2015 Volume 21 Issue 3 Pages 473-477
Soy isoflavones are generally analyzed in accordance with the official method of AOAC International (OMA) 2001.10 and OMA 2008.03. However, this method is time-consuming and requires a large amount of sample. Therefore, the development of a better analytical method for soy isoflavones is required. In this study, OMA 2001.10 and OMA 2008.03 were improved to develop a quantification method for soy isoflavones. As a result, we shortened the analysis time using HPLC. The isoflavone content of a soybean sample was measured according to the improved method in triplicate, and the measurements were repeated on three different days. Both the repeatability and intermediate precision (within laboratory) for the quantification of the isoflavone content (aglycone) of the soy sample were good. The recovery of isoflavones spiked into a soy sample at two different concentrations ranged from 94% to 105%. The multilaboratory validation study with 7 participating laboratories showed satisfactory interlaboratory precision; all HorRat values were < 2.
Soy contains isoflavone aglycones such as genistein, daidzein and glycitein, and their glycosides such as genistin, daidzin and glycitin (Munro et al., 2003). Recently, the antioxidant, cancer-growth inhibition, and other health promotion effects of soy isoflavones have been elucidated (Perabo et al., 2008, Cuevas et al., 2003, Nagata et al., 2002). A microbiota metabolite of isoflavone glycoside showed growth inhibitory activity for cancer cells and anti-inflammatory activity (Atkinson et al., 2005, Hirata et al., 2013, Shimoda and Hamada et al., 2011). Moreover, soy isoflavones are recognized to be plant estrogens, binding estrogen receptors to induce physiological effects. Japanese people consume isoflavones mainly from soy in everyday life; no other Japanese food contains such a high amount of isoflavones. The daily consumption of soy isoflavones is limited to 70∼75 mg aglycone equivalent, as defined by the Food Safety Commission of Japan, because isoflavones are endocrine disruptors of human reproductive systems (Clarke et al., 2003). Therefore, it is important to analyze the isoflavone concentration in common foods and “food for specific health uses” rapidly and with high precision (Cederroth et al., 2012). However, the quantitative analysis of isoflavones by the official methods of AOAC international (OMA) 2001.10 and OMA 2008.03 is both time-consuming and requires a large amount of soy sample (i, Collison et al., 2008). With the OMA 2001.10 and OMA 2008.03 methods, isoflavone content in soy samples can be quantified using 2 to 10 g of sample, and the analysis time by HPLC is 44.5 min and 63 min, respectively. Thus, the conventional method is not suitable for the analysis of large numbers of samples.
Accordingly, the development of a better analysis method for quantification of isoflavones in soy samples is required.
In this study, we developed a quantification method for soy isoflavones requiring shorter analysis time and less sample than the conventional method.
Chemicals Daidzin, daidzein, genistin, genistein, glycitin and glycitein were purchased from Nacalai Tesque (Kyoto, Japan) and used as the standards. All the chemicals were > 97% purity based on HPLC analysis.
Preparation of soybean test samples Soybean test samples were prepared as described by Kuribara et al. (2002). “Fukuyutaka” was harvested in Fukuoka Prefecture, while “Yukipirika” and “Yukihomare” were harvested at the Tokachi Agricultural Experiment Station, Local Independent Administrative Agency Hokkaido Research Organization. “Fukuyutaka” and “Yukihomare” are the most frequently produced in Japan and Hokkaido district, respectively. “Yukipirika” is bred as a high isoflavone content variety. These three soybean varieties were harvested in 2010 and stored at 4°C until use. Soybeans (30 g) were frozen overnight at −80 °C, and then ground using a Rotor-Speed MillP14 (Fritsch GmbH, Idar-Oberstein, Germany), passed through a 0.5 mm filter, and freeze-dried using a freeze dryer. Powdered soybean samples were thoroughly mixed, divided into 8 portions, and placed in plastic bags. Freeze-drying and mixing steps were repeated twice. Finally, soy powder samples were stored in aluminum laminated bags at −80°C. The soy powder particles were analyzed using a laser diffraction particle size analyzer (SALD-2000; Shimadzu Co., Kyoto, Japan). The moisture content was analyzed using a Karl Fischer moisture meter (KF-100; Mitsubishi Chemical Co., Tokyo, Japan).
