Sustainable UAE-HPLC Approach for the Multi-Analyte Profiling of Seven Bioactive Xanthones in Garcinia mangostana L.

Abstract

Standardization of Garcinia mangostana L. products is often hindered by the use of toxic solvents and prolonged chromatographic run times. This study presents a sustainable, Green Analytical Chemistry (GAC)-compliant approach for the simultaneous multi-analyte profiling of seven bioactive xanthones (garcinone D, 8-deoxygartanin, γ-mangostin, gartanin, α-mangostin, 9-hydroxycalabaxanthone, and β-mangostin). Optimization of ultrasound-assisted extraction (UAE) identified 100% ethanol as the superior green solvent, achieving exhaustive extraction in only 5 min with a 0.1 g:5 mL mass-to-volume ratio. Separation was optimized by transitioning from a C18 to a C8 stationary phase to reduce hydrophobic retention. Using an isocratic mobile phase of ethanol and 5% acetic acid (62:38, v/v), baseline resolution of all analytes was achieved within 15 min at 254 nm. The method demonstrated high sensitivity with limits of detection (LODs) ranging from 0.5 to 170 ppb and excellent linearity (R ² > 0.999). Precision was confirmed by intra- and inter-day %RSDs < 5.0%. Accuracy was validated via recovery studies (80–120%) in complex matrices, including commercial mangosteen juices and raw pericarps from diverse Thai regions. The environmental impact was quantitatively validated, achieving an AGREE score of 0.74 and a ComplexMoGAPI score of 76. By replacing acetonitrile and methanol with ethanol, this UAE-HPLC method offers a robust, eco-friendly platform for high-throughput routine quality control and phytochemical standardization in the natural product industry.

1. Introduction

Mangosteen (Garcinia mangostana L.) is a tropical fruit of significant economic importance in Southeast Asia, widely recognized for its diverse profile of secondary metabolites. Among these, xanthonespolyphenolic compounds concentrated primarily in the fruit pericarpserve as the principal bioactive constituents. , To date, over 80 xanthones have been identified, with seven specific compounds (α-mangostin, β-mangostin, γ-mangostin, 8-deoxygartanin, 9-hydroxycalabaxanthone, gartanin, and garcinone D) demonstrating exceptional biological properties, including antioxidant, antitumoral, anti-inflammatory, and antimicrobial activities. These attributes make xanthones highly desirable for pharmaceutical, nutraceutical, and cosmetic applications. However, the commercial mangosteen market, particularly for dietary supplements, is currently hindered by a lack of active ingredient standardization. Existing products often exhibit significant variability in bioactive concentrations, creating a critical need for a robust and accurate quality control method for quantitative determination in both raw pericarp extracts and derived processed products.

The extraction of xanthone compounds from mangosteen pericarps represents a critical bottleneck in the isolation and subsequent quantification of these bioactive molecules. While traditional techniques such as maceration and Soxhlet extraction remain in use, , they are increasingly viewed as incompatible with modern laboratory standards due to excessive extraction times, high solvent volumes, and poor recovery yields. In response to these limitations, ultrasound-assisted extraction (UAE) has emerged as a high-performance alternative, leveraging acoustic cavitation to enhance mass transfer and yield while significantly reducing energy and solvent consumption. , However, despite the application of the UAE to various mangosteen-derived matrices, a significant research gap persists. , Specifically, no existing literature describes a unified UAE protocol for the simultaneous recovery of all seven target xanthones.

Furthermore, this study addresses a critical deficiency in the field: the lack of extraction conditions optimized strictly under Green Analytical Chemistry (GAC) guidelines. While methanol is frequently the default organic solvent, this study utilizes ethanol as a more suitable alternative. Beyond its low toxicity, ethanol provides the optimal polarity range to effectively disrupt the complex plant cell structures of the mangosteen pericarp, enhancing the solubility of moderately lipophilic xanthones more effectively than traditional hazardous solvents. This dual functionality, as both an efficient extractant and a sustainable HPLC mobile phase, positions this method as a robust advancement in green separation science.

High-performance liquid chromatography (HPLC) remains the gold standard for the separation and quantification of xanthones in complex matrices due to its superior resolution, accuracy, and precision. However, traditional HPLC methods for mangosteen analysis are often hindered by significant bottlenecks, including excessive analysis times, complex mobile phase gradients, and high consumption of hazardous solvents. Furthermore, existing literature predominantly focuses on the quantification of a single analyte, typically, α-mangostin, neglecting the broader bioactive profile of the fruit. − Despite the presence of >80 identified xanthones, the selection of these seven target compounds as a representative analytical subset was based on both biological and analytical considerations. Biologically, they represent the bioactive prenylated xanthones that are among the most frequently investigated compounds in pharmacological, phytochemical, and quality-control studies, making them relevant markers for the characterization and standardization of mangosteen-derived products. Analytically, these seven analytes encompass closely related structural analogues that traditionally require extended runtimes or complex gradient elution to resolve. Therefore, they provide a rigorous benchmark for evaluating whether the proposed rapid isocratic system can achieve simultaneous separation while maintaining satisfactory chromatographic performance, thereby demonstrating that a streamlined profile of key bioactive markers is sufficient for robust and routine product standardization without necessitating the highly resource-intensive profiling of all minor derivatives. While Chaivisuthangkura et al. initially reported the simultaneous analysis of seven bioactive xanthones; their protocol presents several analytical limitations: a run time exceeding 40 min, a restricted linear range, and a cumbersome gradient elution system.

This study addresses these challenges by replacing traditional acetonitrile and methanol with ethanol, a sustainable alternative that offers unique selectivity for xanthone separation while adhering to green chemistry principles. By transitioning to a C8 stationary phase, we have circumvented the long retention times associated with C18 columns, reducing the analysis time. The use of ethanol as a green organic modifier not only minimizes environmental toxicity but also facilitates a robust isocratic separation with high resolution for critical pairs, while providing a practical balance between chromatographic performance, analytical throughput, and environmental sustainability.

Thus, the objective of this research was to develop a high-performance, environmentally sustainable analytical platform integrating ultrasound-assisted extraction with HPLC-DAD for the simultaneous quantification of seven key bioactive xanthones. By prioritizing the principles of Green Analytical Chemistry, optimal UAE parameters were established using ethanol as a green solvent to minimize hazardous waste, reduce sample requirements, and lower energy consumptionall while ensuring the quantitative recovery of the target analytes. The resulting HPLC method was meticulously optimized to be both analytically robust and sustainable, effectively addressing the critical need for rapid and accurate standardization of mangosteen constituents. While previous studies have primarily focused on solvent replacement or runtime reduction, the novelty of this work lies in the integration of an ultra-rapid extraction workflow (5 min), the simultaneous multi-analyte quantification of seven bioactive xanthones, and a comprehensive sustainability evaluation using dual green analytical metrics (AGREE and ComplexMoGAPI) within a single analytical platform. To demonstrate the method’s practical utility, it was successfully applied to characterize regional variations in mangosteen pericarps from the southern and eastern regions of Thailand, as well as to quantify bioactive retention in commercial concentrated juices.

