Determination of Flavonol Aglycones in Ginkgo biloba Dietary Supplement Crude Materials and Finished Products by High-Performance Liquid Chromatography: Single Laboratory Validation

Abstract

A single laboratory validation (SLV) was completed for a method to determine the flavonol aglycones quercetin, kaempferol, and isorhamnetin in Ginkgo biloba products. The method calculates total glycosides based on these aglycones formed following acid hydrolysis. Nine matrixes were chosen for the study, including crude leaf material, standardized dry powder extract, single and multiple entity finished products, and ethanol and glycerol tinctures. For the 9 matrixes evaluated as part of this SLV, the method appeared to be selective and specific, with no observed interferences. The simplified 60 min oven heating hydrolysis procedure was effective for each of the matrixes studied, with no apparent or consistent differences between 60, 75, and 90 min at 90°C. A Youden ruggedness trial testing 7 factors with the potential to affect quantitative results showed that 2 factors (volume hydrolyzed and test sample extraction/hydrolysis weight) were the most important parameters for control during sample preparation. The method performed well in terms of precision, with 4 matrixes tested in triplicate over a 3-day period showing an overall repeatability (relative standard deviation, RSD) of 2.3%. Analysis of variance testing at α = 0.05 showed no significant differences among the within- or between-group sources of variation, although comparisons of within-day (S_w), between-day (S_b), and total (S_t) precision showed that a majority of the standard deviation came from within-day determinations for all matrixes. Accuracy testing at 2 levels (approximately 30 and 90% of the determined concentrations in standardized dry powder extract) from 2 complex negative control matrixes showed an overall 96% recovery and RSD of 1.0% for the high spike, and 94% recovery and RSD of 2.5% for the low spike. HorRat scores were within the limits for performance acceptability, ranging from 0.4 to 1.3. Based on the performance results presented herein, it is recommended that this method progress to the collaborative laboratory trial.

In support of the initiative of the National Institutes of Health/Office of Dietary Supplements (NIH/ODS), U.S. Food and Drug Administration (FDA), and AOAC INTERNATIONAL to develop well-validated methods for botanicals and dietary supplements, an expert review panel (ERP) convened to determine the most promising method(s) for further study to determine total flavonol glycosides in Ginkgo biloba (L.) crude material and finished products. After comprehensive review of available and appropriate methods, the ERP chose 2 methods for further study, 1 developed by the NSF International-Institute for Nutraceutical Advancement (NSF–INA; ¹), and another developed at Midwest Research Institute (MRI) in support of the National Institute of Environmental Health Sciences (NIEHS) National Toxicology Program (NTP; ntp-server.niehs.nih.gov). The NSF–INA method was chosen as the overall quantitative method, with sample preparation procedures based on the NTP-MRI method.

Following protocol development and review, the resulting method was evaluated to conclude suitability for the determination of the flavonol aglycones quercetin, kaempferol, and isorhamnetin in Ginkgo biloba products. The 3 primary aglycone bases (Figure 1) are determined following hydrolysis of the flavonol glycosides with HC1 and heating at 90°C. Crude plant material, dry powder and liquid extracts, and finished product dosage forms are extracted and hydrolyzed with ethanol–water–HCl. The analytes are separated by liquid chromatography (LC) and measured by ultraviolet (UV) absorbance at 370 nm against external standards of quercetin, kaempferol, and isorhamnetin. Total flavonol glycoside concentration is then calculated using individual conversion factors for the 3 aglycones.

Figure 1.

Structures of predominant flavonol aglycones following hydrolysis of Ginkgo biloba products.

For the single laboratory validation (SLV), the method was evaluated using 9 matrixes representative of Ginkgo products found in the marketplace, including crude leaf material, standardized dry powder extract, single and multiple entity tablets and capsules, and ethanol and glycerol tinctures (Table 1). In-depth evaluation of hydrolysis duration, ruggedness, precision, and accuracy testing was conducted on these selected matrixes in order to determine potential suitability of the method for further collaborative study.

Table 1.

Ginkgo biloba products evaluated for the flavonol aglycone single laboratory validation

