Sample preparation for the analysis of isoflavones from soybeans and soy foods

Journal of Chromatography A, 1216 (2009) 2–29

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Review

Sample preparation for the analysis of isoflavones from soybeans and soy foods M.A. Rostagno a,∗ , A. Villares a , E. Guillamón a , A. García-Lafuente a , J.A. Martínez a,b a Centro para la Calidad de los Alimentos, Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria (INIA), Campus Universitario “Duques de Soria”, 42004 Soria, Spain b Universidad de Navarra, Dpto. Fisiología y Nutrición, Edificio de Investigación, C/Irunlarrea, 1, 31008 Pamplona, Spain

a r t i c l e

i n f o

Article history: Received 6 August 2008 Received in revised form 3 November 2008 Accepted 13 November 2008 Available online 19 November 2008 Keywords: Reviews Isoflavones Soybeans Sample conservation Sample preparation Extraction Analysis

a b s t r a c t This manuscript provides a review of the actual state and the most recent advances as well as current trends and future prospects in sample preparation and analysis for the quantification of isoflavones from soybeans and soy foods. Individual steps of the procedures used in sample preparation, including sample conservation, extraction techniques and methods, and post-extraction treatment procedures are discussed. The most commonly used methods for extraction of isoflavones with both conventional and “modern” techniques are examined in detail. These modern techniques include ultrasound-assisted extraction, pressurized liquid extraction, supercritical fluid extraction and microwave-assisted extraction. Other aspects such as stability during extraction and analysis by high performance liquid chromatography are also covered. © 2008 Elsevier B.V. All rights reserved.

Contents 1. 2. 3. 4. 5.

6. 7. 8.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . General aspects of soy isoflavones determination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sample stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Extraction techniques and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Solid and semi-solid samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.1. Conventional extraction methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.2. Modern extraction techniques and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2. Liquid samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3. Optimization of extraction conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4. Critical comparison of extraction methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Post-treatment of extracts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Separation approaches/techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 4 4 6 7 7 7 12 21 22 23 25 26 27 28 28

Abbreviations: ε, Dielectric constant; ACE, Acetone; ADi, Acetyl daidzin; AGi, Acetyl genistin; AGly, Acetyl glycitin; ASE, Accelerated solvent extraction; CE, Capillary electromigration techniques; De, Daidzein; Di, Daidzin; DMSO, Dimethylsulfoxide; DSM, Defatted soybean meal; EtOH, Ethanol; Ge, Genistein; Gi, Genistin; Gle, Glycitein; Gly, Glycitin; MAE, Microwave-assisted extraction; MeCN, Acetonitrile; MeOH, Methanol; MGi, Malonyl genistin; MDi, Malonyl Daidzin; MGly, Malonyl glycitin; PLE, Pressurized liquid extraction; PSE, Pressurized solvent extraction; SC-CO2 , Supercritical CO2 ; SFE, Supercritical fluid extraction; SPE, Solid phase extraction; SPI, Soy protein isolate; SPME, Solid phase microextraction; SWE, Superheated water extraction; UAE, Ultrasound-assisted extraction. ∗ Corresponding author. Tel.: +34 975 233204; fax: +34 975 233205. E-mail address: [email protected] (M.A. Rostagno). 0021-9673/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2008.11.035

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

3

1. Introduction Functional foods are one of the most promising fields concerning nutritional sciences. These food-stuffs are interesting from the consumer point of view with the prospect of maintaining health and preventing diseases by using natural foods as part of the habitual diet, and also from the industry point of view, for the added value of the products. There are several raw materials that can be used for healthy purposes and soybeans are among those with the greatest potential. Soybeans are one of most produced and commercialized commodities worldwide. Actually, there are several foods derived or based on soybeans such as soy milk, tofu and tempeh, and the consumption and use of soybeans (texturized soy protein, concentrated soy protein and soy protein isolate) as additives by the food industry is increasing every year [1–4]. The potential of soybeans as a functional food is being currently explored by the food industry. Indeed, soybeans and soy foods, like soymilk, tofu, miso and tofu, are widely promoted and eaten based on assumed relationships between its consumption and beneficial health effects in humans including chemoprevention of breast and prostate cancer, osteoporosis, cardiovascular disease as well as relieving menopausal symptoms. Evidence provided not only by epidemiological studies showing a lower incidence of these health conditions in Asian countries like Japan and China, which have high soy consumption, but also from intervention studies, is the basis of this relationship [5–12]. During the last decades our knowledge about the dietary impact on health and well-being has been highly increased and often related to specific food components. Several classes of phytochemicals have been identified in soybeans, including protease inhibitors, phytosterols, saponins, phenolic acids, phytic acid and isoflavones [13–16]. Of these, isoflavones are particularly noteworthy because soybeans are the only significant dietary source of these compounds. Isoflavone content in soybeans can range from 0.4 mg to 9.5 mg of total isoflavones per gram, which can be influenced by genetics, crop year and growth location [17–19]. More importantly, these compounds have shown several in vitro and in vivo beneficial properties consistent with the potential soybean effects on health. There are several possible mechanisms of action by which isoflavones may act on disease prevention, including estrogenic/ anti-estrogenic activity, cell anti-proliferation, induction of cellcycle arrest and apoptosis, prevention of oxidation, antiinflammatory, regulation of the host immune system, and changes in cellular signaling [7,20–28]. The actual mechanisms in the human organism have not been fully established and metabolism may play an important role. Furthermore, besides of evidence of available epidemiological or intervention studies and “in vitro” observations, there are several reports indicating that several of the specific potential soybean health benefits are linked to isoflavone intake [8,29–32]. However, there is still controversy and an unanimous position about if isoflavones, other soy phytochemicals or components are responsible for the health benefits of soy consumption is still far from being reached. Because the data in humans are not conclusive for any of these possible benefits, it is important to conduct more studies investigating isoflavones and soy foods in the diet to health outcomes. An accurate food composition database is crucial for such studies. That is the reason why there is an increasing interest of scientists focused in developing newer extraction and analysis methods for the characterization of soybean functional components, especially isoflavones, and about the relationships between their consumption and beneficial health effects in humans. Isoflavones are a subclass of flavonoids and are also described as phytoestrogen compounds, since they exhibit estrogenic activ-