2. Results and Discussion

2.1. Optimization of Ultrasound-Assisted Extraction Conditions

The extraction protocol was optimized to maximize the recovery of seven target xanthones (α-mangostin, β-mangostin, γ-mangostin, 8-deoxygartanin, 9-hydroxycalabaxanthone, gartanin, and garcinone D). Due to the limited prior information on their simultaneous extraction behavior from the complex mangosteen pericarp matrix, a one-variable-at-a-time (OVAT) approach was employed as an initial screening step. Key UAE parameters, including solvent composition, mass-to-volume ratio (g/mL), extraction duration, and number of extraction cycles, were systematically evaluated to determine their independent effects, establish baseline trends, and define suitable working ranges. While the OVAT approach does not account for interaction effects between variables, it provides a practical and efficient means for preliminary optimization prior to the application of more comprehensive multivariate designs. To rigorously verify this univariate screening and ensure the statistical reliability of the selected parameters, the experimental optimization data were subjected to one-way ANOVA followed by Tukey’s post hoc test. The total chromatographic peak area of the seven analytes was used as the primary criterion for evaluating extraction efficiency.

The optimization results, summarized in Figure , demonstrate the impact of each variable:

• Solvent selection: Among the tested solvents, 100% (v/v) ethanol (EtOH) exhibited statistically superior extraction efficiency for six of the seven target xanthones, while remaining statistically equivalent to methanol for gartanin (Table S1). This high performance is attributed to ethanol’s optimal polarity, which effectively disrupts plant cell structures and enhances the solubility of moderately lipophilic xanthones.

• Green chemistry advantage: Ethanol was prioritized as a safer, bio-based alternative to toxic solvents like methanol, which pose greater risks to human health and the environment. Although water is the most eco-friendly solvent, its limited capacity to dissolve non-polar xanthone cores resulted in poor efficiency. Regarding ethanol concentration, while one-way ANOVA indicated that varying the concentration from 50% to 100% (v/v) did not significantly alter the extraction yields of six analytes (p > 0.05, Table S2), absolute 100% EtOH provided a statistically superior recovery for β-mangostin (p = 0.029). Consequently, 100% EtOH was selected to maximize multi-analyte recovery while eliminating routine aqueous dilution steps, thereby preventing cumulative volumetric errors.

• Mass-to-volume ratio: A ratio of 0.1 g of dried pericarp to 5 mL of solvent was selected as optimal. Although statistical analysis (Table S3) revealed no significant differences in extraction efficiency among the tested volumes (1–7 mL), lower volumes (such as 1 or 3 mL) were practically insufficient to properly suspend the 0.1 g of dried pericarp powder, which hindered efficient acoustic cavitation and subsequent filtration. Conversely, further increases in solvent volume (e.g., 7 mL) yielded no analytical advantages. Therefore, the 5 mL volume was selected as the optimal practical compromise between ensuring operational feasibility and minimizing solvent consumption.

• Extraction kinetics: Maximum yields were achieved within a short duration of 5 min. This rapid recovery highlights the role of acoustic cavitation in inducing mechanical stress and micro-fractures in the pericarp tissue, leading to immediate compound release. Durations exceeding 5 min (evaluated up to 30 min) did not yield statistically significant improvements (p > 0.05, Table S4), resulting in a stable horizontal plateau (Figure D). This kinetic profile suggests that extraction equilibrium was rapidly established and that the net rate of sonochemical degradation (k _deg) is practically negligible relative to compound release, which is further facilitated by the mild bulk thermal conditions (28 °C to 32 °C). Under these optimized parameters, no chromatographic baseline distortions or extraneous degradation peaks were observed. Although exhaustive forced-degradation stress profiling was not performed within the scope of this routine QC methodology, these collective empirical observations confirm that the streamlined 5 min UAE framework strictly preserves the structural integrity of the target xanthones.

• Process efficiency: A single extraction cycle was sufficient for quantitative recovery across the multi-analyte panel. Subsequent cycles did not contribute significantly to the yields of six out of the seven target xanthones (p > 0.05). Although a statistically significant increment was observed for one minor constituent (9-hydroxycalabaxanthone) after three cycles (Table S5), the single-cycle protocol was strategically retained as an optimal compromise. This confirms the practical efficiency of the initial 5 min sonication and strictly upholds GAC principles by preventing an unfavorable threefold escalation in solvent waste, energy expenditure, and total procedural turnaround time.

1.

Optimization of ultrasound-assisted extraction parameters: solvent types (A), ethanol concentration (B), mass-to-volume ratio (C), extraction time (D), and extraction cycles (E). Compounds: 1 = garcinone D; 2 = 8-deoxygartanin; 3 = γ-mangostin; 4 = gartanin; 5 = α-mangostin; 6 = 9-hydroxycalabaxanthone; and 7 = β-mangostin. Data are presented as the mean ± SD (n = 3). Statistical significance among different operational conditions for each individual analyte was determined using one-way ANOVA followed by Tukey’s post hoc test (p < 0.05), and the detailed results are summarized in Tables S1–S5.

Consequently, the finalized UAE protocol, utilizing a 0.1 g sample in 5 mL of 100% ethanol for 5 minachieved high extraction efficiency with minimal solvent consumption and energy usage, fully adhering to the principles of Green Analytical Chemistry.

2.2. Optimization of HPLC Conditions

The HPLC method was designed to ensure accurate, precise, and high-throughput quantification of the seven target xanthones. Building upon the foundational work of Chaivisuthangkura et al., which utilized a complex gradient system, this study sought to develop a simplified isocratic approach compliant with Green Analytical Chemistry standards. The optimization focused on a systematic evaluation of stationary phase chemistry, organic modifier selection, and mobile phase ratios, using separation efficiency, peak resolution (R _s), and cumulative solvent consumption as primary performance indicators. Quantification was performed at 254 nm, which corresponds directly to the principal UV absorption bands of five out of the seven target xanthones, effectively capturing the intense π → π* electronic transitions inherent to the core xanthone chromophore system. This wavelength was selected to maximize analytical sensitivity while minimizing background noise from the hydro-ethanolic mobile phase and avoiding far-UV interferences from the plant matrix. Although 254 nm does not correspond to the absolute apex of all investigated xanthones, it provided adequate sensitivity for all analytes (empirically demonstrated by the robust LOD/LOQ validation metrics detailed in Section ) while allowing simultaneous multi-analyte quantification under a single detection wavelength. This approach ensures that the method remains simple, robust, and readily transferable to standard HPLC systems equipped with basic UV detectors for routine quality control.