Test sample	Product name	Source	Product description	Label claima	Tab/Cap/serving weight
Gb-SLV-1	Ginkgo leaves TBC	Martin Bauer	Dried and ground leaf material, 500 g/package	>0.5% (flavonol glycosides)	NAb
Gb-SLV-2	Ginkgo dry extract	Indena	Dry powder standardized extract, 200 g/package	240 mg/g (flavonol glycosides)	NA
Gb-SLV-3	Ginkgold	Nature’s Way	Tablets containing Ginkgo standardized extract	60 mg/Tab (Ginkgo extract) containing 24% flavonol glycosides	0.358 g/Tab
Gb-SLV-4	Ginkoba-Mental Sharpness	Pharmaton	Tablets containing Ginkgo standardized extract (50:1)	40 mg/Tab (Ginkgo extract) consisting of 50:1 standardized extract	0.277 g/Tab
Gb-SLV-5	Ginkogin	Blairex Laboratories	Tablets containing Ginkgo standardized extract, ginseng root extract, and garlic extract	60 mg/Tab (Ginkgo extract) containing 27% flavonol glycosides, 100 mg/Tab (ginseng extract), 300 mg/Tab (garlic extract)	1.361 g/Tab
Gb-SLV-6	Ginkgo Awareness	Planetary Formulas	Tablets containing Ginkgo standardized extract, gotu kola leaf extract, and proprietary blend of 7 additional herbs	20 mg/Tab (Ginkgo extract containing 24% flavonol glycosides), 32 mg/Tab (gotu kola extract), 448 mg/Tab (proprietary blend with 7 herbs)	0.811 g/Tab
Gb-SLV-7	Ginkgo Phytosome	PhytoPharmica	Soft gel capsules containing Ginkgo leaf extract and phosphatidylcholine (1:2)	80 mg/Cap (Ginkgo Leaf Phytosome: 1 part Ginkgo extract containing 24% flavonol glycosides, and 2 parts phosphatidylcholine)	0.675 g/Cap (with shell)
Gb-SLV-8	Ginkgo leaf, alcohol-free	Gaia Herbs	Ginkgo leaf extracted in 50–60% vegetable glycerol and spring water	NA	0.795 g/serving, or 0.75 mL/serving
Gb-SLV-9	Ginkgo	Herb Pharm	Ginkgo leaf extracted in ethanol (57–67%) with distilled water and vegetable glycerol	NA	0.624 g/serving, or 0.75 mL/serving

^aTab = Tablet, Cap = Capsule.

^bNA = Specific information regarding the flavanol glycoside concentration was not present in the label claim.

METHOD

Apparatus

1. LC system.—PerkinElmer (Shelton, CT) Series 200 HPLC system with pump, autosampler, and diode array detector (or equivalent). The data were acquired on the TurboChrom™ data system (Wellesley, MA). LC operation conditions: column temperature, 35°C; flow rate, 1.0 mL/min; injection volume, 10 μL; detection, 370 nm; run time, 35 min.

2. LC column.—Phenomenex Prodigy^® ODS(−3), 5 (μm, 100 A, 4.6 × 250 mm (Phenomenex, Torrance, CA), Part No. 00G-4097-E0.

3. Ultrasonic cleaner.—Fisher Scientific (www.fishersci.com) or equivalent.

4. Balance.—Calibrated, readability to 0.1 mg.

5. Grinder.—Waring (New Hartford, CT) blender or equivalent.

6. Syringe, glass, gas-type.—Class A, 0.1, 0.25, 1.0, and 2.5 mL.

7. Volumetric flasks.—Class A, various sizes.

8. Graduated cylinders.—Class A, various sizes.

9. Glass pipets (volumetric).—Class A, various sizes.

10. Serum vials.—30 mL (1 oz) serum bottles, mouth id × od = 13 × 20 mm, Wheaton No. 223743 (Wheaton Science Products, Millville, NJ).

11. Vial caps.—Standard aluminum crimp-cap seals, 20 mm od, Wheaton No. 224183-01.

12. Silicone septa.—Teflon-faced silicone septa, 20 mm od, Wheaton No. 224173.

13. Syringe filters.—0.45 (μm, PVDF membrane, Gelman 4450 or equivalent, available through Fisher Scientific.

14. Syringe.—3cc, plastic, disposable, with Luer-Lok™ tip, from Becton Dickinson Co., available through VWR International (www.vwrsp.com).

15. Sieve.—No. 60 mesh.

16. Laboratory oven.—Able to maintain a temperature of 80° to 100°C.

17. Amber containers.—Amber HDPE wide mouth with screw caps, and amber glass round with screw caps, or equivalent.

18. Autosampler vial.—12 × 32, mm 2 mL amber or clear glass with screw caps.

Reagents

(a) Mobile phase

Isocratic. Methanol–0.85% phosphoric acid (1 + 1, v/v). For the preparation of 1 L mobile phase, transfer methanol into a 500 mL graduated cylinder and fill to volume. Transfer ca 250 mL water (HPLC grade or nanopure) to another 500 mL graduated cylinder, add 5 mL phosphoric acid (85–87%, reagent grade) by serological pipet, and dilute to volume with water. Combine the contents of both graduated cylinders into a 1 or 2 L glass mobile phase container and degas before use. Label Mobile Phase. (HPLC grade methanol was purchased from J.T. Baker, Phillipsburg, NJ, and phosphoric acid, 85% ACS grade, from Fisher Scientific, Fair Lawn, NJ.)

(b) Diluent

Ethanol–water–hydrochloric acid (50 + 20 + 8, v/v/v). For the preparation of ca 150 mL, transfer ethanol to a 100 mL graduated cylinder and fill to volume. Transfer 40 mL water (deionized) by graduated cylinder or volumetric pipet to a 200 mL flask, and add the 100 mL ethanol (EtOH). Carefully pipet 16 mL HCl (37%, reagent grade) to this solution and mix thoroughly. Label Diluent. (95% EtOH, ACS spectrophotometric grade, was purchased from Sigma-Aldrich, St. Louis, MO, and hydrochloric acid, trace metal grade, from Fisher Scientific).