Fig. 1. Chemical structures of soybean isoflavones and abbreviations.

ity (similar effects to estradiol hormones). The basic characteristic isoflavone structure is a flavone nucleus, composed by two benzene rings (A and B) linked to a heterocyclic ring C (Fig. 1). The benzene ring B position is the basis for the categorization of the flavanoid class (position 2) and the isoflavonoid class (position 3). The main isoflavones found in soybeans are genistein (4 ,5,7-trihidroxyisoflavone), daidzein (4 ,7-dihidroxyisoflavone), glycitein (4 ,7-dihidroxy-6-metoxi-isoflavone) and their respective acetyl, malonyl and aglycone forms (Fig. 1) [33–39]. Biochanin A and formononetin (which are derivatives of genistein and daidzein) are generally less abundant in soy than the 12 main forms and which are found mostly in clover and alfalfa sprouts [40]. Isoflavone content of available soy foods in several countries is been intensively investigated. Quantification of isoflavones in the soybeans and soy foods consumed in the USA [40–44], Japan [45,46], Italy [47], UK [48,49], Singapore [43,50], Australia [51], Indonesia [50,51], Brazil [52], and Canada [53] have been published in the last decade. Besides of individual reports, there are food composition databases and compilations from these values specifically focusing on isoflavone distribution [54–62]. These reports supply useful information to investigators determining the intake of phytoestrogens in order to relate intakes to potential biological activities. Also, they can be used by health professionals and consumers to estimate individual phytoestrogens intake and design personalized diets in order to achieve biologically active concentrations of these functional compounds.

4

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

When the intake of isoflavones is estimated, the quality of the food composition database is important. This is critical in the case of foods consumed regularly, in large quantities, or containing ingredients with concentrated amounts of phytoestrogens. Future analyses of the isoflavone content of basic ingredient foods and commercial items commonly consumed in the diet will enable more accurate estimates of phytoestrogen intake and obtain reliable conclusions about their role on health [56,58,63]. Due to the enormous efforts done in the last few years to evaluate isoflavone composition in foods and its relation with nutritional issues and health effects it is of ultimate importance to develop reliable and precise methods for the quantification of these compounds in foods. Because of the increasing complexity of the food supply, there are major challenges in collecting reliable food consumption data for phytoestrogen intake estimates. Several extraction methods have been used for quantification purposes without adequate validation of the extraction procedure and far from optimized extraction conditions, which can lead to erroneous measurements and calculations. Besides, optimal extraction conditions can be used to save time, resources and provide reliable information. Moreover, only scattered data are available in the scientific literature and a review of the subject is needed to provide essential information on the topic and to identify future research fields of action. Therefore, the aim of the present manuscript is to provide a critical review of the actual state, the most recent advances as well as current trends and future prospects in sample preparation and analysis for the quantification of isoflavones from soybeans and related foods. 2. General aspects of soy isoflavones determination The four common steps for any analytical method are sampling, sample preservation, sample preparation and analysis. Fig. 2 presents a general overview of the most common steps for sample preparation for the determination of soy isoflavones. The initial step in any analysis is sampling, where a representative sample is collected from the entire sample matrix that needs to be analyzed. The entire food-stuff should be represented in the sample that will be used for the analysis. Sample preservation is an important step as there is often some delay between sample