2.2.1. Selection of the Stationary Phase

Initial investigations were performed using a C18 column (Synergi 4 μm Hydro-RP, 150 mm × 4.6 mm, 4 μm) with an isocratic mobile phase of acetonitrile and 5% acetic acid (75:25, v/v). While this mobile phase provided adequate resolution for early-eluting peaks, the highly hydrophobic nature of β-mangostin resulted in an excessive retention time of approximately 25 min, making it inefficient for routine high-throughput analysis. Furthermore, attempts to reduce run time by increasing the organic modifier concentration compromised the resolution of critical xanthone pairs.

To address these limitations, a standard C8 stationary phase (Avantor ACE, 150 mm × 4.6 mm, 5 μm) was deliberately chosen. Although specialized stationary phases (such as phenyl-hexyl, biphenyl, or core-shell columns) have been reported to provide enhanced selectivity for certain aromatic analytes and remain an interesting prospect for future ultra-fast high-throughput applications, they offer limited secondary selectivity advantages here since all seven target xanthones share an identical tricyclic aromatic core. Therefore, they were not explored in the present study because the primary objective was to establish a simple, cost-effective, transferable, and broadly applicable method suitable for routine quality-control laboratories. Due to the lower carbon loading and reduced hydrophobicity of the C8 ligand compared to C18, the retention of all seven xanthones was significantly reduced (Table ). For instance, the retention time of β-mangostin decreased from 24.5 min on the C18 column to 9.24 min on the C8 column under identical mobile phase conditions. This significantly shortened the overall analytical run time and minimized organic solvent consumption, strictly aligning with GAC principles without compromising practical utility. However, the use of acetonitrile on the C8 column led to the co-elution of 8-deoxygartanin and γ-mangostin (R _s < 1.5). This dictated the need for an alternative organic modifier to tune selectivity.

1. Retention Times of Each Xanthone Compound on the C18 Column and the C8 Column Using Acetonitrile and 5% Acetic Acid (75:25, v/v) as Mobile Phase.

	Retention time, t R (min)
Analyte	C18 column	C8 column
garcinone D	3.51	2.75
8-deoxygartanin	5.71	4.02
γ-mangostin	7.23	4.20
gartanin	8.10	4.83
α-mangostin	9.08	5.26
9-hydroxycalabaxanthone	19.40	7.32
β-mangostin	24.50	9.24

2.2.2. Effect of Organic Modifier: Selectivity and Sustainability

The selection of the organic modifier is a critical factor in modulating the selectivity and resolution of xanthone compounds. In this study, acetonitrile and ethanol were comparatively evaluated as organic modifiers in an aqueous mobile phase containing 5% acetic acid. Modifier concentrations between 60% and 75% (v/v) were systematically screened to achieve baseline resolution for the most challenging critical pair: 8-deoxygartanin and γ-mangostin.

The experimental results revealed that acetonitrile failed to provide adequate separation for 8-deoxygartanin and γ-mangostin across all tested concentrations, resulting in persistent co-elution. Furthermore, decreasing the acetonitrile content to improve resolution led to a significant increase in the retention time of the final analyte, β-mangostin, from 9.24 to 20.5 min, thereby reducing throughput.

Conversely, ethanol proved to be a superior organic modifier for this multi-analyte separation. At lower ethanol percentages, enhanced separation and improved peak selectivity were observed (Figure ), consistent with the ability of ethanol, a protic solvent, to engage in hydrogen-bonding and dipolar interactions that can effectively tune selectivity for moderately lipophilic xanthone structures. ,

2.

Retention time of a mixed standard xanthone solution using a C8 column with different ethanol and 5% acetic acid ratios as the mobile phase.

Specifically, the baseline resolution of the critical pair (8-deoxygartanin and γ-mangostin) exhibits a noteworthy reversed-phase retention anomaly. Despite γ-mangostin possessing more phenolic hydroxyl groups, its prolonged retention is driven by three-dimensional molecular geometry and localized steric shielding rather than simple polar partitioning. , While the asymmetric C-2/C-4 prenylation of 8-deoxygartanin creates steric congestion that restricts deep intercalation into the C8 alkyl chains, the symmetrical C-2/C-8 prenylation of γ-mangostin preserves core planarity for optimal shape-selective dispersion interactions. Additionally, the C-8 prenyl group in γ-mangostin sterically shields the adjacent C-9 carbonyl oxygen from mobile phase solvation, while its symmetrical hydroxyl layout facilitates strong intramolecular hydrogen bonding, which promotes partial vector cancellation of the net dipole moment, reducing its effective solution polarity compared to the highly asymmetric 8-deoxygartanin.

This geometric selectivity is actively modulated by the hydro-organic mobile phase. The highly acidic environment (comprising 5% acetic acid, which maintains an apparent mobile phase pH* well below the pK _a thresholds of both phenolic solutes and residual silanols) ensures all species remain protonated and unionized, eliminating electrostatic ion-exchange dynamics. Under these conditions, the exposed C-6/C-7 ortho-diphenolic motif of γ-mangostin engages in weak, rapidly reversible dipole-type interactions with accessible silanols. , Here, the high concentration of acetic acid serves as an effective dynamic silanol suppressor, preventing peak tailing by maintaining fast mass transfer kinetics, while ethanol acts as a protic modifier to provide a stabilizing solvation network that preserves these delicate spatial retention differences.

Conversely, the dipolar aprotic nature of acetonitrile limits hydrogen bonding and actively competes for π–π interactions directly with the aromatic core of the analytes in the mobile phase, , compressing the retention gap and causing co-elution (α = 1.04, R _s = 1.25). Beyond its superior selectivity, ethanol aligns with the principles of green analytical chemistry due to its lower toxicity and high biodegradability. , This combined framework of molecular geometry, steric shielding, and secondary silanol interactions offers a coherent, structurally grounded rationalization of the enhanced resolution, strictly consistent with established reversed-phase trends for prenylated xanthones. ,

An optimal isocratic composition of ethanol and 5% acetic acid (62:38, v/v) was selected, enabling the complete separation of all seven xanthones within 15 min. As shown in Figure , the optimized conditions produced a stable baseline and a well-defined elution order, from garcinone D (peak 1) to β-mangostin (peak 7). The chromatogram exhibits sharp and symmetrical peaks with no apparent co-eluting interferences. These observations are supported by the system suitability parameters summarized in Table , particularly the tailing factors (T), which range from 1.03 to 1.06. Furthermore, a statistical summary derived from replicate injections (n = 5)including the relative standard deviations (%RSD) of peak areas, retention times, and resolution reproducibilityhas been incorporated into Table to demonstrate the system’s reliability and stability. Baseline separation was achieved for all analytes, including the adjacent pair of 8-deoxygartanin (peak 2) and γ-mangostin (peak 3), with a resolution (R _s) of 2.74. Furthermore, despite the higher viscosity of the ethanol-based mobile phase, the system backpressure remained stable and well within standard operational limits, primarily facilitated by the use of a conventional 5 μm particle size stationary phase and a moderate flow rate of 1.0 mL min^–1. The implementation of ethanol not only enhances analytical resolution but also supports environmental sustainability and operational safety for routine laboratory analysis.

