figure 3 illustrates the separation of the terpene lactone standards. The methanol extraction was exhaustive, which was verified by analyzing the third extraction in which no measurable amounts of the terpene lactones were found. Table I lists the average area counts with the corresponding concentrations and the run-to-run relative standard deviations (n 5 7) for each of the terpene lactones in the standard solution. For the reproducibility study, Table II reports the Brand A sample area counts along with the standard deviations (6 SD) and relative standard deviations. The relative standard deviations for intraday samples (n 5 6, each injected in triplicate) ranged from 2.31% (bilobalide, day 3) to 9.22% (ginkgolide J, day 1). Dayto-day relative standard deviations ranged from 1.31% (ginkgolide A) to 11.09% (ginkgolide J). As depicted in Figure 4, the limit of detection (S/N. 5) was less than 125 ng on-column for each of the five analytes. Quantitation of the terpene lactones was straightforward using equation 1, and the results reported in Table III. Table IV summarizes the amount of the total terpene lactones per serving size for the Ginkgo biloba herbal supplements. A direct comparison illustrating the absolute and relative amounts of terpene lactones present in each of the four Ginkgo biloba supplements is shown in Figure 5. Both the amounts of the Ginkgo biloba standardized extract and leaf content vary widely from brand to brand, as evidenced in Figure 4. The extract portion is standardized to. 6% total terpene lactones. However, the amounts of the terpene lactones are unregulated in the leaf material. Moreover, the terpene lactone content in the leaf can vary considerably depending upon the season and the geographic location of cultivation (10). The total terpene lactones, the ginkgolides A, B, C, and J, along with bilobalide, in the leaf are not taken into Without hydrolyzation, determination of the flavonol glycosides is impossible due to the different sizes of sugars that are otherwise bound to each of the flavonol glycosides. account in the quantitation listed on the product label, often leading to an underestimation of the amount of the total terpene lactones in the Ginkgo biloba supplement when the leaf is present, as Brands A and C indicate. Brand B, which contains no leaf, contains approximately 69.1% of the total terpene lactones it is purported to have according to the label, while Brand D, the liquid extract, contains 103.6% compared to its label claim, which is within experimental error. The low amount of total terpene lactones in Brand B relative to the label claim could be attributed to label inaccuracies by the manufacturer. Perhaps less relative surface area is present after manual pulverization in Brand B, which is in a tablet form along with fillers and binders. The whole product of Brand D liquid extract is standardized, leading to a more accurate quantitation, as opposed to the capsulated brands A and C, where the leaf portion is neglected for quantitation. As a result, the total terpene lactone content differs extensively from brand to brand, as Table IV summarizes. Table II: Terpene lactones in standard solution of brand A Bilobalide Ginkgolide J Ginkgolide C Ginkgolide A Ginkgolide B Day 1 (n = 6) 27.31 6 1.32 3.47 6 0.32 21.92 6 1.39 124.88 6 7.22 77.17 6 4.67 Intraday 4.84% 9.22% 6.35% 5.78% 6.05% Day 2 (n = 6) 30.32 6 1.25 3.02 6 0.20 21.78 6 0.61 126.80 6 5.25 75.34 6 2.34 Intraday 4.12% 6.55% 2.81% 4.14% 3.10% Day 3 (n = 6) 29.79 6 0.69 2.79 6 0.15 19.67 6 0.55 123.55 6 4.00 65.80 6 1.73 Intraday 2.31% 5.53% 2.77% 3.24% 2.63% Overall (n = 3) 29.14 6 1.61 3.10 6 0.34 21.12 6 1.26 125.08 6 1.63 72.77 6 6.10 Interday 5.52% 11.09% 5.95% 1.31% 8.39%