(c) Reference standards

Quercetin dihydrate, kaempferol, and isorhamnetin (≥98% pure when possible; available from ChromaDex, Santa Ana, CA; Sigma-Aldrich, MO; and PhytoLab GmbH & Co. KG, Hamburg, Germany).

(d) Preparation of standard solutions

Prepare a stock solution in dimethyl sulfoxide (DMSO) and methanol at a final concentration of ca 0.38 mg/mL quercetin dihydrate, 0.38 mg/mL kaempferol, and 0.08 mg/mL isorhamnetin by accurately weighing ca 10.5 mg (±0.06 mg) quercetin dihydrate, ca 9.4 mg kaempferol (±0.06 mg), and ca 2.0 mg (±0.02 mg) isorhamnetin into a 25 mL volumetric flask. (Note: quercetin dihydrate should be converted to quercetin by calculation):

$$\text{Quercetin}\text{dihydrate}(\text{mg})\times 302/338=\text{quercetin}(\text{mg})$$ (1)

(DMSO OmniSolv, high-purity solvent, was purchased from EMD, San Diego, CA.)

Dissolve the standards in 2–3 mL DMSO and sonicate at room temperature for 1–5 min to dissolve solids. Dilute to volume with methanol and mix thoroughly by inversion. Label Stock Solution. Dilute the stock solution with methanol into volumetric flasks to create 5 additional calibration graph concentrations ranging from ca 10 to 150 (μg/mL for quercetin and kaempferol and ca 2 to 30 (μg/mL for isorhamnetin as shown in Table 2.

Table 2.

Preparation of solutions for the calibration graph

Solution No.	Volume of stock standard, mL	Final volume, mL	Quercetin, μg/mL	Kaempferol, μg/mL	Isorhamnetin, μg/mL
1	0.25	10.0	9.4	9.4	2.0
2	0.5	10.0	18.8	18.8	4.0
3	1.0	10.0	37.6	37.6	8.0
4	1.0	5.0	75.2	75.2	16.0
5	2.0	5.0	150.4	150.4	32.0

Label each appropriately with solution Nos. 1–5. Store solutions at 4°C, and bring to room temperature before use.

System Suitability

(a) System suitability

Equilibrate the LC system with the mobile phase at least 30 min, and inject 6 consecutive replicates of standard solution No. 3. Inject diluent blank after the completion of the 6 replicate injections to ensure there is no carryover or interfering peaks. Determine the peak area and retention time of each reference standard. The relative standard deviation (RSD) for the peak areas should be ≤1.5%, and the resolution between kaempferol and isorhamnetin should be ≥1.0:

$$\text{Peak}\text{resolution}=\frac{\mathrm{\Delta }{t}_r}{W_{av}}=\frac{\mathrm{\Delta }{v}_r}{W_{av}}$$ (2)

where Δt_r or Δv_r is the separation between peaks (in units of retention time), and W_av, is the average width of the 2 peaks in corresponding units (peak width measured at the base; ²).

(b) Peak asymmetry

Calculate peak asymmetry (A_s) from 10% peak height; this must be no more than 1.5 for all components in the standard solution chromatogram:

$${\text{A}}_{\text{s}}=\text{B}/\text{A}$$ (3)

where A = the distance from the peak front to the dropline and B = the distance from the dropline to the peak tail.

Calibration

(a) Calibration graph

Run separate 5-point calibration graphs for each of the 3 analytes from freshly prepared standard solutions. Calibration graphs will initially be run at the beginning, midpoint (following ca 12 test samples), and end of the analytical run. Following the initial determination of calibration graph precision from the beginning calibration graph (see below), a check standard (standard solution No. 2) may be run in place of the entire calibration graph at the middle and end of the analytical run for the purposes of this validation.

(b) Linearity

Calculate the relationship between the peak area response and concentration using linear regression and appropriate software, with the LC conditions given above. The determination coefficient, R², for each standard compound must be greater than 0.999.

(c) Precision and stability

Run calibration graphs 3 times (once at the beginning, middle, and end of the analytical run). The middle and ending calibration graphs should not differ from the beginning calibration graph by more than 5%, calculated by comparing the slope of the beginning graph to the slopes of the midpoint and ending calibration graphs. Approximately 12 test samples, not including the blank, may be run following the beginning and middle calibration graphs.

Validation Samples

Nine samples of Ginkgo biloba were chosen and approved by the study monitors for this validation work. Samples were chosen to represent common Ginkgo products present in the marketplace, including crude leaf material, standardized powdered extracts, single and multiple component dosage forms in tablets and capsules, and glycerol and ethanol tinctures. Product names, descriptions, forms, and label claims are presented in Table 1. (Note: All information regarding lot numbers, batch numbers, and certificates of analyses, where applicable, in addition to the location of retention samples, are available from AOAC INTERNATIONAL.)

