Development and Validation of a Stability-Indicating HPLC Method for the Quantification of α-Mangostin in Dietary Supplements

Abstract

Xanthones are the main active constituents isolated from Garcinia mangostana L. (Hypericaceae). Among these, α-mangostin (1) is reported as the principal component of the species, with remarkable pharmacological activities. In this work, a reverse-phase high-performance liquid chromatography (HPLC) method was developed as a suitable stability-indicating procedure for the quantification of 1 in the presence of degradation products. Chromatographic analysis was performed on a Purospher STAR C₁₈ column with a mobile phase containing 0.1% trifluoroacetic acid in water and ACN (20:80) at a flow rate of 1.0 mL/min. The method was linear over the concentration range of 10 to 100 μg/mL. The limits of detection and quantification were 0.13 and 0.4 μg/mL, respectively. Stability studies were carried out at temperature/relative humidity of 30°C/60% and 40°C/75% during 6 months; and no significant changes were observed. According to the global results, a simple HPLC method was developed and fully validated for quantifying 1 in mangosteen dietary supplements.

Garcinia mangostana L. (Hypericaceae), commonly known as mangosteen, is considered as the “queen of tropical fruits.” The species is native to Southeast Asia and highly regarded among connoisseurs as one of the best-tasting fruits. For centuries, the pericarp of this fruit has been marketed for its health benefits as a functional food and in dietary supplements. ^1,2 Prenylated and oxygenated xanthone derivatives are the main secondary metabolites identified and isolated in the species. ^{1 -3} The most abundant are α-mangostin (1) and γ-mangostin (2). ³ α-Mangostin (1) exhibits immunomodulatory, antifibrotic, antioxidant, antitumor, anti-inflammatory, antiallergic, antibacterial, antifungal, antiviral, and antimalarial activities. ^{4 -11}

Unlike conventional drugs, there is no standardization of active ingredients in herbal supplements, resulting in considerable differences between the safety and efficacy of the preparations. ¹² Therefore, investigations have shown that some products do not contain any measurable amount of the substances identified on the label, while others may contain up to 150% of the specified dose. The adulteration, either intentionally or otherwise, of botanical supplements with undeclared conventional pharmaceuticals or heavy metals is even more disturbing. ¹³

Even though G. mangostana is widely commercialized around the world as a crude drug for several herbal preparations, the back-of-pack label for mangosteen supplements often shows the total extract content, ie, the amount of individual xanthones is not specified; and a routinely analytical procedure to determine the content of α-mangostin (1) in dietary supplements has not been established. In this work, a stability-indicating high-performance liquid chromatography (HPLC) method with diode array (PDA) and mass spectra (MS) detection was developed to quantify and identify 1 in dietary supplements available in Mexican markets.

The most adequate separation, with the highest resolution between the major components γ- (2) and α-mangostin (1), was achieved using a Purospher STAR C₁₈ column under isocratic elution with 0.1% trifluoroacetic acid (TFA) in water and ACN (20:80). A representative chromatogram of the methanol extract of mangosteen is shown in Figure 1. The theoretical plates (N = 8815), tailing factor (T = 1.1), capacity factor (k′ = 2.8), and resolution (R = 10.9) meet the requirements stated in the United States Pharmacopeia.

Figure 1

Typical HPLC chromatogram of the methanol extract of mangosteen.

Next, the method was fully validated according to the International Conference on Harmonization (ICH) guidelines and was evaluated for its selectivity, linearity, precision, accuracy, and robustness. ¹⁴ The overall results are summarized in Table 1.

Table 1

Method Validation Parameters.

Parameter	1
Linear range (μg/mL)	10 – 100
Regression equation	y = 45 414.8x – 103 819.8
Determination coefficient (R 2)	0.9999
LOD (μg/mL) a	0.13
LOQ (μg/mL) b	0.40
Recovery (%)	99.1-101.7
Accuracy (regression equation)	y = 1.018x + 0.050
Determination coefficient (R 2)	0.9999
Precision (%RSD)	≤2.0%

RSD, relative standard deviation.