collection and/or preparation and analysis. Proper sample preservation ensures that the sample retains its physical and chemical characteristics from the time it is collected to the time it is analyzed. Sample preparation may consist of multiple steps such as drying, homogenization, sieving, extraction of target compounds, pre-concentration, hydrolysis and derivatization. Sample preparation can seek several objectives: to increase the efficiency of an assay procedure, to eliminate or reduce potential interferences, to enhance the sensitivity of the analytical procedure by increasing the concentration of the analyte in the assay mixture, and sometimes to transform the analyte of interest to a more suitable form that can be easily separated, detected, and/or quantified. Isoflavone determination is complex since its concentration in the sample depends of several variables which may difficult the determination. Overall, the ultimate goal is to obtain a concentrated extract with all isoflavones and free of interfering compounds from the matrix [64–66]. The quantification of isoflavones in solid samples is usually performed by extracting isoflavones from the food matrix using a certain solvent and then analyzing the extract by one of the several analysis techniques available, including gas chromatography, high performance liquid chromatography (HPLC) and immunoassay, among others. The most used analysis technique is, without doubt, reverse-phase HPLC using C18 based columns with water and methanol or acetonitrile containing small amounts of acid as the mobile phase. The extraction phase is extremely important and the process will depend of analyte liberation from the matrix, which will allow quantitative determinations of target compounds. Moreover, the extract should mimetic the original isoflavone composition and profile as much as possible. For the efficient extraction several parameters should be defined like the solvent, temperature, sample amount and time. Optimization of the extraction conditions is normally accomplished using the classical one-variable-at-a-time method, in which the optimization is directly assessed by systematic alteration of one variable, while the others are kept constant. Some authors use experimental designs for the determination of interactions between parameters and selecting the most suitable extraction conditions while minimizing the number of experiments. In the experimental design strategies the values of all the factors under study are varied in each assay in a programmed and rational way. It is thus possible to detect the influencing factors while the number of trials can be kept to a minimum [67,68]. 3. Sample stability

Fig. 2. Most common steps for sample preparation for the determination of soy isoflavones.

In analytical practices, the importance of sample conservation must be emphasized. Indeed, if not carefully controlled can lead to errors that cannot be corrected afterwards since will consequently affect the outcome of the final analysis. Thus, the results obtained, instead of being the source of information, can produce misinformation. Often too little attention is given to the handling of soybean, soy foods or isoflavone extract samples after their collection and before the actual instrumental analysis. How and for how long different samples can be stored to preserve their original isoflavone profile is particularly important since some isoflavones have a relatively unstable character. Chemical changes of isoflavone structures have been reported to occur during the processing of soybeans and soy products. The most frequently observed chemical changes of isoflavones during the processing are decarboxilation of malonyl glucosides to acetyl glucosides and ester hydrolysis of malonyl and acetyl glucosides to underivatized glucoside. It is also possible

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

5

Fig. 3. Most common possible degradation paths of soybean isoflavones.

for all different conjugated forms to generate the aglycone forms by cleavage of the glucosidic bond [69–73]. An overview of the most common possible degradation paths of soy isoflavones are presented in Fig. 3. However, only there are only a few studies about isoflavone stability during storage of soybeans, soy foods and extracts. In fact, only recently the stability of isoflavones in soybeans stored under different conditions was investigated [74–76]. Information from the few reports available indicates that storage of soybeans and soy foods for prolonged times at room temperature can affect isoflavone distribution and content. Generally, the concentrations of individual isoflavones either significantly decrease or increase during storage for long periods. With storage, malonyl glucosides concentration tends to decrease while concentration of glucosides and aglycones tend to increase. Concentration of malonyl glucosides can decrease by about 2 times, whereas glucosides and aglycone concentration can increase up to 3–4 times during storage for 2 years [74]. However, storage at room temperature may, in some cases, decrease glucoside and aglycone content [75]. Moreover, not only the isoflavone profile may be affected by the course of time, but also total isoflavone concentration, especially in the first year of storage. Afterwards, storage (up to 2 years) only slightly changes total isoflavone content but still affects isoflavone profile of the samples [74]. Storage for prolonged periods reduces total isoflavone concentration and the reduction level depends of the soybean cultivar. While some cultivars show only a slight decrease on total isoflavone concentration, others present a severe decrease on concentration of these compounds [75]. Furthermore, the level and type of the modification on isoflavone profile and losses caused by storage may be dependent of temperature, relative humidity and the soybean cultivar among other factors. The variation on isoflavone concentrations were positively correlated with storage temperature and total isoflavones were related with the amount of malonyl glucoside and glucoside groups. Storage at low temperature can result in changes in isoflavone levels similar to those observed during processing [75] or may not affect isoflavone distribution [76]. Endogenous glucosidases, humidity