Test Samples Preparation

(a) Crude material (dried leaves of Ginkgo biloba)

Mix the original laboratory test sample and transfer ca 250 g to a tightly sealed container for storage as reference material. Grind ca 10 g in a Waring blender to create a powder consistency capable of passing through a No. 60 sieve. Record the weight and transfer the contents into an amber container with a screw cap and mix thoroughly.

(b) Dry powder extract of Ginkgo biloba

Mix the original laboratory test sample and transfer ca 100 g to a tightly sealed container for storage as reference material. Record the weight, transfer ca 10 g into an amber container with a screw cap, and mix thoroughly.

(c) Capsules containing dry powder extract of Ginkgo biloba (single or multiple component)

Weigh 20 capsules, empty the contents, and weigh the empty capsules. Calculate the net content/capsule. Record the weight and transfer the contents into an amber container with a screw cap, and mix thoroughly.

(d) Tablets containing dry powder extract of Ginkgo biloba (single or multiple component)

Weigh 20 tablets and record weight. Grind the tablets with a grinder to create a powder consistency capable of passing through a No. 60 sieve. Record weight and transfer contents into an amber container with screw cap and mix thoroughly.

(e) Tinctures of hydroethanolic (with and without glycerol) extracts of dried or fresh Ginkgo biloba leaves

Combine two, 2 oz bottles of tincture into an amber glass round container with a screw cap and mix thoroughly by inversion.

Test Samples Hydrolysis Procedure

(a) Crude material (dried leaves of Ginkgo biloba)

Weigh 200 mg (±5.0 mg) crude material into a 30 mL clear glass serum vial. Record the weight. Add 12 mL diluent, sonicate for 5 min, and seal using an aluminum crimp cap and Teflon septum. Hydrolyze the contents by placing in an oven set at 90°C (±2°C) for 60 min (±3 min). Remove and cool to ambient temperature. Transfer the contents into a 25 mL volumetric flask, rinse and dilute to volume with methanol, and mix well. Filter a portion of the solution [0.45 μm polyvinylidene fluoride (PVDF)] into an autosampler vial.

(b) Dry powder extract of Ginkgo biloba, including capsules and tablets containing dry powder extract of Ginkgo biloba)

Weigh 15 mg (±1.5 mg) dry powder extract into a 30 mL clear glass serum vial. Record the weight. Add 5 mL diluent into the vial and sonicate for 5 min. Seal using an aluminum crimp cap and Teflon septum. Hydrolyze the contents by placing in an oven set at 90°C (±2°C) for 60 min (±3 min). Remove and cool to ambient temperature. Transfer the contents into a 10 mL volumetric flask, rinse and dilute to volume with methanol, and mix well. Filter a portion of the solution (0.45 μmPVDF) into an autosampler vial.

(c) Soft gel capsules containing extract of Ginkgo biloba

Weigh a soft gel capsule and place into a 30 mL clear glass serum vial. Record the weight. Add 10 mL diluent into the vial and sonicate for 40–60 min (adjust this time if necessary in order to achieve complete dissolution of the capsule). Seal using an aluminum crimp cap and Teflon septum. Hydrolyze the contents by placing in an oven set at 90°C (±2°C) for 60 min (±3 min). Remove and cool to ambient temperature. Transfer the contents into a 25 mL volumetric flask, rinse and dilute to volume with methanol, and mix well. Filter a portion of the solution (0.45 μm PVDF) into an autosampler vial.

(d) Capsules and tablets containing multiple components including dry powder extract of Ginkgo biloba

Weigh powder from tablets (or capsules) equivalent to 15 mg (±1.5 mg) dry powder extract into a 30 mL clear glass serum vial. Record actual weight. Add 5 mL diluent and sonicate for 5–15 min. Seal using an aluminum crimp cap and Teflon septum. Hydrolyze the contents by placing in an oven set at 90°C (±2°C) for 60 min (±3 min). Remove and cool to ambient temperature. Transfer the contents into a 10 mL volumetric flask, rinse and dilute to the volume with methanol, and mix well. Filter a portion of the solution (0.45 μm PVDF) into an autosampler vial.

(e) Tinctures of hydroethanolic (with and without glycerol) extracts of dried or fresh Ginkgo biloba leaves

Transfer 1 mL tincture into a 30 mL clear glass serum vial. Add 5 mL diluent and mix well. Seal using an aluminum crimp cap and Teflon septum. Hydrolyze the contents by placing in an oven set at 90°C (±2°C) for 60 min (±3 min). Remove and cool to ambient temperature. Transfer the contents into a 10 mL volumetric flask, rinse and dilute to volume with methanol, and mix well. Filter a portion of the solution (0.45 μm PVDF) into an autosampler vial.