Quantification of isoflavone content in soy samples Soybean test samples (“Fukuyutaka”, “Yukipirika”, and “Yukihomare”) were extracted following a method for the analysis of crude materials in OMA 2001.10 (?), with modifications. Approximately 500 mg of soybean powder was weighed into a glass vial (30 mL) and 16 mL of extraction solvent (methanol/water = 80:20 (v/v)) was added. After mixing by sonication, extraction was performed by heating at 65 ± 2°C for 120 min in the vial capped with a Teflon-faced septum. After cooling down to room temperature, 1.2 mL of 2 mol/L NaOH was added. After mixing for 10 min, 0.4 mL of glacial acetic acid was added to hydrolyze ester linkages of isoflavones. Then, the extract volume was adjusted to 25 mL with the extraction solvent, and the content was mixed thoroughly. The extract was then transferred to a centrifuge tube and centrifuged at 7,000×g for 5 min. The resulting supernatant (250 µL) was collected in a centrifuge tube (1.5 mL), methanol (300 µL) and water (450 µL) were added, and the sample was mixed using a vortex mixer. The mixture was centrifuged at 7,000×g for 5 min, and the supernatant was used for HPLC analysis. The isoflavones in the extract were separated and quantified by HPLC using a Phenomenex Prodigy ODS (3) column (4.6 × 250 mm; Phenomenex, Torrance, CA, USA), a Shiseido Capcell Pak C18 Type MG II column (4.6 × 200 mm; Shiseido, Tokyo, Japan) or a YMC-Pack ODS-AM column (4.6 × 250 mm; YMC, Kyoto, Japan). An aliquot of the isoflavone extract (20 µL) was injected and the flow rate of the mobile phase was 1.5 mL/min. The mobile phase, consisted of eluent A: 86.5% water, 10% acetonitrile, 3% methanol, and 0.5% glacial acetic acid, and eluent B: 45.6% water, 50% acetonitrile, 3% methanol, and 0.5% glacial acetic acid, and was freshly prepared each time of use. The sample was eluted by gradient elution as follows: 12.5 – 32.5% of B (linear), 0 – 10 min; 32.5 – 67.5% of B (linear), 10 – 12 min; 67.5 – 100% of B (linear), 12 – 14 min; 100% of B, 14 – 16 min. Then, the mobile phase returned to the initial composition (12.5% of B, 16.0 – 23 min) and conditioning. The total analysis time was 23 min. The isoflavones were detected using a UV detector set at 260 nm. Isoflavones content was expressed on a total aglycone basis by summing the concentrations of aglycone isoflavones (genistein, glycitein, and daidzein) and the aglycone equivalents of the corresponding glucoside forms (genistin, glycitin, and daidzin). A typical HPLC chromatogram of the isoflavones is shown in Fig. 1. We shortened the analysis time by HPLC in OMA 2001.10 and OMA 2008.03 to 21.5 min and 40 min, respectively.
Typical HPLC chromatogram of the isoflabones isolated from soybean test sample: daidzin (1), glycitin (2), genistin (3), daidzein (4), glycitein (5), genistein (6)
Single laboratory validation A methanol solution containing daidzin, daidzein, genistin, genistein, glycitin, and glycitein (isoflavone mixture) was prepared at concentrations of 84, 371, 98, 373, 33, and 64 µg/mL, respectively. The methanol solution was spiked into the soy samples (“Fukuyutaka”) at 0 µg/g (non-spiked, 4 mL of methanol was added and air dried), 831 µg/g, and 1663 µg/g total aglycone (in which two doses of isoflavone mixture methanol solution were added and air dried). Spiked and non-spiked samples were extracted as described above, and then the isoflavone concentration (total aglycone) and recovery were calculated. Triplicate measurements were obtained on three different days.
Multilaboratory validation The linearity of the standard curve of isoflavones (daidzin, glycitin, genistin, daidzein, glycitein and genistein) were assessed by the coefficient of determination (R2).
A multilaboratory validation study was conducted with seven participating laboratories and three kinds of soy powder samples (“Fukuyutaka”, “Yukipirika”, and “Yukihomare”)
Sample homogeneity was verified prior to the multilaboratory validation study. Approximately 500 mg of each soy sample were drawn from an aluminum laminated bag, divided into two portions (250 mg each) and placed in glass vials. Further extraction and analysis were conducted as described above.
For the multilaboratory validation study, the participating laboratories were part of the grant-in-aid for the research project shown in the acknowledgments section. The blind duplicate samples were randomized. Participants were also provided with a method protocol as well as an electronic evaluation and reporting sheet (MS Excel format). Six test samples were analyzed once under repeatability conditions.