and influence of soybean variety as observed by Kim et al. [75] may, partially, explain differences observed in these studies. Relative humidity as well as temperature can influence the changes on isoflavone profile during storage. Storage of soybeans under high relative humidity (84%) and temperature (30 ◦ C) conditions for extended periods of time (9 months) causes the interconversion between aglycones and ␤-glucosides. Storage under these conditions can affect isoflavone profile to a point that the major constituents can become the minor constituents, and vice-versa. Storage under milder storage conditions (57% relative humidity and 20 ◦ C) causes only the interconversion between ␤glucosides and malonyl glucosides [76]. It has also been demonstrated that some isoflavones in soymilk are subjected to degradation [77] during storage. For example, Gi is labile to degradation during storage at room temperature, although at a low rate. Losses of Gi with time showed typical first-order kinetics and increased with storage temperature. The Di concentration was not influenced by storage between 15 ◦ C and 37 ◦ C. However, degradation of Di was not discarded, since it was possible that a combination of deacetylation of ADi to Di and Di degradation was taking place simultaneously. At early stages of soymilk storage at 80–90 ◦ C, ADi concentration increased, followed by a slow decrease. However, malonyl isoflavones, which are more sensitive to degradation, were not studied. Therefore, more research it is still needed on the effects of storage environments, such as humidity and temperature, on the transformation and losses of isoflavone groups. The characterization of the differences between soybeans cultivars related with the change of isoflavones, with special emphasis on endogenous glucosidases and to identify suitable conservation methods are important pending tasks. Also, more research aimed at different soy products is required in the same direction. Finally, it is imperative that authors conducting quantification studies be specific about sample conservation aspects. It must be clear for how long the soybean or soy food sample have been stored before actual analysis and the conditions such as temperature, humidity, etc.

6

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

Also, while studying isoflavone profiles and distribution in foods and over different cropping years it is highly recommendable to perform the determinations after harvest, and not analyze all the samples at the same time (after all samples were harvested), with the inherent differences and errors caused by storage after 1 year or more, even if samples are stored at low temperature. It is also recommendable to refer isoflavone content to dry weight since variation of sample humidity may influence concentration. On the other hand, storage of samples after extraction and before analysis can also affect isoflavone profiles and result in avoidable analytical errors. Due to the relatively unstable character some isoflavones as well as by the action of native ␤-glucosidases, resulting in a rapid degradation or interconversion between chemical forms, quantification of isoflavones is a complicated procedure. The most susceptible to degradation isoflavones seems to be the malonyl forms. Barnes et al. [78] noted that isoflavones in 80% MeOH extracts of soy samples kept at room temperature were converted gradually from malonyl glucosides to ␤-glucosides. Coward et al. [71] reported a slight conversion of the malonyl glucosides to the ␤-glucosides conjugates at room temperature and that malonyl glucoside conjugates are stable at 4 ◦ C for 24 h, but prolonged storage also causes conversion to the ␤-glucosides conjugates. Later, Murphy et al. [42] reported a conversion rate of 0.2–0.3 mol% per hour of malonyl forms to glucosides in soy isoflavone extracts at room temperature. Evidences show that prompt analysis of the extracts after extraction or other strategies, such as maintenance of auto sampler at low temperatures (4–5 ◦ C) are necessary to minimize degradation of malonyl isoflavones. Although these procedures can elude potential analysis errors it is essential to consider the stability of soy isoflavones extracts under storage conditions to allow better planning of routine analysis of large number of samples and avoid analytical errors due to degradation and conversion between forms (i.e. malonyl to glucosides, malonyl to acetyl glucosides, etc.) [79]. In one of the few published reports dealing specifically with the storage of soy isoflavone extracts, Rostagno et al. [79] evaluated the influence of several factors (temperature, storage time, head space and UV light) on short-term stability of samples kept on HPLC vials. The conclusion was that samples can be stored up to 1 week with no significant degradation if kept at temperatures lower than 10 ◦ C and protected from light. On the other hand, Rijke et al. [80] evaluated the stability of isoflavone extracts obtained from red clover and observed that samples can be stored up to 2 weeks at −20 ◦ C and if samples are kept at room temperature or if are stored dry at −20 ◦ C, degradation starts almost immediately. Curiously, they also observed that in LC separated fractions, red clover malonyl isoflavones are more stable when stored at low temperature after evaporation to dryness. Aside the fact that the report did not include most common isoflavones present in soybeans it indicates that more research is needed to find more suitable sample conservation methods and to evaluate longer storage of soy isoflavone extracts under different conditions before analysis. 4. Hydrolysis As previously discussed, there are different isoflavone chemical structures, and interconversion can occur between forms depending of storage, processing and extraction conditions. Not only sample preparation is complicated, but also the analysis step may be critical. The accurate quantification of the total content of isoflavones is hampered by the feasibility of chromatographically separating all the possible forms of these compounds and to find the corresponding reference standards. Some isoflavones are par-