Calculations

Flavonol glycoside determination

Total flavonol glycoside concentration is determined from the combined peak areas of quercetin, kaempferol, and isorhamnetin at 370 nm. Aglycone conversion factors were derived from previous work (3). Flavonol glycosides present as coumaroyl derivatives of the 3 aglycones are determined as either percentage individual glycoside (%, Equation 4) or individual glycoside/tablet (μg, Equation 5):

$$\%\text{Individual}\text{glycoside}(\%\text{w}/\text{w})=(\text{C}\times \text{FV}\times \text{F}\times 100)/\text{W}$$ (4)

where C = aglycone concentration (μg/mL) determined from standard curve; FV = final volume; F = conversion factor (2.504 for quercetin, 2.588 for kaempferol, and 2.437 for isorhamnetin); and W = weight of test sample (mg).

$$\text{Individual}\text{glycoside}/\text{tablet}(\mu \text{g})=[(\text{C}\times \text{FV}\times \text{F})/\text{W}]\times \text{T}$$ (5)

where T = average tablet weight (mg); (other parameters as described in Equation 4).

Test Samples Hydrolysis Experiment

Oven heating experiment

Evaluate hydrolysis times of 60, 75, and 90 min for each matrix in order to determine the optimal duration of hydrolysis at 90°C. Prepare test samples according to the previous section [see Test Samples Hydrolysis Procedure (a) through (e)]. Prepare and evaluate test samples from each matrix at each heating time on a single day.

Ruggedness Testing

Youden ruggedness trial

Design a ruggedness trial for 7 factors with the potential to affect quantitative results, including heating time, temperature, acid concentration in the diluent, volume hydrolyzed, column temperature, flow rate, and test sample weight. Prepare 2 replicates of test sample Gb-SLV-2 for each experiment. Individual calibration graphs are necessary to run when flow rates or column temperatures are varied. Evaluate samples and chart results based on the parameters and equations presented in Table 3.

Table 3.

Youden ruggedness trial design and calculations

Factor	High	Low
Heating time, MIN	65 (A)	55 (a)
Temperature, °C	95 (B)	80 (b)
Acid concentration (EtOH-H2O-HCl, v/v/v)	50 + 20 +9 (C)	50 + 20 + 7 (c)
Volume hydrolyzed, mL	6(0)	4(d)
Column temperature, °C	40 (E)	30 (e)
Flow rate, mL/min	1.2(F)	0.8 (f)
Test sample weight, dry extract, mg	17(G)	12(g)
Experiment No.	Combination	Measurement
1	ABCDEFG	X1
2	ABcDefg	X2
3	AbCdEfg	X3
4	AbcdeFG	X4
5	aBCdeFg	X5
6	aBcdEfG	X6
7	abCDefG	X7
8	abcDEFg	X8
Effect	Equation
Effect of “A” and “a”	= [(X1 + X2 + X3 + X4]/4] − [(X5 + X6 + X7 + X8]/4]
Effect of “B” and “b”	= [(X1 + X2 + X5 + X6]/4] − [(X3 + X4 + X7 + X8]/4]
Effect of “C” and “c”	= [(X1 + X3 + X5 + X7]/4] − [(X2 + X4 + X6 + X8]/4]
Effect of “D” and “d”	= [(X1 + X2 + X7 + X8]/4] − [(X3 + X4 + X5 + X6]/4]
Effect of “E” and “e”	= [(X1 + X3 + X6 + X8]/4] − [(X2 + X4 + X5 + X7]/4]
Effect of T” and T	= [(X1 + X4 + X5 + X8]/4] − [(X2 + X3 + X6 + X7]/4]
Effect of “G” and “g”	= [(X1 + X4 + X6 + X7]/4] − [(X2 + X3 + X5 + X8]/4]

Test Material Precision and Accuracy Testing

(a) Repeatability standard deviation

Assay crude leaf material (Gb-SLV-1), dry powder extract (Gb-SLV-2), dry powder extract in a single component tablet (Gb-SLV-3), and multicomponent soft gel capsule (Gb-SLV-7) for quercetin, kaempferol, and isorhamnetin using the following design: r = number of replicate test preparations within day (3 per matrix); n = number of test sample matrixes (4 matrixes); d = number of days (3 days).

(b) Negative control recovery

There are no extant species in the genus Ginkgo other than biloba, and given the biosynthetic pathway and ubiquity of the flavonol glycosides, no leaf material is likely to be devoid of these compounds. The negative controls were developed from ingredients in the chosen matrixes. Two negative controls were selected to mimic the main excipients in 2 finished products: (1) 95% methyl cellulose and 5% starch and (2) soybean oil.

Fortify 30 mg portions of material (capable of passing through No. 60 sieve mesh size) with the Stock Solution at ca 30 and 90% of the determined concentrations in Ginkgo dry powder extract (Gb-SLV-2). Prepare all test samples and blanks as described previously (Test Samples Hydrolysis Procedure, section (b) for Gb-SLV-2).