^aLimit of detection (S/N = 3).

^bLimit of quantification (S/N = 10).

The linearity of the system was determined using calibration curves. All the curves were plotted based on a least squares regression analysis of peak area (y) vs concentration (x; μg/mL) of the standard solution at 5 different concentrations in the range of 10 to 100 µg/mL with correlation coefficient >0.9999.

The accuracy of the method was evaluated by recovery experiments spiking 5 concentrations of 1 to the MeOH extract (10, 25, 50, 75, and 100 µg/mL). All samples were analyzed in triplicate (Table 1). The average recoveries ranged between 99.1% and 101.7% (relative standard deviation [RSD] ≤2.0%) indicating that the method is accurate (98.0%-102.0%).

The intra- and interday precision RSDs were under 2.0%. The repeatability variation was also under 2.0%. Altogether, the results indicate that the method was precise, accurate, and linear for the quantification of 1 in the MeOH extract. Under the experimental conditions, the limits of detection (LOD) and quantification (LOQ) were 0.13 and 0.4 µg/mL, respectively.

Until now, no stability-indicating method has been reported for analyzing 1 in dosage forms. ¹⁵ In this work, forced degradation testing processes were carried out under acid, alkaline, oxidative, and UV stress conditions (Table 2). None of the degradation products interfered with the quantification of 1, evidencing the ability of the method to unequivocally assess 1 in the presence of potential interferences. During acid hydrolysis, α-mangostin (1) was almost destroyed and 9 unknown degradation products appeared. During photolytic UV light degradation, oxidation, and alkaline hydrolysis, only minor degradation was observed. Furthermore, the high peak purity (>0.999) index confirmed that 1 was pure and homogeneous in all the analyses.

Table 2

Stress Degradation Studies Results.

Degradation condition	% Degradation
Acid hydrolysis	89.7
Base hydrolysis	14.5
Oxidation	9.6
Photolytic degradation	4.4

Thus, the proposed method was successfully applied for long-term and accelerated stability studies of 1 in some formulations. ¹⁵

A robustness test was performed to evaluate the effect of small, but deliberate variation in a method parameter test. For robustness assessment, a 2³ factorial design was used as statistical tool, and the adequacy of the model was evaluated showing that when acetonitrile and flow rate decrease, an improvement in resolution occurs (P value <0.05). In addition, the fit of the model to the data was evaluated by analysis of variance. The high value of the coefficient of determination for the response evaluated (R ² = 0.988) indicated the suitability of the model. Lastly, the mathematical model afforded to define the best chromatographic system for quantification as 80% of acetonitrile content in the mobile phase, flow rate of 1.0 mL/min, and 10 µL for injection size.

Finally, the developed and validated method was used for the quantification of 1 in 5 dietary supplements (F01-F05; Table 3). The amounts detected ranged from 2.5 to 89.6 µg per capsule, which were inconsistent with the content indicated in the label of each product. F01 and F03 claimed 100 mg of mangosteen extract, but the content of the extract was, respectively, 413.8 and 509.1 mg, demonstrating the widespread inconsistency among G. mangostana supplement manufacturers.

Table 3

Content of α-Mangostin (1) in Mangosteen Dietary Supplements.

Sample	Origin	α-Mangostin (μg) per capsule ± SD	Average content (mg)
F01	Mexico City	4.40 ± 0.149	413.8
F02	Mexico City	5.53 ± 0.163	424.7
F03	Mexico City	89.55 ± 1.493	509.1
F04	Mexico City	2.45 ± 0.025	834.8
F05	Mexico City	ND	454.6

UPLC-/MS was employed to confirm the identity of α-mangostin (1). The results agree with previous studies describing the fragmentation pathway of prenylated xanthones. ¹⁶ The MS spectra obtained for the acid hydrolysis sample is characterized by a 68 uma loss (Figure 2); 1 showed a peak at m/z 411 and fragments at m/z 355 [C₄H₈]⁺ and 299 [2C₄H₈]⁺ (Figure 2).

Figure 2

Typical ion current chromatogram in positive and negative modes (ESI) of α-mangostin (1) for the stability studies.