The isoflavone contents of each sample were measured under repeatability conditions, i.e., the same method was conducted on identical test items in the same laboratory by the same operator using the same equipment within a short period of time.
Statistical analysis The repeatability standard deviation, reproducibility standard deviation, intermediate standard deviation and recovery were determined by a one-way analysis of variance (ANOVA).
Precision within a laboratory The isoflavone content of a soy sample (Fukuyutaka) was measured in triplicate, and the measurements were carried out on three different days. The mean isoflavone content (total aglycone) was 1400 µg/g dry weight, the relative repeatability standard deviation (RSDr) was 2.2%, and the relative intermediate standard deviation (RSDint) was 2.0% (Table 1). The RSDr value was compared with the predicted level of precision obtained from the Horwitz equation.
| Isoflavone content/µg g−1 | |||
|---|---|---|---|
| Day1 | Day 2 | Day 3 | |
| 1 | 1369 | 1414 | 1462 |
| 2 | 1404 | 1401 | 1388 |
| 3 | 1385 | 1411 | 1366 |
| Mean / mg g−1 | 1400 | ||
| Sr / µg g−1 a | 31 | ||
| RSDr % b | 2.2 | ||
| Sint / µg g−1 c | 31 | ||
| RSDint % d | 2.0 | ||
| PRSDr % e | 2.6 | ||
 |
where C is the most commonly measured concentration of the analyte in the sample, expressed as a mass fraction. The predicted RSDr (PRSDr) was calculated to be 2.6%. The HorRat values for the isoflavone contents in the “Fukuyutaka” sample was 1.6. The HorRat values for the soy samples were < 2. Thus, both repeatability and intermediate precision within a laboratory for isoflavone (total aglycone) quantification of the soy sample were good.
Spike and recovery test The average recoveries of spiked isoflavone at concentrations of 831 and 1663 µg/g were 105% and 94%, and the RSDr and RSDint were 5.0% and 3.0%, and 6.1% and 5.3%, respectively (Table 2). According to the single laboratory validation guidelines of AOAC (ii), the recovery limit of the spiked analytes was 80 – 110%. These results clearly indicate that the trueness of the method for quantifying the isoflavone content of soy samples can be judged to be good.
| Spiked concentration, µg g−1 | Recovery, % | |||||
|---|---|---|---|---|---|---|
| 831 | 1663 | |||||
| Day1 | Day2 | Day3 | Day1 | Day2 | Day3 | |
| 1 | 90 | 120 | 109 | 97 | 95 | 88 |
| 2 | 101 | 115 | 101 | 96 | 96 | 86 |
| 3 | 109 | 106 | 97 | 93 | 101 | 90 |
| Mean recovery, % | 105 | 94 | ||||
| Sr a | 5.2 | 2.8 | ||||
| RSDr %b | 5.0 | 3.0 | ||||
| Sint c | 6.4 | 5.0 | ||||
| RSDint % d | 6.1 | 5.3 | ||||
Multilaboratory validation Standard curves were obtained for all isoflavones with high linearity (R2 ≥ 0.99, data not shown). Linearity of daidzin, glycitin, genistin, daidzein, glycitein and genistein were obtained in the range of 0.48 – 7.46 µg/mL, 0.2 – 3.21 µg/mL, 0.49 – 8.15 µg/mL, 2.02 – 31.9 µg/mL, 0.49 – 7.69 µg/mL, and 1.95 – 32 µg/mL, respectively.
For the homogeneity test, each sample was analyzed in the same laboratory by the same operator using the same equipment within a short period of time. The homogeneity of the isoflavone contents of the three soy samples was analyzed according to the FAPAS protocol (Thompson et al., 2006). F1σ2all in each soy sample was confirmed to be smaller than F2San2, indicating that the samples were homogeneous (Table 3).
