Abstract
An ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) method was developed for the determination of nobiletin and tangeretin in rat plasma, and the plasma was processed by a simple liquid-liquid extraction method with ethyl acetate. The chromatographic column was UPLC HSS T3 (50 × 2.1 mm, 1.7 μm), the mobile phase was acetonitrile-water (containing 0.1% formic acid). Multiple reaction monitoring mode (MRM) was used for quantitative analysis, nobiletin m/z 403.29 → 373.14 (cone voltage 22v, collision voltage 28v), tangeretin m/z 373.28 → 343.17 (cone voltage 20v, collision voltage 28V), tangeretin m/z 373.28 → 343.17 (cone voltage 20V, collision voltage 28V) and internal standard vitexin m/z 433.14 → 313.03 (cone voltage 32v, collision voltage 26v). The pharmacokinetics of nobiletin and tangeretin were evaluated in rats. The established UPLC-MS/MS method in the range of 2–2,000 ng mL−1 was successfully applied to the pharmacokinetics, and the calculated bioavailability of nobiletin and tangeretin was 63.9 and 46.1%, respectively.
Introduction
The dry and immature fruits of Citrus aurantium L. and its cultivated varieties are widely used in Chinese herbal medicines [1–3]. They have anti-inflammatory, anti-oxidation, and anti-tumor effects [4, 5]. Among them, flavonoids are the main bioactive components. Naringin and neohesperidin are the target compounds for the quality inspection of C. aurantium L. [6, 7]. In addition, the polymethoxyflavonoids, nobiletin and tangeretin, have attracted the attention of the pharmaceutical and nutrition industry due to their antibacterial, anti-inflammatory, neuroprotective, anti-Alzheimer's disease, anti-tumor, and other biological activities [8, 9]. Nobilietin and tannetin are the main active compounds in C. aurantium L. Therefore, monitoring the plasma concentrations of nobiletin and tangeretin is of great significance for the clinical and food application of C. aurantium L.
The detection sensitivity of HPLC method is low, while UPLC-MS/MS has low detection limit [10–12]. Studies on the pharmacokinetics of nobiletin or tangeretin in vivo have been reported [13–18], but the simultaneous detection of nobiletin and tangeretin in plasma by UPLC-MS/MS has not been reported.
The aim of this study was to establish an UPLC-MS/MS method for the determination of nobiletin and tangeretin in rat plasma, and to study pharmacokinetics of nobiletin and tangeretin, and to calculate the absolute bioavailability.
Experimental
Reagents and animals
Nobiletin, tangeretin and vitexin (internal standard) (purity ≥98%, Fig. 1), were purchased from Chengdu Manst Pharmaceutical Co., LTD. HPLC pure formic acid, acetonitrile and methanol were purchased from Merck LTD. Ultrapure water (resistance >18 MΩ) for the experiments was prepared by a Milli-Q purification system, USA. Sprague-dawley rats (male, 220–250g) were from the Animal Experiment Center of Wenzhou Medical University.
Chemical structure of nobiletin (A), tangeretin (B) and vitexin (C)
Citation: Acta Chromatographica 36, 4; 10.1556/1326.2024.01158
Condition of instrument
A Waters XEVO TQ-S micro triple quadrupole tandem mass spectrometer (Waters Corp.) was used to detect nobiletin and tangeretin.
Chromatographic conditions: The column was UPLC HSS T3 (50 × 2.1 mm, 1.7 μm), and the column temperature was set at 40 °C. The mobile phase was acetonitrile-water (containing 0.1% formic acid). 0–0.2 min, acetonitrile 10%; 0.2–1.0 min, acetonitrile 10–70%; 1.0–2.5 min, acetonitrile 70–90%; 2.5–2.8 min, acetonitrile 90–10%; and kept at 10% acetonitrile for 2.8–4.0 min, at a flow rate of 0.4 mL min−1.
Mass conditions: Nitrogen was used as a conical gas (50 L h−1) and desolvation gas (900 L h−1). The capillary voltage was set at 2.5 kV, ion source temperature at 150 °C and desolvation temperature at 450 °C. Electrospray ionization (ESI) positive ion mode was used for detection, and multiple reaction monitoring (MRM) mode was used for quantitative analysis: nobiletin m/z 403.29 → 373.14 (cone voltage 22v, collision voltage 28v), tangeretin m/z 373.28 → 343.17 (cone voltage 20v, collision voltage 26v) and vitexin m/z 433.14 → 313.03 (cone voltage 32v, collision voltage 26v), Fig. 2.