3.

HPLC chromatogram of a mixed standard solution (10 ppm) of seven xanthone compounds using a C8 column with ethanol and 5% acetic acid (62:38, v/v) as the mobile phase. Peaks: 1 = garcinone D; 2 = 8-deoxygartanin; 3 = γ-mangostin; 4 = gartanin; 5 = α-mangostin; 6 = 9-hydroxycalabaxanthone; and 7 = β-mangostin.

2. Statistical Summary of System Suitability Parameters of the Proposed UAE-HPLC Method for the Determination of Seven Xanthones (n = 5).

Analytes	Retention time (t R, min) (%RSD)	Peak Area (mAU·s) (%RSD)	Resolution (R s) (%RSD)	Tailing factor (T)	Theoretical plates (N)
garcinone D	3.47 (0.35%)	662.96 (2.91%)	–	1.06	3502
8-deoxygartanin	5.22 (0.16%)	364.58 (2.53%)	10.8 (1.89%)	1.04	4379
γ-mangostin	5.78 (0.29%)	680.82 (3.33%)	2.74 (2.44%)	1.03	3816
gartanin	7.22 (0.21%)	185.97 (1.51%)	5.96 (2.03%)	1.05	4244
α-mangostin	8.04 (0.15%)	778.35 (2.38%)	2.91 (2.96%)	1.04	4167
9-hydroxycalabaxanthone	11.1 (0.72%)	322.87 (1.54%)	8.67 (1.65%)	1.05	4225
β-mangostin	13.4 (0.77%)	771.58 (1.89%)	5.09 (2.55%)	1.04	4055

Moreover, the robustness of the developed method was evaluated by introducing deliberate, minor variations to the mobile phase composition (±2% ethanol), flow rate (±0.2 mL/min), and column temperature (22–25 °C). These specific variation ranges were selected to represent realistic day-to-day operational fluctuations within standard analytical laboratories. Among the parameters tested, variations in the ethanol concentration exerted the most noticeable influence on the absolute retention times of the analytes. This behavior is typical in reversed-phase chromatography. Nevertheless, these deliberate variations did not compromise the method’s analytical performance. Complete baseline separation (R _s > 1.5) was consistently maintained across all tested conditions without any peak co-elution. These findings indicate satisfactory separation reproducibility despite minor variations in the mobile phase composition during routine preparation. These results, detailed in Tables S6–S8, confirm that the UAE-HPLC method possesses sufficient robustness to withstand typical day-to-day operational fluctuations. It should be noted that the evaluated temperature range (22–25 °C) reflects standard air-conditioned laboratory environments. Testing under wider, non-controlled ambient extremes (e.g., >30 °C) was not pursued, as operating precision HPLC instrumentation under such conditions introduces systemic hardware vulnerabilitiessuch as detector baseline drift and mobile phase degassing issuesthat obscure the true chemical robustness of the separation method. Therefore, for routine application in laboratories experiencing significant ambient thermal fluctuations, the use of an active column thermostat is strongly recommended to ensure reproducible analyte retention.

2.3. Method Validation

The developed UAE-HPLC method was rigorously validated in accordance with International Conference on Harmonisation (ICH) Q2­(R1) guidelines to verify its suitability for the simultaneous quantification of seven bioactive xanthones in complex mangosteen matrices. The comprehensive validation results are summarized in Tables , , and .

3. Linear Ranges, Limits of Detection, and Limits of Quantification of the Investigated Xanthones.

Analyte	Linear Range (ppm)	R 2	LOD (ppb)	LOQ (ppb)
garcinone D	0.020–10	0.9990	10	20
8-deoxygartanin	0.025–10	0.9999	5.0	25
γ-mangostin	0.010–100	0.9997	5.0	10
gartanin	0.050–100	0.9999	10	50
α-mangostin	0.0010–1000	0.9994	0.5	1.0
9-hydroxycalabaxanthone	0.50–10	0.9999	170	500
β-mangostin	0.10–100	0.9998	20	100

4. Comparison of the Limits of Detection (LOD) of the Proposed Method with Previously Reported Methods.

Analytes	This work LOD (ppb)	Literature LOD (ppb)
α-mangostin −	0.50	14–133
8-deoxygartanin	5.0	221
gartanin	10	248
β-mangostin	10	104
9-hydroxycalabaxanthone	170	131
γ-mangostin	5.0	Not reported
garcinone D	10	Not reported

5. Results of Repeatability and Intermediate Precision for the Seven Xanthones (n = 5).

	Repeatability (%RSD)		Intermediate precision (%RSD)
Analyte	Retention time	Peak area	Retention time	Peak area
garcinone D	0.20–0.67	2.8–4.9	0.55–0.62	3.0–3.6
8-deoxygartanin	0.14–0.68	2.0–4.4	0.20–0.58	0.9–4.7
γ-mangostin	0.12–0.34	2.4–4.2	0.66–1.10	3.2–4.2
gartanin	0.19–0.23	1.5–2.5	0.28–0.65	1.3–4.3
α-mangostin	0.086–0.15	2.2–4.5	0.75–1.30	3.5–4.1
9-hydroxycalabaxanthone	0.63–1.30	1.5–4.8	1.10–1.40	2.5–4.7
β-mangostin	0.42–1.30	1.6–3.1	0.83–1.30	4.0–4.1

2.3.1. Linearity and Sensitivity

Linearity was established for each of the seven xanthones across their respective concentration ranges. The calibration curves yielded correlation coefficients (R ²) exceeding 0.999, confirming a robust linear relationship between peak area and analyte concentration. Notably, the method demonstrated exceptional sensitivity, with limits of detection and limits of quantification in the ranges of 0.5–170 ppb and 1.0–500 ppb, respectively (Table ). The extremely low LOD for the primary bioactive marker, α-mangostin (0.5 ppb), highlights the method’s capability to reliably analyze highly diluted commercial products.