| Isoflavone content/µg g−1 | ||||||
|---|---|---|---|---|---|---|
| Sample No. | Fukuyutaka | Yukipirika | Yukihomare | |||
| 1 | 2 | 1 | 2 | 1 | 2 | |
| 1 | 1827 | 1849 | 2731 | 2709 | 1652 | 1577 |
| 2 | 1843 | 1869 | 2958 | 3105 | 1794 | 1567 |
| 3 | 1858 | 1858 | 2931 | 2761 | 1736 | 1809 |
| 4 | 1890 | 1882 | 2790 | 3023 | 1855 | 1756 |
| 5 | 1844 | 1875 | 2947 | 3011 | 1751 | 1623 |
| 6 | 1813 | 1843 | 3064 | 3050 | 1763 | 1752 |
| 7 | 1845 | 1860 | 3088 | 3032 | 1693 | 1778 |
| 8 | 1884 | 1877 | 3063 | 3023 | 1681 | 1822 |
| 9 | 1829 | 1851 | 2631 | 3051 | 1794 | 1660 |
| 10 | 1854 | 1878 | 3039 | 2870 | 1801 | 1617 |
| Mean / µg g−1 | 1857 | 2944 | 1724 | |||
| San2 a | 2197 | 83636 | 159579 | |||
| Ssam2 b | 1912 | 2944 | 1724 | |||
| σp c | 95 | 89 | 141 | |||
| σ2all d | 17216 | 15180 | 37672 | |||
| C e | 19436 | 198878 | 99669 | |||
The isoflavone contents of the three soy samples were measured in duplicate in seven different laboratories. The mean isoflavone contents on a total aglycone basis of the “Fukuyutaka”, “Yukipirika”, and “Yukihomare” samples were 1673, 2911, and 1675 µg/g dry weight, respectively, and the RSDr and RSDR were 1.4%, 1.3%, and 1.7%, and 9.9%, 7.4%, and 7.3%, respectively (Table 4). The precision data obtained in the interlaboratory study were compared with the predicted levels of precision obtained from the Horwitz equation.
| Lab. No. | Fukuyutaka | Yukipirika | Yukihomare | |||
|---|---|---|---|---|---|---|
| 1 | 2 | 1 | 2 | 1 | 2 | |
| 1 | 1803 | 1795 | 3014 | 3041 | 1773 | 1803 |
| 2 | 1741 | 1729 | 3045 | 3071 | 1717 | 1699 |
| 3 | 1784 | 1783 | 3028 | 2973 | 1749 | 1779 |
| 4 | 1378 | 1329 | 2461 | 2452 | 1421 | 1485 |
| 5 | 1571 | 1504 | 2867 | 2784 | 1535 | 1591 |
| 6 | 1725 | 1763 | 2950 | 3044 | 1694 | 1749 |
| 7 | 1760 | 1755 | 3002 | 3027 | 1725 | 1729 |
| Mean / µg g−1 | 1673 | 2911 | 1675 | |||
| Number of laboratories | 7 | 7 | 7 | |||
| Sr / µg g−1a | 24 | 38 | 29 | |||
| RSDr % b | 1.4 | 1.3 | 1.7 | |||
| SR / µg g−1c | 166 | 215 | 123 | |||
| RSDR % d | 9.9 | 7.4 | 7.3 | |||
| PRSDR % e | 5.2 | 4.8 | 5.2 | |||
| HorRat f | 1.9 | 1.5 | 1.4 | |||
| k × SR g | 322 | 430 | 246 | |||
 |
where C is the most commonly measured concentration of the analytes in the samples, expressed as a mass fraction. The PRSDR values for the “Fukuyutaka”, “Yukipirika”, and “Yukihomare” samples were 5.2%, 4.8%, and 5.2%, respectively.
The HorRat values for the isoflavone contents in “Fukuyutaka”, “Yukipirika”, and “Yukihomare” were 1.9, 1.5, and 1.4, respectively (Table 4). All the HorRat values for the soy samples were < 2, indicating that our improved procedure can be used to quantify the isoflavone contents in soy samples with satisfactory interlaboratory precision.
In this study, we confirmed that validation of our improved procedure for the determination of total isoflavone concentration in the range of 1673 – 2911 µg/g in “Fukuyutaka”, “Yukipirika”, and “Yukihomare”. Expanded measurement uncertainty (k × SR) computed from SR (reproducibility standard deviation) and coverage factor (k = 2.2622) obtained from our improved procedure were in the range of 278 – 488 µg/g (Table 4).
Acknowledgements This work was supported by a Research Project for New Demand Creation of Agricultural Products of the Ministry of Agriculture, Forestry and Fisheries. We express our appreciation to the following collaborators for their participation in the study: Takashi Yamagishi (National University Corporation Kitami Institute of Technology), Mayumi Onishi-Kameyama and Hiroshi Yada (National Agriculture and Food Research Organization, National Food Research Institute). We would also like to thank Ms. Yoshiko TOYAMA and Ms. Michiko YAMAMOTO for their technical assistance in our laboratory.