To investigate the possible stability issues of the extract, an accelerated stability study was carried out according to the ICH regulation at 40°C/75 relative humidity (RH) during 1, 2, 3, and 6 months. Evaluation of the extracts showed no significant changes in the content of α-mangostin (1) after 3-month storage under controlled room temperature and humidity (Table 4). The stability of α-mangostin (1) was also tested in intermediate stability studies by keeping the samples at 30°C/60 RH. The results indicate that the extract is stable at long-term or accelerated study conditions.

Table 4

Stability Testing of α-Mangostin (1) in Garcinia mangostana Extract Under Controlled Room Temperature and Humidity.

Storage conditions	Amount found (μg/mL)	Recovery (%)
Initial concentration	52.5 ± 0.22
30°C/60% RH
1 month	53.6 ± 0.07	102.1
2 months	52.5 ± 0.09	100.0
3 months	52.9 ± 0.13	100.8
6 months	54.1 ± 0.16	103.0
40°C/75% RH
1 month	52.7 ± 0.12	100.3
2 months	52.0 ± 0.22	99.1
3 months	51.4 ± 0.31	97.9
6 months	51.7 ± 0.29	98.4

RH, relative humidity.

According to the global results, a simple and reliable HPLC analytical method was developed and fully validated for quantifying α-mangostin (1). Forced degradation studies provided knowledge about potential degradation products that may form under storage conditions. These studies are essential in the development of the stability-indicating method as part of a validation protocol. It is recommended that this kind of analysis be carried out as early as possible during the manufacturing process in order to obtain valuable information that can be used to assess the inherent stability of a drug and improve formulations and processing. Stability tests of the extract were performed as per ICH guidelines Q1A (R2) by analyzing the initial α-mangostin (1) content and then at 1, 2, 3 and 6 months at 30°C ± 2°C/60 ± 5 RH and accelerated conditions 40°C ± 2°C/75 ± 5 RH. The degradation of 1 was found to be lowest in the extract over a study period; it can therefore be concluded that 1 is stable in the extract.

Furthermore, the method was highly specific for the quantification of 1 in dietary supplements and can be useful for quality control procedures.

Experimental

Plant Material and Reagents

All solvents used were of analytical and HPLC-grade from Honeywell Burdick and Jackson (Morristown, NJ, USA). TFA, formic acid, and α-mangostin (1) standard were acquired from Sigma-Aldrich-Fluka (St Louis, MO, USA). Ultrapure water for HPLC was deionized and filtered using a Maxima ultrapure water purifier system (Elgastat, Bucks, UK). Dry powder of mangosteen pericarp was obtained from DPN International Inc. Co. (Whittier, CA, USA). Dietary supplements (F01-F05) were purchased at different botanical stores in March 2016 in Mexico City, Mexico.

Dry powdered mangosteen pericarp (100 g) was exhaustively extracted by maceration with methanol at room temperature. The extract was filtered and the solvent was evaporated to dryness under reduced pressure at 70°C using a rotary evaporator yielding the crude extract.