Results and Discussion

System Suitability

System suitability (instrument precision) showed that the LC system used for this validation study met the acceptance criteria for quercetin, kaempferol, and isorhamnetin with an RSD <1.5% for all analytes (1.4,1.3, and 1.3%, respectively).

An adequate resolution of 1.4 between kaempferol and isorhamnetin was achieved, and peak asymmetry was calculated to be less than 1.5 (quercetin, kaempferol, and isorhamnetin = 1.4).

Standard Curve Precision

The R² for each standard compound for a 5-point calibration graph (approximately 10 to 150 μg/mL for quercetin and kaempferol and approximately 2 to 30 μg/mL for isorhamnetin) was greater than 0.999, with a mean (n = 3) of 0.99994, 0.99998, and 0.99999 for quercetin, kaempferol, and isorhamnetin, respectively. The calibration graph for each aglycone standard covered the analytical range for each of the 9 matrixes tested, and precision testing showed that the slopes of triplicate calibration graphs differed by less than 1.5%.

Test Samples Hydrolysis Results

Glycoside hydrolysis evaluation of each test material at 90 °C showed that there was no notable or consistent difference in total glycoside results between 60, 75, and 90 min (Table 4). No degradants or interfering peaks were determined to be present in any matrix at any hydrolysis time (Figure 2). With the exception of 2 matrixes (test materials Gb-SLV-4 and Gb-SLV-5), RSDs for grouped hydrolysis times (RSD of 60, 75, and 90 min) were less than 4%. For those matrixes with grouped RSDs of >4%, results showed that the 60 min hydrolysis specified in the method consistently gave the greatest total glycoside results (Table 4).

Table 4.

Hydrolysis testing results (mean, SD^a, and RSD) for 3 durations for each of the 9 matrixes evaluated

Test sampleb	Hydrolysis time, min	Tab/Cap/Sample weight, gc	Total flavonol glycosidesd	Sample weight, mg	Volume, mL	Glycosides concn, μg/mL	Resultse, mg/g, mg/Tab, or mg/Cap)
Gb-SLV-1	60	1	>5 mg/g	195.8	25	72.1	9.2
	75	1	>5 mg/g	194.9	25	68.3	8.8
	90	1	>5 mg/g	199.6	25	72.5	9.1
Mean						71.0	9.0
SD						2.3	0.2
RSD						3.2	2.2
Gb-SLV-2	60	1	240 mg/g	15.8	10	370.8	234.7
	75	1	240 mg/g	15.4	10	352.5	228.9
	90	1	240 mg/g	15.2	10	342.5	225.3
Mean						355.3	229.6
SD						14.4	4.7
RSD						4.1	2.0
Gb-SLV-3	60	0.358	14.4 mg/Tab	60.1	10	273.4	16.3
	75	0.358	14.4 mg/Tab	61.0	10	276.7	16.2
	90	0.358	14.4 mg/Tab	61.9	10	271.7	15.7
Mean						273.9	16.1
SD						2.5	0.3
RSD						0.9	1.9
Gb-SLV-4	60	0.277	9.6 mg/Tab	105.8	10	268.2	7.0
	75	0.277	9.6 mg/Tab	105.9	10	236.2	6.2
	90	0.277	9.6 mg/Tab	106.6	10	257.8	6.7
Mean						254.1	6.6
SD						16.3	0.4
RSD						6.4	6.1
Gb-SLV-5	60	1.361	16.2 mg/Tab	341.4	10	227.9	9.1
	75	1.361	16.2 mg/Tab	341.7	10	192.6	7.7
	90	1.361	16.2 mg/Tab	342.8	10	197.6	7.8
Mean						206.0	8.2
SD						19.1	0.8
RSD						9.3	9.8
Gb-SLV-6	60	0.811	4.8 mg/Tab	609.0	10	381.4	5.1
	75	0.811	4.8 mg/Tab	601.2	10	391.6	5.3
	90	0.811	4.8 mg/Tab	606.8	10	407.2	5.4
Mean						393.4	5.3
SD						13.0	0.2
RSD						3.3	3.8
Gb-SLV-7	60	0.675	6.4 mg/Cap	678.4	25	266.9	6.6
	75	0.675	6.4 mg/Cap	674.3	25	258.9	6.5
	90	0.675	6.4 mg/Cap	666.8	25	256.1	6.5
Mean						260.6	6.5
SD						5.6	0.1
RSD						2.1	1.5
Gb-SLV-8	60	0.795	NA mg/g	795.0	10	162.7	1.6
	75	0.795	NA mg/g	795.0	10	155.3	1.6
	90	0.795	NA mg/g	795.0	10	154.5	1.5
Mean						157.5	1.6
SD						4.5	0.0
RSD						2.9	0.0
Gb-SLV-9	60	0.6236	NA mg/g	887.3	10	417.9	2.9
	75	0.6236	NA mg/g	885.4	10	423.4	3.0
	90	0.6236	NA mg/g	892.8	10	405.0	2.8
Mean						415.4	2.9
SD						9.4	0.1
RSD						2.3	3.4

^aSD = Standard deviation.