ticularly difficult to separate from each other (i.e. MGi, AGly and De) [81], while others (i.e. malonyl and acetyl isoflavones), due their relative unstable character, are not widely commercially available. Coelution of other substances present in the extracts may also add difficulty to the troublesome determination of soy isoflavones. Furthermore, some isoflavones might occur in as yet unidentified forms. A possible solution to these analytical problems is to perform adequate sample treatment involving hydrolysis in order to reduce the number of isoflavone chemical forms occurring in the sample. The hydrolysis procedure itself can be carried out before, during or after the extraction using different conditions and agents. There are two main procedures to perform the hydrolysis of isoflavones reported in the literature, basic or acidic hydrolysis. Basic hydrolysis act on ester bonds, removing acid groups that are linked to the sugar moiety of the isoflavone glucosides. As a result, the malonyl and acetyl glucoside isoflavone forms are converted to their respective glucosides. Acid hydrolysis breaks the bond between the isoflavone and the glucoside moieties, transforming all the isoflavone derivatives, into their aglycone forms [82]. Although reaction times and temperatures for the acidic hydrolysis conditions vary a great deal, these procedures usually involve treating the extract or food sample itself with inorganic acid (HCl) at high temperatures in aqueous or alcoholic solvents with reaction times ranging from a few minutes to several hours. Basic hydrolysis entails treating the sample with a solution of NaOH and allowing standing at room temperature from a few minutes to overnight [82–87]. Hydrolysis through the use of enzymes or a combination of enzyme and acid [88,89] has also been used, although these methods are less frequently used than acid or basic hydrolysis. The enzymatic hydrolysis consists of incubating the sample with enzymes for long periods of time, ranging from a few hours to overnight. Different enzymes have been used for the hydrolysis of isoflavones, including endougenous soy ␤-glucosidases, ␤-glucuronidases, sulfatases and cellulases. Similarly to basic and acid hydrolysis, conditions vary a great deal and several different methods have been reported [88–93]. There are advantages and disadvantages with the use of hydrolytic methods. The most obvious disadvantage is the inclusion of an additional step, with the inherent complication of the sample preparation procedure and the possible added analytical variability. Also, there is indication that Ge is not entirely stable under acid hydrolysis conditions [93]. The limited information obtained using hydrolytic methods can also be decisive, since only a few chemical forms are quantified, while using non-hydrolytic methods full information can be accessed. Although the hydrolysis step creates new questions with respect with sample preparation, analyte stability and recoverability, it greatly simplifies the analysis by reducing the number of derivatives. The chromatographic analysis time is considerably shorter and separation of target compounds is easier since there are fewer compounds occurring in the sample. Acid hydrolysis results in the inclusion in the quantification of isoflavones that are linked to sugars other than glucose, and of glucosides of isoflavones that are not commercially available or difficult to acquire. Acid hydrolysis is useful for the analysis of complex samples, and may be used to identify sugar-isoflavones by comparison of these results with those from basic hydrolysis. The analysis of acid hydrolyzed extracts is preferred when analyzing samples of unknown origin, because it includes in the quantification the glucoside derivatives of all isoflavones available only as aglycones [82]. Moreover, the use of hydrolytic methods may reduce the analytical variability caused by stability issues during extraction since the most unstable isoflavones (malonyl glucosides) are not quantified

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

as such. However, it is of crucial importance when using hydrolytic methods that authors make the necessary corrections and normalize the results by molecular weight to the aglycone forms, since the molecular weight of the glucosides is greater than aglycones, and therefore reported total isoflavone amount can be significantly less than the value of non-normalized data. Although the available evidence in the literature suggests that the biological effects of soy isoflavones depend upon aglycone form, for analysis of soy foods for isoflavonoids, the recent trend has been to avoid hydrolysis. Using non-hydrolytic methods provide valuable information about the exact distribution of all chemical forms present in soybeans and soy foods. The different isoflavones may have differing pharmacokinetics and bioactivity [94–96] and this may be a key factor in understanding their biological effects. Another logical reason to avoid hydrolysis is to minimize sample handling, simplifying the extraction and overall analytical procedure and to shorten, as much as possible, the time required from sampling to actual analysis. However, it is important to highlight that the valuable information about the total isoflavone concentration provided by hydrolytic methods is an essential screening measurement and that isoflavone profiles are very important in an advanced metrological step. 5. Extraction techniques and methods 5.1. Solid and semi-solid samples Optimal solid–liquid extraction involves the intimate contact between a solid material, usually finely grinded, and a solvent that has a maximal solubility for the analyte of interest and a minimal solubility for the matrix, using additional external forces and heating to speed up the extraction process. Solid soy samples, such as soybeans and soy protein, require only grinding before extraction, but sometimes are freeze-dried to provide a homogenous powder. Liquid samples are most often freeze-dried and also treated as solid samples. Common methods for the extraction of the isoflavones from solid samples include organic solvent extraction with pure or aqueous methanol (MeOH), ethanol (EtOH), acetonitrile (MeCN) or acetone (ACE) with and without the addition of small amounts of acids using simple soaking, mixing, shaking or soxhlet extraction. The extraction time may range from 2 h to 24 h and the extraction temperature from 4 ◦ C to 80 ◦ C. More recently, “modern” extraction methods, such as ultrasonically assisted extraction (UAE), pressurized liquid extraction (PLE), supercritical fluid extraction (SFE) and microwave-assisted extraction (MAE) have been used for the extraction of soy isoflavones using similar solvents. In many cases, besides of filtration and centrifugation, further purification and/or pre-concentration of the target compound fraction is applied. In these cases, evaporation to dryness and re-dissolution on another solvent or solid phase extraction (SPE) are the most commonly used methods. Another common procedure during sample preparation is the hydrolysis after the extraction (see Section 4). 5.1.1. Conventional extraction methods Among the conventional extraction techniques soxhlet, shaking, and stirring are the most commonly used for the extraction of isoflavones from soybeans and soy foods. There are numerous available extraction methods using these techniques with different conditions, and most of them without an appropriate method optimization. Several parameters can influence the extraction of organic compounds such as polarity and amount of the solvent, temperature, mass and kind of sample and extraction duration. In the specific case of isoflavones, optimum solubility of the analyte in the