To systematically evaluate the analytical performance of the proposed approach, the limits of detection (LODs) were directly compared with previously reported multi-analyte methods for the same target xanthones (Table ). A direct evaluation demonstrates that the proposed method offers significantly enhanced sensitivity for most of the target analytes. It should be noted that the relatively higher LOD observed for 9-hydroxycalabaxanthone (170 ppb) is specifically attributed to its lower intrinsic molar absorptivity at the selected universal consensus wavelength of 254 nm. Nevertheless, its corresponding LOQ of 500 ppb remains highly sufficient for routine quantification, as the natural phytochemical abundance of this compound in the investigated sample matrices far exceeds this analytical threshold. Moreover, for 9-hydroxycalabaxanthone, the LOD is comparable to the literature, which is an acceptable trade-off for the ability to simultaneously detect multiple analytes within a single run. Furthermore, by establishing detection limits for two previously unreported analytes, the proposed method provides a broader and more comprehensive analytical window for trace xanthone determination. However, it should be explicitly noted that for samples containing specific xanthones at concentrations substantially below their respective validated LOQs (e.g., ultra-trace levels of minor constituents), the current conventional HPLC configuration may yield limited quantitative reliability. In such specific ultra-trace scenarios, the application of sample preconcentration techniques or more sensitive hyphenated detectors (e.g., LC-MS/MS) would be recommended.

2.3.2. Precision and Stability

Method precision was evaluated through repeatability (intra-day) and intermediate precision (inter-day) studies conducted at three concentration levels (Table ). The percentage relative standard deviations for both retention times (t _R) and detector responses (peak areas) were calculated. In all instances, the %RSD values remained consistently below 5.0%, demonstrating the excellent precision and reliability of the optimized isocratic ethanol-based system for routine analysis.

To ensure method reliability, the short-term stability of the seven standard xanthones and the sample extracts (derived from both mangosteen pericarp and concentrated juice) prepared in the mobile phase was evaluated. As summarized in Table S9, the results demonstrated that both the standard solutions and the sample extracts remained quantitatively stable when stored at 25 °C for up to 48 h, exhibiting relative standard deviations (RSDs) of less than 4.0%. Because long-term stability evaluations beyond this 48 h window fell outside the established methodological scope of the present study, it is recommended that working standard solutions and mobile phases be freshly prepared for routine analytical sequences, or independently validated by the adopting laboratory for extended storage under their specific environmental conditions.

2.4. Determination of Xanthones in Real Samples and Accuracy Validation

The optimized UAE-HPLC method was applied to quantify the seven target xanthones in both natural sources and processed mangosteen products. This included raw mangosteen pericarps harvested from the southern and eastern regions of Thailand, as well as two commercial brands of concentrated juice. The pericarp samples provided a baseline for natural bioactive abundance, while the juice samples allowed for the evaluation of xanthone retention following industrial processing.

2.4.1. Quantification in Natural and Processed Matrices

The concentrations of the seven xanthones in dried pericarp samples are expressed in mg/g (Table ), where values are reported as mean ± SD, and the corresponding 95% confidence intervals (CIs) are provided in parentheses to reflect statistical variability. Consistent with existing literature, α-mangostin was identified as the predominant constituent across all samples. Samples from the eastern region of Thailand exhibited higher total xanthone levels than those from the southern region. However, the relatively wide 95% CIs observed for some analytes highlight the extent of biological variability among independent orchards (n = 3). Although this variability suggests potential regional differences, the overlap of confidence intervals indicates that these differences should be interpreted with caution, and a larger sample size would be required to establish statistically significant geographical distinctions.

6. Concentration of Xanthones in Mangosteen Pericarp from Different Regions of Thailand .

	Southern region samples		Eastern region samples
Analyte	Concentration (mg/g) Mean ± SD	Recovery (%) Mean ± SD	Concentration (mg/g) Mean ± SD	Recovery (%) Mean ± SD
garcinone D	1.21 ± 0.03 (1.14–1.28)	99.6 ± 7.8 (80.2–119.0)	1.68 ± 0.05 (1.55–1.81)	93.8 ± 2.2 (88.3–99.3)
8-deoxygartanin	1.57 ± 0.05 (1.46–1.68)	107.0 ± 1.2 (104.0–110.0)	1.80 ± 0.04 (1.70–1.90)	102.0 ± 7.2 (84.1–119.9)
γ-mangostin	5.34 ± 0.13 (5.02–5.66)	98.6 ± 3.3 (90.4–106.8)	7.56 ± 0.24 (6.96–8.16)	88.4 ± 3.4 (80.0–96.8)
gartanin	7.27 ± 0.09 (7.04–7.50)	101.0 ± 7.0 (83.6–118.4)	9.52 ± 0.42 (8.48–10.56)	98.1 ± 5.8 (83.7–112.5)
α-mangostin	37.7 ± 1.2 (34.7–40.7)	85.7 ± 2.1 (80.5–90.9)	51.2 ± 1.7 (47.0–55.4)	86.7 ± 1.9 (82.0–91.4)
9-hydroxycalabaxanthone	2.73 ± 0.26 (2.08–3.38)	109.0 ± 2.9 (101.8–116.2)	2.80 ± 0.08 (2.60–3.00)	103.0 ± 1.6 (99.0–107.0)
β-mangostin	1.72 ± 0.05 (1.60–1.84)	111.0 ± 2.6 (104.6–117.4)	2.03 ± 0.07 (1.86–2.20)	106.0 ± 2.0 (101.0–111.0)

^aData are based on biological replicates from independent orchards per region (n = 3). Values in parentheses indicate 95% confidence intervals calculated using the t-distribution.

In processed products, α-mangostin remained the major xanthone component in both commercial juice brands analyzed (Table ). Notably, the total xanthone content in 1 g of pericarp was comparable in magnitude to that found in 1 L of concentrated juice. This substantial difference in concentration density is biologically expected; xanthones are hydrophobic defense metabolites localized predominantly in the pericarp, whereas commercial juices are primarily aqueous formulations derived from the inner flesh (aril), which naturally contains negligible amounts of these compounds. The successful detection of all seven analytes in these complex matrices demonstrates the method’s precision, selectivity, and applicability to complex matrices.