www.chromatographyonline.com This HPLC method is 40% faster than other methods reviewed. Separation of the five terpene lactones occurs in less than 14 min. Both Polymer Laboratories (Amherst, Massachusetts) (11) and Ganzera and colleagues (7) managed to resolve the lactones within 25 min. In addition, only three terpene lactones were assessed in the Polymer Laboratories study (11). The binary mobile phase, comprising methanol, water, and trifluoroacetic acid, is much simpler in this method. A ternary gradient is used in the Polymer application, while Ganzera and his colleagues (7) used a buffer, an acid, two alcohols, and water in their mobile phase. Sample preparation is simpler when compared to other published methods. Li and Fitzloff (6) attempted to determine both the terpene lactones and flavonol glycosides in Ginkgo biloba simultaneously. Their sample preparation involved a liquid liquid extraction, subsequent evaporation, and resuspension. In addition, the flavonol glycosides were not acid hydrolyzed into their respective aglycones. Without hydrolyzation, determination of the flavonol glycosides is impossible due to the different sizes of sugars that are otherwise bound to each of the flavonol glycosides, resulting in numerous peaks with varying retention times for each MAY 2004 LCGC NORTH AMERICA VOLUME 22 NUMBER 5 461 Table III Determination of Terpene Lactones (TLs) in Ginkgo Biloba Supplements Brand A Bilobalide Ginkgo. J Ginkgo.C Ginkgo. A Ginkgo. B Total TLs mg/capsule 541.6 55.9 388.5 1319.1 934.3 3239.4 mg/g 1.40 0.145 1.06 3.41 2.42 8.43 %(w/w) 0.14% 0.014% 0.11% 0.34% 0.24% 0.84% Brand B Bilobalide Ginkgo. J Ginkgo.C Ginkgo. A Ginkgo. B Total TLs mg/tablet 44.4 15.9 149.1 1480.5 798.1 2488.0 mg/g 0.14 0.050 0.47 4.67 2.52 7.85 %(w/w) 0.01% 0.005% 0.05% 0.47% 0.25% 0.79% Brand C Bilobalide Ginkgo. J Ginkgo.C Ginkgo. A Ginkgo. B Total TLs mg/capsule 790.1 91.7 639.4 1923.2 889.3 4333.7 mg/g 2.88 0.335 2.33 7.02 3.25 15.82 %(w/w) 0.29% 0.034% 0.23% 0.70% 0.32% 1.57% Brand D Bilobalide Ginkgo. J Ginkgo.C Ginkgo. A Ginkgo. B Total TLs mg/ml 4993.1 511.6 2320.8 1644.5 471.6 9941.6 mg/g 5.24 0.54 2.44 1.73 0.50 10.5 %(w/w) 0.52% 0.05% 0.24% 0.17% 0.05% 1.03% flavonol glycoside (12). The limit of detection for the terpene lactones using ELSD is 125 ng on-column for the terpene lactones. This is much lower than both Polymer Laboratories (11) and Ganzera and colleagues (7) limits of detection, which were 250 and 203 ng on-column, respectively. Conclusion The total terpene lactones present in the Ginkgo biloba supplements analyzed ranged from 0.79% (w/w) for Brand B to 1.57% (w/w) for Brand C. The concentrations of the individual terpene lactones varied from 0.005% (w/w) for ginkgolide J in Brand B to 0.70% (w/w) for ginkgolide A in

462 LCGC NORTH AMERICA VOLUME 22 NUMBER 5 MAY 2004 www.chromatographyonline.com Table IV: Comparison of total terpene lactones (TTLs) Brand Serving Size (SS) ActualTTLs (µg/ss) Label Claim TTLs (µg/ss) A 2 capsules 6478.8 2400.0 B 1 tablet 2488.0 3600.0 C 2 capsules 8667.4 7200.0 D 1 ml 9941.6 9600.0 Brand C. The HPLC method described here allows analysis of Ginkgo biloba much more rapidly than has been reported previously without compromising sensitivity and reproducibility. As a result, routine quality control in high-throughput laboratories within the nutraceutical industry can be performed much more efficiently. Acknowledgements The author would like to thank the following employees of Alltech Associates: Michael Li, Stephen Liu, Dennis McCreary, Sharon McKinley, Gary Nixon, Cary Thrall, Mark Waksmonski, Bob Wiedemer, and Bob Ziegler. He would also like to thank all of Alltech Associates for their invaluable contributions to this project. References (1) A.Y. Leung and S. Foster, Encyclopedia of Common Natural Ingredients Used in Food, Drugs, and Cosmetics, 2nd ed. (Wiley & Sons, Hoboken, New Jersey, 1996). (2) C. Mar and S. Bent, Western J. Med. 171, 168 171 (1999). (3) P.L. Le Bars, M.M. Katz, N. Berman, T.M. Itil, A.M. Freedman, and A.F. Schatzberg, J. Am. Med. Assoc. 278, 1327 1332 (1997). (4) P.F. Smith, C.L. MacLennan, and J. Darlington, J. Ethnopharmacol. 50, 131 139 (1996). (5) M. Blumenthal, Herbalgram 47, 64 65 (1999). (6) W. Li and J. Fitzloff, J. Pharm. Biomed. Anal. 30, 67 75 (2002). (7) M. Ganzera, J. Zhao, and I. Khan, Chem. Pharm. Bull. 49, 1170 1173 (2001). (8) E. Lolla, A. Paletti, and F. Peterlongo, Fitoterapia 69, 513 519 (1998). (9) Alltech Associates, Inc., Alltech ELSD 2000 Evaporative Light Scattering Detector Operating Manual (2001). (10) T.A. Van Beek, H.A. Scheeren, T. Rantio, W.C. Melger, and G.P. Lelyveld. J. Chromatogr. 543, 375 387 (1991). (11) Polymer Laboratories, Inc., The Application Notebook, LCGC, 44 45 (February 2003). (12) D. McCreary, personal communication (2002). Tim Herring is with Alltech Associates, Inc., 2701 Carolean Industrial Drive, State College, Pennsylvania 16801, e-mail therring@alltechemail.com.