7

extraction solvent, one of the key parameters of any extraction method, is very difficult to achieve since there are several chemical forms with different solubility coefficients in a given solvent. Most methods so far developed evaluate different solvents trying to reach an optimal condition where extraction of all isoflavones is maximized. Although abundant research on soy isoflavones quantification is available only a few reports deals with the development and optimization of extraction methods for quantification studies. An overview of developed methods and evaluated parameters using conventional techniques for the extraction of isoflavones from soybeans and soy foods is presented in Table 1. One of the first studies about the extraction of soy isoflavone was published by Eldridge [97], where pure MeOH and EtOH and with different water proportions, Ethyl acetate and MeCN were evaluated for refluxing extraction of isoflavones from defatted soybeans. From the evaluated solvent systems, 80% MeOH gave the highest isoflavone extraction yields and the most reproducible results. Using this solvent, 4 h seemed to be sufficient for extracting the isoflavones from soybean meal and no significant differences in the extraction efficiency was reported when using different solvent:sample ratios (14:1 and 45:1). Once extraction conditions were established, the method was used for the determination of isoflavones from soybean flours, protein concentrates and isolates. The same method was also used for the study of the effect of environment and variety on the composition of soybean isoflavones [98]. Another pioneer study about the extraction of isoflavones was carried out by Murphy [99], who compared several solvents systems (MeOH, ACE, MeCN, and chloroform-MeOH) for the extraction of isoflavones from toasted defatted soy flakes using wrist-action shaker. The results indicated that extraction with pure solvents gave low yields and that the addition of water or acid greatly improve the extraction efficiency of all isoflavones examined (Gi, Ge, Di and De). In terms of total isoflavones and coextractives, MeCN with water or acid was more efficient for the extraction of isoflavones all other solvents systems examined and no marked difference between these two solvents was observed in terms of total isoflavones. As a result of these two pioneer studies, 80% methanol and acidified 83% acetonitrile became the most commonly used extraction solvents in isoflavone analysis. The method developed by Murphy [99] has been extensively used with slight modifications in sample amount, solvent volume, addition of water to the extracting solvent and shaking technique [17–19,41,42,72,74,100–103]. However, these slight modifications of the method have an important impact on extraction efficiency and should not be used lightly, since extraction conditions require optimization for each different sample. As an example, Song et al. [101] reevaluated the method by Murphy [99] and reported that using water in addition to HCl and MeCN increased recovery. For different soy samples different amounts of water may be necessary maximize isoflavone extraction. For most soy foods, 7 mL of water was sufficient to maximize extraction using a solvent volume:sample ratio higher than 6 mL g−1 . It was also recommended that the solvent volume:sample ratio should be adjusted for soy products with high concentration of isoflavones, particularly for isoflavone supplements, which have more than 10 mg isoflavones/g. These investigators gave the example of soy germ, which have high isoflavone content (>10 mg g−1 ), and reported that the normal extraction procedure would underestimate the isoflavone content by 10–20%. They found that adjusting the ratio of solvent to sample weight to 95 mL g−1 resulted in more efficient extraction of isoflavones from the soy germ sample. Following the evidence of the effect of the amount of water of the extraction solvent on isoflavone extractability, Murphy et al. [42], reevaluated the same method and confirmed that adding a

8

Table 1 Developed methods and evaluated parameters using conventional techniques for the extraction of isoflavones from soybeans and soy foods. Sample used for evaluation of the method

Defatted soybeans

Isoflavones

Di, Gi, Gly, De, Ge and Gle

Fixed extraction conditions Technique: refluxing Sample: 1 g Solvent: 25 mL Temperature: boiling point of solvent

Evaluated parameters Solvent: EtOH, 50% EtOH, 80% EtOH MeOH, 50% MeOH, 80% MeOH CH3 CN Ethyl acetate

Selected conditions

Reference

80% MeOH, 4 h

[78]

80% CH3 CN and 80% CH3 CN (HCl)

[99]

The amount of water optimized depending of the sample ranged from 5 mL to 10 mL of water

[42]

Extraction time: 1–5 h Technique: Wrist-action shaker Sample: 5 g Toasted defatted soy flakes

Gi, Ge, Di and De

Solvent: 25 mL of pure solvent or: 5 mL (H2 O or HCl 0.1N) + 20 mL (solvent) Temperature: RT