7. Concentration of Xanthones in Concentrated Juice from Different Commercial Brands .

	Brand 1 samples		Brand 2 samples
Analyte	Concentration (mg/L) Mean ± SD	Recovery (%) Mean ± SD	Concentration (mg/L) Mean ± SD	Recovery (%) Mean ± SD
garcinone D	0.65 ± 0.01 (0.63–0.67)	106.0 ± 1.4 (102.5–109.5)	3.10 ± 0.01 (3.08–3.12)	88.0 ± 2.8 (81.0–95.0)
8-deoxygartanin	1.34 ± 0.10 (1.09–1.59)	103.0 ± 4.3 (92.3–113.7)	0.68 ± 0.08 (0.48–0.88)	104.0 ± 5.8 (89.6–118.4)
γ-mangostin	2.91 ± 0.02 (2.86–2.96)	85.8 ± 1.7 (81.6–90.0)	1.03 ± 0.03 (0.96–1.10)	93.4 ± 2.1 (88.2–98.6)
gartanin	11.9 ± 1.5 (8.2–15.6)	84.5 ± 1.8 (80.0–89.0)	4.36 ± 0.52 (3.07–5.65)	85.1 ± 2.0 (80.1–90.1)
α-mangostin	17.8 ± 0.4 (16.9–18.7)	87.2 ± 1.3 (84.0–90.4)	16.4 ± 0.2 (15.9–16.9)	89.4 ± 1.1 (86.7–92.1)
9-hydroxycalabaxanthone	9.03 ± 0.34 (8.19–9.87)	94.8 ± 5.2 (81.9–107.7)	3.41 ± 0.30 (2.66–4.16)	111.0 ± 3.9 (101.3–120.7)
β-mangostin	2.92 ± 0.01 (2.90–2.94)	108.5 ± 4.7 (96.8–120.2)	1.74 ± 0.03 (1.67–1.81)	120.0 ± 0.2 (119.5–120.5)

^aValues in parentheses indicate 95% confidence intervals calculated using the t-distribution.

However, the routine application of this UAE-HPLC protocol to finished commercial dietary supplementssuch as dry powders, capsules, or compressed tabletswas not directly evaluated in the current work. While the established extraction and chromatographic parameters provide a solid analytical baseline, commercial supplements frequently incorporate diverse formulation excipients (e.g., binders, lubricants, and anti-caking agents) or multi-botanical blends that introduce unpredictable matrix variables. Therefore, adapting this method for finished dietary supplements requires prior product-specific partial validation, with particular emphasis on confirming chromatographic resolution against potential excipient interferences and verifying analyte recovery to ensure reliable quantification.

2.4.2. Recovery and Assessment of Matrix Effects

To validate the accuracy of the method and assess potential matrix effects in real-world applications, recovery studies were performed by spiking known amounts of xanthone standards into both pericarp extracts and juice matrices. Sample preparation incorporated the established dilution protocol (Section ) that effectively minimized matrix-induced signal anomalies. As summarized in Tables and , the mean percentage recoveries consistently remained strictly within the acceptable 80% to 120% validation range. Although the upper bound of the 95% confidence interval exhibited a minor statistical tail reaching 120.5% in concentrated juice matrices, this upper statistical boundarywhile indicative of potential mild matrix-induced signal enhancementremains strictly within practically acceptable thresholds for routine botanical and food quality control. These findings indicate that matrix effects, such as signal suppression or enhancement from sugars and other polyphenols, did not significantly compromise chromatographic quantification, demonstrating that the method is fit-for-purpose using conventional external standard calibration.

Furthermore, the quantitative specificity of the method was confirmed by peak purity analysis (Agilent OpenLab CDS), where all target xanthones exhibited a purity factor greater than 990. In addition, analyte specificity in complex matrices was visually supported by representative DAD spectral overlays (Figure S1). The selected analytes included 8-deoxygartanin and γ-mangostin, which represent the critical chromatographic pair in the developed separation, alongside β-mangostin, a representative late-eluting and highly hydrophobic xanthone. The DAD spectral profiles of these selected peaks obtained from mangosteen pericarp extract and concentrated juice samples were highly comparable with those of the corresponding standards, providing robust qualitative evidence for peak assignment. Crucially, while routine industrial execution of this protocol relies solely on standard single-wavelength UV detection, this developmental DAD verification provides robust qualitative proof of peak homogeneity. Consequently, the developed UAE-HPLC approach is a robust and reliable tool for the accurate quality control of mangosteen-derived products in both research and industrial settings.

Overall, analytical accuracy was practically established via thorough spike-recovery experiments across highly complex botanical matrices. Although the empirical recovery distributions confirm that major matrix-induced quantitative bias was successfully avoided, we transparently acknowledge that formal matrix-matched calibration curves or standard addition protocols were not executed. These approaches represent a methodological boundary of the current routine QC framework and should be prioritized in future exhaustive validation investigations to provide a direct, absolute mathematical determination of signal suppression or enhancement factorsparticularly when deploying this chromatographic platform to novel, highly viscous mangosteen-derived nutraceutical formulations.

2.5. Greenness Assessment and Method Sustainability

The environmental impact of the developed UAE-HPLC method was evaluated against the principles of Green Analytical Chemistry. To ensure a comprehensive evaluation, the environmental sustainability of the proposed method was rigorously characterized using a dual-metric approach involving AGREE and ComplexMoGAPI, providing a more holistic perspective than single-metric approaches. This assessment revealed a high level of concordance between the two frameworks, with the method yielding an AGREE score of 0.74 and a ComplexMoGAPI score of 76 (Figure ). All underlying assumptions, specific input parameters, and detailed scoring breakdowns for these calculations are detailed in the Supporting Information (Tables S10 and S11). The calculated AGREE and ComplexMoGAPI scores are expected to be highly reproducible when this protocol is implemented in independent laboratories, as they are predominantly governed by fixed analytical parameters such as sample mass, bio-based solvent utilization, extraction volume, and total run time. However, it should be noted that minor score variations may occur during inter-laboratory method transfer due to instrument-specific hardware factors, such as differences in exact electrical energy consumption between varying HPLC models or variations in internal system dead volumes affecting routine equilibration waste. These favorable values confirm the robust green profile of the procedure and underscore its suitability as a sustainable alternative to established chromatographic methods. The method’s sustainability is underpinned by three key factors. While many previously reported methods often prioritize specific green aspects, the present approach simultaneously addresses solvent toxicity, energy consumption, analytical throughput, and waste generation.

• Solvent replacement and toxicity reduction: A primary achievement of this study is the complete replacement of hazardous organic solvents. While conventional methods for xanthone analysis rely on acetonitrile and methanol, the current protocol utilizes ethanol as the sole organic modifier for both ultrasound-assisted extraction and the HPLC mobile phase. Ethanol is recognized for its lower toxicity, higher biodegradability, and superior safety for laboratory personnel.

• Energy and time efficiency: The integration of UAE significantly optimized the extraction kinetics, achieving quantitative recovery within only 5 min. This represents a substantial reduction in energy consumption compared to traditional maceration or Soxhlet techniques. Furthermore, the transition to an isocratic C8-based separation reduced the total chromatographic analysis time to 15 min per sample, enhancing sample throughput while minimizing the carbon footprint of the instrumentation.

• Waste reduction through miniaturization: The method minimizes hazardous waste generation by employing an optimized mass-to-volume ratio. By using only 5 mL of ethanol per 0.1 g of sample, the protocol reduces solvent waste by several orders of magnitude compared to conventional large-scale extraction methods.

4.