Mass spectrum of nobiletin (A), tangeretin (B) and vitexin (C)
Citation: Acta Chromatographica 36, 4; 10.1556/1326.2024.01158
Standard curve
Stock solutions of nobiletin, tangerein and vitexin (500 μg mL−1) were prepared with methanol, respectively. The stock solutions of nobiletin and tangerein were diluted with methanol to obtain a series of concentration standard working solutions of nobiletin and tangeretin (20, 50, 100, 200, 500, 1,000, 2,000, 5,000, 10,000, 20,000 ng mL−1). The stock solutions of vitexin diluted with methanol to obtain a working solution of vitexin (1,000 ng mL−1). Both the stock solution and the working solution were stored at 4 °C.
An appropriate amount of nobiletin and tangeretin working solution was added to blank rat plasma to be 2, 5, 10, 20, 50, 100, 200, 500, 1,000, 2,000 ng mL−1. Quality control (QC) samples of three plasma concentrations (4, 160 and 1,600 ng mL−1) were prepared by the same method.
Sample pretreatment
Add 100 μL of plasma sample to a 1.5 mL eppendorf tube, add 10 μL of internal standard vitexin (1,000 ng mL−1), add 1 mL of ethyl acetate, vortex mixture for 1.0 min, centrifuge separation (13,000 r min−1, 4 °C, 5 min), take the supernatant, air dry under 60 °C flow. The supernatant was redissolved in 100 μL methanol and centrifuged at 13,000 r min−1 at 4 °C for 5 min. The supernatant was injected into the inner tube of the injection bottle with an injection volume of 2 μL for UPLC-MS/MS analysis.
Methodology validation
The method was validated from the aspects of method selectivity [19], standard curve and lower limit of quantification (LLOQ), precision and accuracy, stability, recovery, and matrix effect [20].
Selectivity
Rat blank plasma from six different sources and corresponding blank plasma prepared LLOQ samples were collected for LC-MS/MS analysis. Whether endogenous substances in plasma interfere with the determination of the tested substance and its internal standard was investigated.
Standard curve
According to “sample pretreatment”, the concentration of each analyte was used as the abscissa, the peak area ratio of the analyte to the internal standard was used as the ordinate, and the weighted (W = 1/x2) least square method was used for regression operation. The linear regression equation obtained was the standard curve.
Precision and accuracy
Three QC samples of low, medium, high concentration and LLOQ were taken and operated under the “sample pretreatment”. Six samples per concentration were analyzed for 3 consecutive days to obtain the intra-day and inter-day precision and accuracy.
Matrix effect
Six samples of blank plasma from different sources were taken, except without internal standard, and the operation was performed under “sample pretreatment”. All the supernatant was taken, and the corresponding concentration of control quality control solution and internal standard solution were added. After mixing by vortex, the supernatant was taken in another clean test tube, and three samples were analyzed for each concentration to obtain the corresponding average peak area. The blank plasma was replaced by water in the same method and processed as above. Three samples were analyzed for each concentration to obtain the corresponding average peak area. The matrix effect was calculated by the peak area ratio of the two treatment methods.
Recovery
Rat blank plasma was used to prepare low, medium, high concentration quality control and LLOQ samples. According to “sample pretreatment”, 6 samples were analyzed for each concentration. At the same time, another 100 µL blank plasma was taken, except for the internal standard solution, according to the “sample pretreatment” under the operation, take all the supernatant, add the corresponding concentration of control quality control solution and internal standard solution, vortex mixing, take the supernatant in another clean test tube, vortex mixing after injection analysis. The treatment recovery was calculated as the ratio of the peak areas of the two treatments at each concentration.
Stability
Plasma sample storage and pretreatment were performed in the dark. In this experiment, the stability of nobiletin and tangeretin plasma samples at low and high concentrations were investigated after being stored in the dark for 48 h at room temperature, and rat plasma samples were stored in the dark for 6 h at room temperature after three freeze-thaw cycles and long-term storage at −20 °C.