Soy isolate, tofu, soybeans and miso

Ge, De, Gi, Di, Gly, MGi, MDi, MGly and AGi

Technique: Stirring Sample: 2 g, Solvent: 12–22 mL (12 mL CH3 CN + 2 mL HCl 0.1N + water)

Solvent: Different amounts of water (0–10 mL) added to the solvent (CH3 CN)

Extraction time: 2 h Temperature: RT

Toasted soy flour

Soy protein

Di, Gi, Gly, De, Ge, Gle, MDi, ADi, MGi, AGi and MGly

Di, Gi, Gly, De, Ge, Gle, MDi, MGi, MGly, ADi, AGi and AGly

Technique: tumbling mixer Sample: 0.5 g Solvent: 4 mL

Technique: rotary mixer Sample: 1 g or amount containing 10 mg total isoflavones (always less than 1 g) Solvent: ∼17 mL Extraction time: 2 h Temperature: RT Technique: Stirring Sample: 2 g

Soy flour, tempeh, TVP and soy germ

Di, Gi, Gly, De, Ge, Gle, MDi, MGi, MGly, ADi, AGi and AGly

Solvent: 19 mL (10 mL solvent + 2 mL (HCl 0.1N or water) + 7 mL water Extraction time: 2 h Temperature: RT Technique: Stirring Sample: 2 g,

Solvent: 80% MeOH and 80% CH3 CN (0.1% HCl) Extraction time: 1, 2 and 24 h Temperature: RT, 60 ◦ C and 80 ◦ C Solvent: 10 mL CH3 CN + 6 mL H2 O + 0.5 mL DMSO (IS)

10 mL CH3 CN + 2 mL HCl 0.1 M + 5 mL H2 O 80% MeOH Water % (10–100% CH3 CN) Solvent: 53% CH3 CN, 53% ACE, 53% EtOH, 53% MeOH With and without acid addition

Solvent: 83% CH3 CN, 83% CH3 CN (+0.1N HCl)

1 h, RT

[73]

10 mL CH3 CN + 6 mL H2 O + 0.5 mL DMSO (IS)

[104]

53% CH3 CN without acidification

[102]

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

Extraction time: 2 h

Solvent: MeOH, 80% MeOH, 80% MeOH (HCl) Chroloform–MeOH (90:10), 80% chroloform–MeOH (90:10), 80% chroloform–MeOH (90:10) (HCl) CH3 CN, 80% CH3 CN, 80% CH3 CN (HCl) ACE, 80% ACE and 80% ACE (HCl)

Soybeans

Di, Gi, Gly, De, Ge, Gle, MDi, MGi, MGly, ADi, AGi and AGly

Solvent: 12 mL Extraction time: 2 h

58% CH3 CN, 58% CH3 CN (+0.1N HCl) 80% MeOH, 80% MeOH (+0.1N HCl)

58% CH3 CN without acidification

[105]

50% EtOH, 60 ◦ C

[106]

80% CH3 CN–HCl 0.1N, 5 sequential extractions

[107]

Proposed method

[103]

99.99% EtOH, 3:1 mL g−1 , 80 ◦ C and 8h

[20]

Temperature: RT

Freeze-dried soybeans

Defatted soybean meal, soy protein isolate

Di, Gi, Gly and MGi

Di, Gi, Gly, De, Ge, Gle, MDia , MGia and MGlya

Di, Gi, Gly, MDi, MGly, MGi, De and Ge

Soybean flour

Ge and De

Solvent: CH3 CN (30–70%) EtOH (30–70%) MeOH (30–70%) Temperature: 10 and 60 ◦ C

Technique: Shaking Sample: 2 g Solvent: 10 mL

Solvent: 80% CH3 CN–HCl 0.1N 80% MeOH

Extraction time: 2 h Temperature: RT

80% EtOH Number of extractions: 1 and 5

Technique: homogenization probe and hand agitation Sample: 0.1 g Solvent: 4 mL (80% MeOH) (homogenization) + 1 mL (agitation) Extraction time: 1 min (homogenization) + 30 min (agitation) Temperature: RT (homogenization) and 70 ◦ C (agitation) Technique: stirring Solvent: 4 mL (80% MeOH) (Homogenization) + 1 mL (agitation) Extraction time: 1 min (homogenization) + 30 min (agitation) Temperature: RT (Homogenization) and 70 ◦ C (agitation)

Proposed method and reference method (modified Murphy method)

Solvent: 40–99.99% EtOH Volume:sample ratio: 1:1 to 10:1 (mL g−1 ) Temperature: 40–90 ◦ C

Extraction time: 2–24 h

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

Soybean flour

Technique: Stirring Sample: 0.5 g Solvent: 25 mL Extraction time: 10 min

De: daidzein, Ge: genistein, Gle: glycitein, Di: daidzin, Gi: genistin, Gly: glycitin, MDi: malonyl daidzin, MGi: malonyl genistin, MGly: malonyl glycitin, ADi: acetyl daidzin, AGi: acetyl genistin, AGly: acetyl glycitin, MeOH: methanol, EtOH: ethanol, CH3 CN: acetonitrile, RT: room temperature. a Tentatively identified by literature.