AGREE (A) and ComplexMoGAPI (B) assessment scores for the developed method.

While ultra-high-performance liquid chromatography (UHPLC) systems inherently offer chromatographic advantages regarding rapid column elution and trace-level sensitivity, the proposed conventional HPLC method prioritizes practical accessibility, instrument transferability, and operational cost-effectiveness for routine quality-control laboratories. In terms of sensitivity, although UHPLC platforms achieve lower detection thresholds, the established limits of detection and quantification of the proposed HPLC approach are strictly fit-for-purpose for profiling abundant secondary metabolites in botanical pericarps and commercial juices. To further highlight the sustainability and unique practical advantages of the proposed method, a comprehensive comparison with previously reported multi-analyte xanthone methods, including recent green analytical approaches and a UPLC-based method, was conducted (Table ). As shown in Table , the proposed method effectively analyzes both mangosteen pericarps and concentrated juice while substantially reducing total solvent consumption and overall analysis time. In comparison, recently reported green extraction-based methods require total analysis times ranging from >49 min to over 5 h, whereas the proposed method completes the entire workflow, including extraction and quantification, within 25 min. Notably, as outlined in Table , although the comparative UPLC-UV DES-based approach operates on advanced instrumentation with a slightly lower solvent volume (17.5 mL), its scope is restricted to quantifying only 2 analytes and requires a total analytical turnaround time exceeding 3.50 h due to intensive sample extraction protocols. In contrast, the proposed streamlined HPLC framework enables the simultaneous multi-analyte quantification of 7 bioactive xanthones within a complete turnaround workflow of only 25 min and a highly comparable total solvent consumption (21 mL), successfully bridging the gap between green analytical principles and routine industrial utility. These results demonstrate that the proposed method offers a favorable balance between analytical performance, time efficiency, and solvent usage, supporting its suitability as a rapid and environmentally considerate approach for xanthone analysis. Although the environmental sustainability of the proposed method was comprehensively assessed using AGREE and ComplexMoGAPI, a formal life-cycle assessment (LCA) and the incorporation of additional green metrics, such as Analytical Eco-Scale and GAPI, represent valuable directions for future investigations to provide complementary sustainability evaluations.

8. Quantitative Comparison of the Proposed Method with Previously Reported Multi-Analyte Methods for the Determination of Xanthones.

Method	Analytes	Sample type	Extraction method	Total analysis time	Total solvent consumption (mL)
HPLC-DAD	6	Dietary supplements	UAE	1.78 h	43.5
HPLC-DAD	6	Mangosteen pericarps	Solvent extraction	1.08 h	85
HPLC-DAD	14	Mangosteen pericarps	Maceration	37 h	310
UPLC-UV	2	Mangosteen pericarps	DES-UAE	>3.50 h	17.5
UV–Vis	1	Mangosteen pericarps	NADES-MAE	>49.46 min	Not reported
HPLC	8	Mangosteen pericarps	SFE	5.13 h	>68
Proposed method	7	Mangosteen pericarps	UAE	25 min	21
		Concentrated juice	Filtration

^aDeep eutectic solvent-based ultrasound-assisted extraction.

^bNatural deep eutectic solvent-based microwave-assisted extraction.

^cSupercritical fluid extraction.

3. Conclusions

This study successfully established and validated a high-performance analytical platform integrating ultrasound-assisted extraction with HPLC-DAD for the simultaneous quantification of seven bioactive xanthones. The optimized method achieved baseline separation with high resolution for all analytes, including the challenging critical pair of 8-deoxygartanin and γ-mangostin, within a rapid 15 min isocratic run.

The analytical performance and practical utility of the UAE-HPLC approach were demonstrated through its application to complex matrices, including raw mangosteen pericarps from diverse regions of Thailand and commercial concentrated juices. The method proved to be robust, sensitive, and accurate, with recovery values consistently within the 80–120% range, indicating minimal matrix interference. Although a formal evaluation of measurement uncertainty was beyond the scope of this validation protocol, the high precision and accuracy demonstrated the method’s current reliability. The incorporation of a comprehensive measurement uncertainty budget following internationally recognized frameworks remains an important objective for future prospective studies. Although the one-variable-at-a-time approach employed in this study successfully established reliable baseline conditions, it does not account for potential interaction effects among the extraction parameters. Therefore, future studies should consider employing multivariate optimization strategies, such as Response Surface Methodology (RSM) or Design of Experiments (DoE), to comprehensively evaluate these interactions and further refine the extraction process. Although the developed method demonstrated satisfactory performance for Garcinia mangostana matrices, additional validation and potential re-optimization may be required before application to other Garcinia species possessing different phytochemical compositions and matrix characteristics. Nevertheless, this green platform serves as an excellent, eco-friendly blueprint for high-throughput natural product analysis. Furthermore, while satisfactory reproducibility was achieved using the specific ultrasonic bath setup in this study, minor re-optimization may be required when transferring the method to other ultrasonic systems with different operating frequencies, power distributions, and cavitation characteristics.

Crucially, this method advances the goals of Green Analytical Chemistry by replacing hazardous solvents such as acetonitrile and methanol with ethanol for both the extraction process and the chromatographic mobile phase. From an economic and operational perspective, the feasibility of implementing this method in routine industrial quality control is highly favorable. First, it utilizes conventional HPLC instrumentation and a standard C8 column, avoiding the need for specialized equipment or high capital investment. Second, the simple isocratic elution and short 15 min run time significantly improve analytical throughput while reducing overall energy and solvent consumption. Finally, while the procurement cost of HPLC-grade ethanol is comparable to standard organic modifiers, replacing toxic acetonitrile drastically reduces hazardous waste disposal costsa major financial and regulatory burden for industrial laboratories operating at high capacities. By significantly reducing analysis time, solvent toxicity, and energy consumption during sample preparation, this developed method offers a sustainable, cost-effective, and efficient tool for routine quality control and phytochemical standardization. While currently validated at the laboratory scale, the use of ethanol, short extraction times, and a simple isocratic system provides a strong foundation for future adaptation and scale-up into industrial applications.

4. Experimental Section

4.1. Chemicals and Materials

Methanol (HPLC grade), ethanol (HPLC grade), and acetonitrile (HPLC grade) were purchased from RCI Labscan (Bangkok, Thailand). Acetic acid glacial (AR grade) was purchased from Sigma-Aldrich (Germany). Deionized water was obtained from a water purification system (model Direct-Q 3 UV, MERCK MILLIPORE, USA). Standard seven xanthone compounds, including α-mangostin, β-mangostin, γ-mangostin, 8-deoxygartanin, 9-hydroxycalabaxanthone, gartanin, and garcinone D were previously isolated, purified, and structurally confirmed using NMR and ESIMS spectroscopic techniques, as reported by Suksamrarn et al. ,,

4.2. Sample Preparation

Ripe mangosteen (Garcinia mangostana L.) fruits were procured from three different local farms per region in the southern and eastern regions of Thailand. The pericarps were separated from the pulp and processed immediately upon collection. To remove surface impurities, the pericarps were thoroughly washed with deionized water and oven-dried at 50 °C until a constant weight was reached. The dried material was pulverized into a fine powder using a high-speed electric grinder (Multi-functional Grinder, West Kitchen) and stored in a desiccator at ambient temperature to prevent moisture absorption and ensure phytochemical stability prior to extraction.