Pharmacokinetics
The rats were kept in the environment for 5 days, the temperature was 22–25 °C, the humidity was 55 ± 10%, and no food was given 12 h before the experiment. The animal experiments were conducted according to the animal management guidelines of Wenzhou Medical University. All experimental procedures and protocols were approved by the Animal Care Committee of Wenzhou Medical University (Wydw 2023-0449). Nobiletin and tangeretin were dissolve in 2% DMSO physiological saline solution (1.0 mg mL−1), respectively. Nobiletin and tangeretin were administered intravenously (iv) 1 mg kg−1 and orally (po) 5 mg kg−1, respectively, with 6 rats in each group, for a total of 24 rats. At 0.083 3, 0.5, 1, 2, 3, 4 h, 6, 8 and 12 h time points, 0.3 mL blood was collected from the tail vein, collected in heparinized tubes, and centrifuged at 13,000 r min−1 for 10 min. Then 100 µL of the rat plasma was transferred to a new 1.5 mL eppendorf tube and maintained at −20 °C until analysis. The pharmacokinetic parameters were statistically calculated using the pharmacokinetics software (DAS 2.0).
Results
Selectivity
According to Fig. 3, the retention time of nobiletin, tangeretin and internal standard were 2.14, 2.25 and 1.57 min, respectively. No interference of endogenous components was observed in the retention time of nobiletin, tangeretin and internal standard.
UPLC-MS/MS of nobiletin, tangeretin and vitexin in rat plasma, (A) blank plasma, (B) blank plasma spiked with LLOQ of nobiletin, tangeretin and vitexin, (C) a sample after administration of nobiletin and tangeretin
Citation: Acta Chromatographica 36, 4; 10.1556/1326.2024.01158
Standard curve
Calibration curves for nobiletin and tangeretin in rat plasma showed good linearity in the range of 2–2,000 ng mL−1. The typical regression equation of nobiletin in rat plasma was y1 = 1.4402x1 + 1.303 (r = 0.9976), x1 represents the concentration of nobiletin in plasma, and y1 represents the ratio of the peak area of nobiletin to the internal standard. The typical regression equation of tangeretin in rat plasma was y2 = 1.9339x2 + 2.285 (r = 0.9961), x2 represents the concentration of tangeretin in plasma, and y2 represents the ratio of the peak area of tangeretin to the internal standard. The LLOQ of nobiletin and tangeretin in rat plasma was 2 ng mL−1, and the limit of detection was 0.5 ng mL−1.
Precision, accuracy, recovery, and matrix effect
The intra-day and inter-day precision of nobiletin was within 15%, the intra-day and inter-day accuracy was 90–108%, the recovery was better than 77%, and the matrix effect ranged from 92 to 110%. The intra-day and inter-day precision of tangeretin was within 13%, the intra-day and inter-day accuracy was 92–109%, the recovery was better than 82%, and the matrix effect ranged from 101 to 110%, Table 1.
Accuracy, precision, matrix effect and recovery of nobiletin and tangeretin in rat plasma
| Compound | Concentration (ng mL−1) | Accuracy (%) | Precision (RSD %) | Matrix effect (%) | Recovery (%) | ||
| Intra-day | Inter-day | Intra-day | Inter-day | ||||
| Nobiletin | 2 | 98.9 | 108.0 | 14.0 | 14.0 | 96.9 | 79.5 |
| 4 | 98.0 | 102.6 | 7.7 | 4.0 | 109.4 | 88.4 | |
| 160 | 102.7 | 90.5 | 12.7 | 4.5 | 92.6 | 77.2 | |
| 1,600 | 107.8 | 95.2 | 5.5 | 8.4 | 101.6 | 82.6 | |
| Tangeretin | 2 | 108.6 | 91.6 | 12.0 | 13.9 | 106.0 | 91.2 |
| 4 | 96.9 | 98.4 | 6.5 | 8.1 | 109.6 | 82.2 | |
| 160 | 100.9 | 106.6 | 6.1 | 13.4 | 109.1 | 91.1 | |
| 1,600 | 92.3 | 103.3 | 4.6 | 9.5 | 101.4 | 95.1 | |
Stability
The accuracy of nobiletin was between 85 and 113%, and the RSD was within 15%. The accuracy of tangeretin ranged from 86 to 112%, and the RSD was within 14%, Table 2. These results indicated that nobiletin and tangerein had good stability.