^bFull descriptions of test samples are provided in Table 1.

^cTab = Tablet, Cap = capsule.

^dTotal flavonol glycosides = amount of Ginkgo extract in product × percentage of flavonol glycosides in extract (see Table 1 label claim), presented in mg/g, mg/Tab, mg/Cap or NA where appropriate.

^eResults = Total determined flavonol glycosides presented in mg/g or mg/Tab, corresponding to total flavonol glycosides.

^fNA = Specific information regarding flavonol glycoside concentration was not present on the label claim (Table 1).

Figure 2.

Representative aglycone chromatograms of hydrolyzed Ginkgo biloba products evaluated in the SLV. (Q = quercetin; K = kaempferol; I = isorhamnetin.) Test sample chromatograms from top to bottom: Gb-SLV-1, Gb-SLV-8, Gb-SLV-4, Gb-SLV-7, Gb-SLV-5, Gb-SLV-2, Gb-SLV-6, Gb-SLV-3, Gb-SLV-9, and reference standard mixture. (Full descriptions of test samples are provided in Table 1.)

As shown in Table 4, the average total glycoside content of 4 of the test products (Gb-SLV-2, 3, 6, and 7) ranged from 96 to 111% of their label claim. However, 2 of the test products (Gb-SLV-4 and 5) were determined to contain an average of 68 and 51%, respectively, of their label claims for total glycosides. While the exact reason for these discrepancies are currently unknown, there are 2 possibilities proposed at this time: (1) the lower observed concentration in Gb-SLV-4 could have been due to the quality of the crude materials used in the sample, incorrect sample formulation, or both; and (2) for Gb-SLV-5, it is possible that the high concentration of ginseng saponins (100 mg/tablet listed on the label) consumed acid during sample preparation and hydrolysis, decreasing the effective hydrolysis of the glycosides. For Scenario 1 (Gb-SLV-4), the dry extract starting material and overall product formulation were not evaluated as part of this SLV With respect to Scenario 2 (Gb-SLV-5), further method development beyond the scope of this SLV may be necessary for products containing ginseng saponins.

Ruggedness Trial Results

For the 7 tested factors in the Youden (4) ruggedness trial (“high” and “low” parameters for heating time, hydrolysis temperature, HCl concentration, hydrolyzed volume, column temperature, flow rate, and test sample weight), 2 factors were found to have the greatest effects on total glycoside results (Figure 3). The method was determined to be most sensitive to changes in factors “D/d” (volume hydrolyzed 6 or 4 mL) and “G/g” (test sample weight 17 or 12 mg) based on the calculated difference from zero compared to the other factors. Factors “E/e” (column temperature 40° and 30°C) and “B/b” (oven temperature 95° and 80°C) affected the results to a lesser relative extent. The remaining factors “A/a” (65 and 55 min heating time for hydrolysis), “C/c” (HCl concentration of 11 and 9% in the diluent), and “F/f” (flow rates of 1.2 and 0.8 mL/min) were determined to be factors least likely to substantially affect quantitation with slight variation.

Figure 3.

Results of Youden ruggedness trial. (“A” and “a” = heating duration of 65 and 55 min; “B” and “b” = heating temperature of 95° and 80°C; “C” and “c” = acid concentration of EtOH-H₂O-HCl of (50 + 20 + 9, v/v/v) and (50 + 20 + 7, v/v/v); “D” and “d” = volume hydrolyzed of 6 and 4 mL; “E” and “e” = column temperature of 40° and 30°C; “F” and “f” = flow rate (mL/min) of 1.2 and 0.8; “G” and “g” = test sample weight (dry extract) of 17 and 12 mg.)

Test Material Precision and Accuracy Results

Precision experiments were performed on 4 matrixes (Gb-SLV-1, Gb-SLV-2, Gb-SLV-3, and Gb-SLV-7), with triplicate sample preparations over a consecutive 3-day period. Within-day precision (estimated by S_w), between-day precision (estimated by S_b), and total precision (estimated by S_t) were determined in part through analysis of variance (ANOVA) and are presented in Table 5.

Table 5.

Repeatability of triplicate sample preparations for 4 matrixes over a consecutive 3-day period, with overall mean concentration (n = 9), RSD, % of label claim, and estimated standard deviations for within (S_w), between (S_b), and total (S_t) precision

Test sample	Mean concn, mg/Taba	RSD	Label claim, %	Swb	Qbc	Std
Gb-SLV-1	9.7	1.9	NAe	0.2	0.1	0.2
Gb-SLV-2	229.4	3.0	96	5.3	5.1	7.4
Gb-SLV-3	16.5	1.5	114	0.3	0.1	0.3
Gb-SLV-7	6.6	2.6	103	0.2	0.1	0.2

^aMean concentration determined from 9 determinations/matrix (3 replicates × 3 days); Tab = tablet.

^bS_w = Estimate of standard deviation within day; square root of mean square (MS) from ANOVA table.