For processed products, two commercial brands of concentrated mangosteen juice were purchased from a local supermarket in Bangkok, Thailand. To rigorously evaluate the analytical applicability of the developed method, commercial concentrated mangosteen juice was intentionally selected as a highly challenging complex matrix model due to its high viscosity, suspended solids, and sugar content. This selection aimed to challenge the chromatographic resolving power and matrix tolerance, rather than to assess the batch-to-batch consistency of the manufacturer. The juice samples were initially filtered through Whatman No. 1 filter paper to remove suspended particulate matter. Before HPLC analysis, the filtered juices were diluted 10-fold and passed through a 0.22 μm syringe filter to ensure sample homogeneity and protect the chromatographic system.

4.3. Ultrasound-Assisted Extraction Conditions

For the optimized extraction of the seven bioactive xanthones, 0.1 g of powdered mangosteen pericarp was accurately weighed and suspended in 5.0 mL of 100% (v/v) ethanol within a conical flask. The mixture was subjected to ultrasound-assisted extraction using an ultrasonic bath (Model D13, GT SONIC) operated at a fixed frequency of 40 kHz and a power output of 300 W. During ultrasonication, the temperature of the bath water increased from approximately 28 °C before sonication to 32 °C after 5 min of extraction, indicating only a modest temperature rise under the optimized UAE conditions.

Exhaustive extraction was achieved within a 5 min duration at ambient temperature. Following sonication, the suspension was filtered through Whatman No. 1 paper to remove insoluble plant residue. The resulting crude extract was diluted 10-fold with HPLC-grade ethanol and passed through a 0.22 μm membrane filter to ensure suitability for chromatographic analysis.

4.4. HPLC Instrumentation and Chromatographic Conditions

Chromatographic analysis was conducted using an Agilent 1260 Infinity Series HPLC system (Agilent Technologies, USA) equipped with a binary pump, a manual injector, and a diode array detector (DAD). The simultaneous separation of the seven target xanthones was achieved using a C8 stationary phase (Avantor ACE, 150 mm × 4.6 mm, 5 μm particle size).

The mobile phase consisted of an isocratic mixture of ethanol and 5% aqueous acetic acid (62:38, v/v). The choice of ethanol as the organic modifier was prioritized to align with Green Analytical Chemistry principles, replacing traditional hazardous solvents like acetonitrile. The flow rate was maintained at 1.00 mL/min, and the injection volume was 20 μL for all samples and standards. Detection was performed at a fixed wavelength of 254 nm. The column was operated at ambient temperature (22–25 °C) without thermostatic control.

4.5. Method Validation

The developed UAE-HPLC method was validated in accordance with the International Conference on Harmonisation guidelines. The evaluated parameters included linearity, sensitivity (LOD and LOQ), precision (repeatability and intermediate precision), and accuracy through recovery studies.

4.5.1. Linearity and Sensitivity

Linearity was assessed using mixed standard solutions of the seven xanthones at five distinct concentration levels. Calibration curves were constructed by plotting the chromatographic peak areas against the analyte concentrations, followed by least-squares linear regression analysis. Method sensitivity was evaluated by calculating the LOD and LOQ, which were based on signal-to-noise (S/N) ratios of approximately 3:1 and 10:1, respectively.

4.5.2. Precision and Accuracy

Method precision was evaluated at three concentration levels (low, medium, and high) covering the validated linear range for each analyte (Table ). Intra-day precision (repeatability) was determined using five replicate preparations within a single day, while inter-day precision (intermediate precision) was assessed over five non-consecutive days. Results were expressed as the percentage relative standard deviation (%RSD). Accuracy was determined via recovery studies using both concentrated mangosteen juice and dried pericarp matrices. Known concentrations of xanthone standards were spiked into each matrix: 5.0 ppm for garcinone D, 8-deoxygartanin, 9-hydroxycalabaxanthone, and β-mangostin; 10 ppm for γ-mangostin and gartanin; and 100 ppm for α-mangostin. Spiked samples were processed following the procedures outlined in the sample preparation section for the juice matrix and ultrasound-assisted extraction conditions section for the pericarp. The percentage recovery was calculated as the ratio of the measured concentration to the theoretical spiked amount.

9. Concentrations of Seven Xanthones for Precision Analysis.

	Concentration level (ppm)
Analyte	Low	Medium	High
garcinone D	0.5	2.0	10
8-deoxygartanin	0.025	5.0	10
γ-mangostin	1.0	20	100
gartanin	0.050	25	50
α-mangostin	5.0	100	500
9-hydroxycalabaxanthone	0.50	2.0	10
β-mangostin	1.0	20	100

4.5.3. Stability Studies

The stability of the standard solutions and sample extracts (mangosteen pericarp and concentrated mangosteen juice) prepared in the mobile phase was evaluated to ensure their integrity during analysis. The solutions were stored at 25 °C and manually injected into the HPLC system at 0, 12, 24, and 48 h. Stability was determined by comparing the peak responses of the stored samples with those of freshly prepared solutions.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.6c01350.

• Statistical summary of solvent type optimization (Table S1); statistical summary of ethanol concentration optimization (Table S2); statistical summary of mass-to-volume ratio optimization (Table S3); statistical summary of extraction time optimization (Table S4); statistical summary of extraction cycle optimization (Table S5); robustness evaluation: effect of minor variations in mobile phase composition (Table S6); robustness evaluation: effect of minor variations in flow rate (Table S7); robustness evaluation: effect of minor variations in column temperature (Table S8); stability of seven standard xanthones and sample extracts (Table S9); detailed input parameters for the AGREE assessment (Table S10); detailed input parameters for the ComplexMoGAPI assessment (Table S11); representative DAD spectral overlays obtained from standard solution, mangosteen pericarp extract, and concentrated juice sample (Figure S1) (PDF)

D.K.: Investigation, Methodology, Visualization, WritingOriginal Draft. S.R.: Formal analysis, Validation. J.B.: Formal analysis, Validation. W.S.: WritingReview & Editing. S.S.: Funding acquisition, Resources. P.J.: Conceptualization, Supervision, WritingReview & Editing.

The authors declare no competing financial interest.