Stability of nobiletin and tangeretin in rat plasma
| Compound | Concentration (ng mL−1) | Autosampler (4 °C, 12 h) | Ambient (2 h) | −20 °C (30 d) | Freeze-thaw | ||||
| Accuracy | RSD | Accuracy | RSD | Accuracy | RSD | Accuracy | RSD | ||
| Nobiletin | 4 | 100.9 | 4.6 | 92.1 | 8.9 | 103.6 | 9.7 | 96.5 | 12.5 |
| 160 | 85.9 | 5.9 | 91.3 | 5.0 | 104.6 | 9.8 | 112.9 | 11.9 | |
| 1,600 | 111.9 | 4.1 | 101.5 | 8.5 | 87.6 | 14.1 | 98.7 | 13.9 | |
| Tangeretin | 4 | 107.7 | 5.9 | 105.8 | 12.5 | 102.0 | 8.1 | 96.6 | 9.9 |
| 160 | 105.4 | 8.7 | 95.3 | 3.2 | 109.7 | 7.0 | 86.5 | 10.9 | |
| 1,600 | 91.3 | 3.2 | 111.0 | 9.8 | 109.9 | 13.6 | 88.1 | 13.2 | |
Pharmacokinetics
The concentration-time curves of nobiletin and tangeretin in rat plasma are shown in Fig. 4. The main pharmacokinetic parameters are listed in Table 3, and oral bioavailability was 63.9 and 46.1%, respectively.
The concentration-time curve of rats after intravenous (iv, 1 mg kg−1) and oral (po, 5 mg kg−1) administration of nobiletin and tangeretin
Citation: Acta Chromatographica 36, 4; 10.1556/1326.2024.01158
Main pharmacokinetic parameters after oral (5 mg kg−1) and intravenous (1 mg kg−1) administration of tangeretin and nobiletin in rats
| Parameters | Unit | Nobiletin | Tangeretin | ||
| po | iv | po | iv | ||
| AUC(0-t) | ng mL−1*h | 2077.0 ± 141.6 | 650.3 ± 35.4 | 167.7 ± 15.1 | 72.7 ± 15.2 |
| AUC(0-∞) | ng mL−1*h | 2079.3 ± 141.9 | 653.7 ± 31.7 | 174.6 ± 19.7 | 78.6 ± 15.9 |
| t1/2z | H | 2.4 ± 1.6 | 4.1 ± 3.4 | 5.3 ± 1.8 | 5.5 ± 2.9 |
| Tmax | H | 1.0 | – | 2.0 | – |
| CLz/F | L h−1 kg−1 | 2.4 ± 0.2 | 1.5 ± 0.1 | 28.9 ± 3.3 | 13.1 ± 2.4 |
| Vz/F | L kg−1 | 8.2 ± 5.6 | 9.3 ± 8.2 | 214.5 ± 56.3 | 101.9 ± 54.1 |
| Cmax | ng mL−1 | 1254.6 ± 167.6 | 555.2 ± 101.1 | 44.7 ± 2.5 | 69.7 ± 22.2 |
AUC: area under the plasma concentration curve; t1/2: elimination half-life; tmax: Peak Time; CL: clearance rate; V: apparent distribution volume; Cmax: maximum plasma concentration.
Discussion
In order to obtain the best mass spectrometry conditions, nobiletin and tangeretin were monitored by positive and negative ion modes, and the responsivity of nobiletin and tangeretin in positive ion mode was higher. Through the standard sample, the capillary voltage and impact energy were optimized and finally obtained.
In order to select the appropriate plasma processing method, the effects of liquid-liquid extraction, solid-liquid extraction, and protein precipitation on the extraction efficiency of analytes were investigated and compared. The solid-liquid extraction method has more extraction recovery and matrix effect results, but the finished product was high and the operation was complex. The protein precipitation method was simple and fast to process plasma samples, but the recovery was not high. Liquid-liquid extraction has good selectivity, small matrix effect can be obtained, and the recovery of extraction was also good. Under the requirements of detection sensitivity, the ethyl acetate liquid-liquid extraction method was selected to process the plasma samples in this experiment.
According to the physicochemical properties and chromatographic behavior of the index components, the elution systems composed of methanol, acetonitrile, water and formic acid were compared. The results show that the separation effect of methanol-water or acid water as the system components is not ideal, and the optimized acetonitrile-formic acid in water system components separation effect was better, each chromatographic peak separation and the theoretical number of plates meet the requirements, the baseline was stable. In addition, the authors choose the gradient elution condition, which can change the retention time of the analyte, so that the analysis can be separated from the co-effluent to reduce the matrix effect.