Instrumentation and Chromatographic Conditions

The HPLC system consisted of a Shimadzu instrument equipped with a degasser (DGU14A), a quaternary pump (LC-10ATVP), an autosampler (SIL-10ADVP), a column oven (CTO-10A), a diode array detector (SPD-10A), and a computer with the LC Solution software for data analysis. All analyses were performed using a C₁₈ stationary phase in a Purospher STAR column (150 mm × 4.6 mm, 5 µm particle size). The mobile phase was an isocratic binary phase consisting of 0.1% TFA in water (A) and acetonitrile (B) (20:80) at a flow rate of 1.0 mL/min with an injection volume of 10 µL. The detection was monitored at 340 nm. The retention time (R _T) of 1 was 7.5 minutes under the conditions described. The xanthone was identified based on the retention time, ultraviolet spectrum, and MS spectrum and by comparing with an authentic sample. UPLC-ESI/MS analyses were performed on a Waters Acquity UPLC-H class system with a quaternary pump, column oven, sample manager, and photodiode array detector interfaced with an SQD2 single-quadrupole mass spectrometer detector with an electrospray ion source. LC separation was accomplished on a 100 mm × 2.1 mm, 1.7 µm, Acquity UPLC BEH C₁₈ column (Waters) with the following elution conditions: flow rate 0.3 mL/min, injection volume of 3 µL, and column temperature maintained at 40°C using a column oven. The mobile phase consisted of 0.1% formic acid in water (solvent A) and acetonitrile (solvent B) in the following gradient elution: 10% to 90% B in 11 minutes. The MS parameters were the following: cone and capillary voltages were set at 35.0 and 4.0 kV, respectively; source temperature and desolvation flow at 400°C and 650 L/h, respectively. MS spectra were obtained using nitrogen as the collision gas within a mass range of m/z 150 to 750 (scan duration of 0.5 s). Data acquisition and processing was done with the Masslynx 4.1 software; each run was followed by an equilibration time of 5 minutes.

Accelerated and intermediate studies were performed according to ICH guidelines. ¹⁵ BINDER KBF stability chambers were used for all the stability studies. The storage conditions were 30°C and 65% RH for intermediate studies and 40°C and 75% RH for accelerated studies. The extract was placed in stability chambers that were taken out periodically in different months.

Preparation of Standard Solutions

The standard stock solutions were prepared by stepwise dilution of a solution with 10 mg of α-mangostin (1) in 10 mL of methanol, to reach the concentration range of 10 to 100 µg/mL. All solutions were filtered through 0.45 µm PVDF membrane filters prior to injection into the HPLC system.

Sample Solution (Soft-Gelatin Capsules)

Test solutions were prepared by emptying the content of 20 individual dosage forms of each commercial product (soft-gelatin capsules) to determine their average weight. Then, an amount equivalent to 100 mg of extract was placed in a volumetric flask and 100 mL of methanol was added. This solution was filtered through a 0.45-µm PVDF membrane syringe filter for HPLC analysis.

Method Validation

The method was validated according to ICH guidelines. ¹⁴ The following characteristics were evaluated: selectivity, accuracy, precision, linearity, and robustness. The selectivity was assessed based on the peak purity function of the photolytic, oxidative, and hydrolytic degradation of 1. For this, 1 mL of hydrochloric acid (1 M), sodium hydroxide (1 M), and hydrogen peroxide (0.3%; v/v) were added separately to 1 mL of 1, at a concentration of 1.0 mg/mL, and heated to reflux for 60 minutes.

For accuracy, 3 replicates of the extract sample were spiked with standard compounds. The calibration curves were prepared in the concentration range expected for 1 (10-100 µg/mL). The mean recoveries and their standard deviations were calculated.

Linearity of the calibration plot was evaluated by a least squares regression according to the correlation coefficients using Statgraphics XVI versión 16.1.03 software (Statpoints Technologies, Warrenton, VA, USA). Thus, the linearity of 1 was evaluated by analyzing 5 different concentrations in the range of 10 to 100 µg/mL in triplicate.

Precision was determined based on the repeatability (intraday precision) and intermediate precision (interday precision) using the standard adding method. For system repeatability, 6 replicate analyses of the same solution containing 50 µg/mL of 1 were evaluated on the same day. The intermediate precision was estimated from a set of 6-fold analyses of the same solution of 1 on different days and by 2 analysts. The RSD was calculated; in the precision study for 1 it was less than 2.0%, which confirmed that the method was precise.

The LOD was calculated as 3.3 times the SD at signal-to-noise (S/N) ratio and the LOQ was calculated as 10 times the SD.

Robustness of the method was assessed using a 2³ factorial design. The factors evaluated were the following: acetonitrile content in the mobile phase; flow rate (mL/min); injection size (μL). The response evaluated was the resolution of peaks of 3-isomangostin and α-mangostin. The experiments were carried out in randomized order and in replicate (2 blocks). All analyses were performed using Statgraphics XVI versión 16.1.03 software. (StatPoint Technologies Inc, 2009, Warrenton, VA, USA).