^cS_b = Estimate of standard deviation between day; square root of (MS between groups − MS within groups)/replicates. For Gb-SLV-3 and Gb-SLV-7, subtracted MS totals were multiplied by −1 prior to square root.

^dS_t = Estimate of total standard deviation; square root of S_w² + S_b².

^eNA = Specific information regarding flavonol glycoside concentration was not present in the label claim.

For the 4 matrixes (m) tested in triplicate (r) over a 3-day period (d), the overall repeatability (RSD) was 2.3% (Table 5). ANOVA testing at α = 0.05 showed no significant differences among the within- or between-group sources of variation, although Gb-SLV-2 was borderline with a calculated p value of 0.09. Comparisons of S_w, S_b, and S_t showed that a majority of the standard deviation came from within-day determinations for all matrixes.

Negative control recovery was determined from spiking all 3 reference compounds in duplicate into 2 excipient matrixes at 2 concentrations (“low” at approximately 30% and “high” at approximately 90% of the determined concentrations in dry powder extract, Gb-SLV-2) over a single day. These spiking concentrations were also equivalent to approximately 25 to over 400% of the total aglycones determined in the other matrixes.

Negative control 1 consisted of a matrix of 95% methylcellulose and 5% starch. Total aglycone recovery from this matrix at the low spike concentration was 94% with an RSD of 1.8%, and recovery from the high spike concentration was 95% with an RSD of 1.0%. Negative control 2 consisted of a matrix of soybean oil, a common excipient found in gelatin capsules (consistent with Ginkgo product matrix Gb-SLV-7). Total aglycone recovery from this matrix at the low spike concentration was 95% with an RSD of 3.1%, and recovery from the high spike concentration was 96% with an RSD of 0.9 (Table 6).

Table 6.

Accuracy results from negative control recovery experiments consisting of duplicate sample preparations for 2 spike levels (“high” and “low”) in 2 matrixes

Negative control matrixa	Quercetin, μg/mL	Kaempferol, μg/mL	Isorhamnetin, μg/mL	Total aglycones, μg/mL	HorRatb
Low spike matrix 1a	17.2	17.7	3.6	38.5
Low spike matrix 1b	16.9	17.1	3.6	37.5
Mean	17.0	17.4	3.6	38.0
SD	0.2	0.4	0.0	0.7
%RSD	1.2	2.3	0.0	1.8	0.8
STD spikedc	18.2	18.4	3.8	40.5
% Recovery	93.4	94.6	94.7	93.8
High spike matrix 1a	51.3	52.7	10.9	114.8
High spike matrix 1b	52.3	52.9	11.1	116.3
Mean	51.8	52.8	11.0	115.6
SD	0.7	0.2	0.2	1.1
%RSD	1.4	0.4	1.8	1.0	0.5
STD spiked	54.7	55.2	11.5	121.4
% Recovery	94.7	95.7	95.7	95.2
Low spike matrix 2a	16.3	17.9	3.5	37.7
Low spike matrix 2b	17.6	18.0	3.7	39.4
Mean	16.9	18.0	3.6	38.5
SD	0.9	0.1	0.2	1.2
%RSD	5.3	0.6	5.6	3.1	1.3
STD spiked	18.2	18.4	3.8	40.5
% Recovery	92.9	97.8	94.7	95.1
High spike matrix 2a	51.6	53.3	11.0	115.8
High spike matrix 2b	52.6	53.5	11.2	117.3
Mean	52.1	53.4	11.1	116.6
SD	0.8	0.2	0.1	1.1
%RSD	1.5	0.4	0.9	0.9	0.4
STD spiked	54.7	55.2	11.5	121.4
% Recovery	95.2	96.7	96.5	96.0

^aMatrix 1 = 95% methyl cellulose + 5% starch; matrix 2 = soybean oil.

^bHorRat = %RSD/%PRSD, where %PRSD = percentage predicted repeatability standard deviation; = 2C^−0.1505; C = estimated mean concentration expressed as a decimal fraction, low spike samples C = 0.30, and high spike samples C = 0.90.

^cSTD = Standard.

The overall average recovery from the high spike levels was 96% with an RSD of 1.0%, which compares favorably to the AOAC recommended guidelines of repeatability (1%) and recovery (98–101%; ⁵). The overall average recovery from the low spike levels was 94% with an RSD of 2.5%, which again compares favorably, although not ideally, to the recommended repeatability of 1.5% and recovery of 95–102% (5). All calculated HorRat scores were within the limits for performance acceptability (0.5 to 2.0; ⁵), ranging from 0.4 to 1.3 (Table 6) and indicating satisfactory precision.

Conclusions

The SLV study presented was completed for a method to determine total flavonol glycosides in 9 selected Ginkgo biloba products, including crude leaf material, standardized dry powder extract, single and multiple entity tablets and capsules, and ethanol and glycerol tinctures. The method performed well in terms of ruggedness, hydrolysis testing results, precision, and accuracy. Based on the performance results presented, it is recommended that this method progress to a collaborative laboratory trial.