In the quantitative analysis of biological samples, the deuterated standard was the best internal standard to correct the loss in the extraction process. However, this internal standard was expensive and not always available for sale. Therefore, it was preferred to select a compound with similar structure, extraction and recovery effect and mass spectrum ion response to the analyte. The polarity, mass spectrometry analysis conditions and ion response degree of vitexin were like those of nobiletin and tangeretin. Therefore, vitexin was selected as the internal standard in this experiment.
Li et al. used astragaloside IV as the internal standard, a sensitive and reliable LC-MS/MS method was established for the determination of paeoniflorin, nobiletin, tangeretin, liquiritigenin in plasma and tissues of rats after oral administration of Si-Ni-San extract [13]. However, the bioavailability of nobiletin and tangeretin was not report.
Kumar et al. developed a highly sensitive and rapid method for the determination of nobiletin in rat plasma by LC-MS/MS [14]. The assay was performed by extracting nobiletin and citalopram (internal standard) from rat plasma using the liquid-liquid method. An isodense mobile phase (0.2% formic acid-acetonitrile, 20:80, v/v) was used for chromatographic separation. The flow rate was 0.6 mL min−1 on an Atlantis C18 column (maintained at 40 °C) with a total running time of 2.0 min. Absolute oral bioavailability of nobiletin was 35.9%.
Elhennawyd et al. developed a simple HPLC method for the quantification of tangeretin in the plasma of Sprague-Dawley rats[17]. The LLOQ of tangeretin was 15 ng mL−1, and the pharmacokinetic characteristics of tangeretin were studied. After a single intravenous administration (10 mg kg−1), tangeretin was rapidly cleared (Cl = 94.1 ± 20.2 mL min−1 kg−1) with a moderate terminal elimination half-life (t1/2 λz = 166 ± 42 min). When administered as a suspension (50 mg kg−1), the absolute oral bioavailability of tangeretin was poor but unstable (mean <3.05%). However, when tangeretin was treated with randomly methylated β-cyclodextrin (50 mg kg−1), its plasma exposure was at least doubled (mean bioavailability: 6.02%). The aqueous solubility obviously hinders the oral absorption of tangeretin and acts as a barrier to its oral bioavailability.
MANTHEY et al. used HPLC-ESI-MS method to determination of nobiletin, tangeretin and their main metabolites after gavage and intraperitoneal (ip) injection to rats [16]. Administration of nobiletin dissolved in corn oil by gavage produced a rapid appearance of nobiletin (9.3 μg mL−1 at 0.5 h) in the rat blood sera, in which the levels of nobiletin remained >4 μg mL−1 at 4 h, >3 μg mL−1 at 8 h, and ∼2 μg mL−1 at 12 h. However, the bioavailability of nobiletin and tangeretin was not calculated, and the HPLC-ESI-MS was not full validated.
Huang et al. investigated the pharmacokinetics, bioavailability, distribution, and excretion of tangeretin in rats by HPLC [21]. After oral administration of 50 mg kg−1 tangeretin to rats, the Cmax, Tmax and t1/2 were 0.87 ± 0.33 mg mL−1, 340.00 ± 48.99 min, and 342.43 ± 71.27 min, respectively. Absolute oral bioavailability was calculated to be 27.11%.
In this study, after oral administration of 5 mg kg−1 nobiletin to rats, the Cmax, Tmax and t1/2 were 1254.6 ± 167.6 ng mL−1, 1.0 h, and 2.4 ± 1.6 h. After oral administration of 5 mg kg−1 tangeretin to rats, the Cmax, Tmax and t1/2 were 44.7 ± 2.5 ng mL−1, 2.0 h, and 5.3 ± 1.8 h. Tmax of tangeretin was not same as reported by Huang [21], and the t1/2 almost the same. The bioavailability of nobiletin and tangeretin was 63.9 and 46.1% in this study, respectively. Absolute oral bioavailability of nobiletin was 35.9% reported by Kumar [14], and tangeretin, was 27.11% reported by Huang [21]. However, there are no studies on the simultaneous determination of nobiletin and tangeretin by UPLC-MS/MS.
Conclusion
In this study, an UPLC-MS/MS method for the determination of nobiletin and tangeretin in rat plasma was established. The method has been verified, and applied to the pharmacokinetic study of rat plasma, and the bioavailability was calculated to be 63.9 and 46.1%.
Conflict of interest statement
The author(s) declare(s) that there is no conflict of interest regarding the publication of this paper.
Data availability statement
The data used to support the findings of this study are included within the article.
Acknowledgements
This work was supported by the Scientific Research Project of Affiliated Hospital of Zhejiang Chinese Medical University (2022FSYYZY25), Zhejiang Provincial Key Laboratory of Traditional Chinese Medicine.
References
- 1.↑
Gong, Y.; Liang, X.; Dai, Y.; Huang, X.; Su, Q.; Ma, Y.; Chen, F.; Wang, S. Pharm. Biol. 2023, 61, 345–355.
- 2.
Zhou, P.; Li, X.; Jiang, Z.; Zhou, J.; He, G.; Qu, L. Int. J. Biol. Macromol 2023, 227, 241–251.
- 3.
Miao, W.; Liu, X.; Li, N.; Bian, X.; Zhao, Y.; He, J.; Zhou, T.; Wu, J. L. Food Chem. 2023, 405, 134988.
- 4.↑
Ghods, A. A.; Sotodeh-Asl, N.; Zia, H.; Ghorbani, R.; Soleimani, M.; Vaismoradi, M. Healthcare (Basel) 2022, 10.
- 5.↑
Gao, L.; Zhang, H.; Yuan, C. H.; Zeng, L. H.; Xiang, Z.; Song, J. F.; Wang, H. G.; Jiang, J. P. Front Pharmacol. 2022, 13, 983470.
- 6.↑
Yan, Y.; Zhou, H.; Wu, C.; Feng, X.; Han, C.; Chen, H.; Liu, Y.; Li, Y. Prep. Biochem. Biotechnol. 2021, 51, 780–791.
- 8.↑
Wang, M.; Meng, D.; Zhang, P.; Wang, X.; Du, G.; Brennan, C.; Li, S.; Ho, C. T.; Zhao, H. J. Agric. Food Chem. 2018, 66, 3155–3160.
- 11.
Chen, H.; Wang, H.; Liu, H.; Xu, X.; Wen, C.; Li, X. Latin Am. J. Pharm. 2019, 38, 2428–2433.
- 12.
Li, Y.; Lin, Z.; Peng, X.; Chen, W.; Wen, C.; Song, H.; Tong, S.; Wang, S. Latin Am. J. Pharm. 2016, 35, 2309–2313.
- 13.↑
Li, T.; Yan, Z.; Zhou, C.; Sun, J.; Jiang, C.; Yang, X. Biomed. Chromatogr. 2013, 27, 1041–1053.
- 14.↑
Kumar, A.; Devaraj, V. C.; Giri, K. C.; Giri, S.; Rajagopal, S.; Mullangi, R. Biomed. Chromatogr. 2012, 26, 1464–1471.
- 15.
Singh, S. P.; Wahajuddin; Tewari, D.; Patel, K.; Jain, G. K. Fitoterapia 2011, 82, 1206–1214.
- 16.↑
Manthey, J. A.; Cesar, T. B.; Jackson, E.; Mertens-Talcott, S. J. Agric. Food Chem. 2011, 59, 145–151.
- 18.
Li, W.; Wang, S.; Wang, W.; Gong, L.; Ni, D.; Li, Y.; Wu, W.; Zhang, Y.; Xu, X.; Jiang, Q.; Zhang, J.; Zhang, T. J. Sep. Sci. 2022, 45, 3985–3994.
- 19.↑
(FDA), U. F. a. D. A., 2018. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/bioanalytical-method-validation-guidance-industry.
- 20.↑
Chen, F.; Ma, Y.; Cui, Y.; Wang, W.; Mei, C.; Nie, J.; Wen, C.; Shen, X.; Zhou, X. Int. J. Anal. Chem. 2023, 2023, 4747771.
- 21.↑
Hung, W. L.; Chang, W. S.; Lu, W. C.; Wei, G. J.; Wang, Y.; Ho, C. T.; Hwang, L. S. J. Food Drug Anal 2018, 26, 849–857.