9

10

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

certain amount of water could optimize the total extraction. Extraction conditions were optimized for each soy sample. The amount of water had a significant effect of the amount of isoflavone extracted and varied with the food extracted. The amount of water optimized, depending of the food matrix, ranged from 5 mL to 10 mL (isolate, 10 mL; tofu, 10 mL, soybeans, 7 mL, miso, 5 mL) using 2 g samples. Also, the question of which extraction solvent is more efficient is difficult to answer since it will depend of several factors such as the technique, sample, amount of water, time, sample:solvent ratios, temperature, etc. For example, while Murphy [99] observed that 80% MeCN (with and without acid) was more efficient than 80% MeOH (with and without acid), Barnes et al. [78] did not observe significant differences between 80% MeOH and acidified 80% MeCN for the extraction of isoflavones from toasted soy flour toasted soy flour using a different solvent to sample ratio. Later, Griffith and Collison [104] proposed an improved procedure for the extraction of isoflavones from different soy samples using 60% MeCN with 3% dimethylsulfoxide (DMSO) (v/v) and compared this solvent with 80% MeOH. This procedure was also compared with the modified Murphy [99] method used by Song et al. [101]. 80% MeOH was less efficient than MeCN (with and without acidification + DMSO) in extracting most isoflavones and differences between MeCN solvents (with and without acidification + DMSO) were smaller, with the primary difference in the extraction efficiency of more hydrophobic isoflavones (AGi, Ge, De and Gle). Afterwards, different water proportions of the extraction solvent were tested and 60% MeCN proved to be the most efficient solvent for two different soy protein samples (high and low in malonyl isoflavones). It was also observed an improvement (0.7–10.6%) in the extraction efficiency of different isoflavones from soy samples extracted with DMSO. The authors suggested that marginal increase in isoflavone content might be attributed to the lack of acid or to the presence of small quantity of DMSO. It is clear that more research is still needed to evaluate the influence of DMSO on extraction efficiency of isoflavones and examine the observations reported in this study. Another interesting result was the small effect of extraction time and the observation that the vast majority of isoflavones were extracted in the first 5 min of extraction. This is strong evidence that the extraction time of similar methods can be drastically reduced from 2 h and this parameter can be further optimized. Following the matter about the choice of the extraction solvent, Murphy et al. [102], reviewed the extraction method and further investigated MeCN, EtOH, ACE and MeOH in a 53% aqueous solution with and without acid addition using the same method and concluded that MeCN was more efficient than the other solvents and that MeOH was the least efficient solvent in extracting the 12 main isoflavone forms in raw soy flour, tofu, tempeh, texturized vegetable protein and soy germ. They also observed that the different solvents have different abilities to extract the different isoflavone forms and that the food matrix configuration may have an important impact on the extractability of the isoflavone forms. Depending of the sample, some solvents may underestimate individual isoflavone content up to 35% and total isoflavones up to 20%. Another important remark was that addition of acid reduced the extracted amount of some isoflavones and increased the extraction of others depending of the sample matrix. The authors suggested that in order to simplify the extraction protocol, it is probably better not to use acid in the extraction medium for these food matrices. In fact, the addition of small quantities of acid to the extracting solvent used by Murphy et al. [41,42,99,100] have been questioned since no clear differences or systematic pattern for all foods or for all isoflavone forms have been demonstrated and as evidenced by Griffith and Collison [104].

The initial purpose for the addition of small amount of acid was to increase the extraction efficiency and minimize coextractives and give clean HPLC chromatograms. However, in the initial report [99], non-acidified MeCN extracted lower amounts of coextractives with similar efficiency than acidified MeCN. Therefore, the use of acidified MeCN seems not to make sense. Further evidence is provided by Lin and Giusti [105], who evaluated the effects of solvent polarity and acidity on the extraction efficiency of isoflavones from soybeans. In this report, acidified solvents either extracted significantly (p < 0.05) lower amounts of isoflavones or did not significantly differ from solvents without acid. Non-acidified solvents were more efficient in extracting malonyl isoflavones. For glucosides isoflavones, the acidification showed a less significant effect on Gi and Gly and no relevant effect on Di. Also, no remarkable effect of acidification was found in the extraction of AGi and aglycones (Ge and De). The differences in the total isoflavones obtained between acidified and non-acidified solvents mainly reflected the differences in malonyl isoflavones. This may, in part, explain the results obtained in the first report of Murphy [99] regarding the use of acidified solvents, since malonyl isoflavones were not measured in this study. Moreover, a significant polarity–acidity interaction was found for aglycone extraction, which suggests that the effect of the acid was not the same in the solvents with different polarities. Another important observation was that acidification of the extraction solvent favored isoflavone transformations during the extraction and therefore should be avoided for quantification of intact isoflavones [105]. Regarding the extraction efficiency of the solvents, results indicated that for all glucoside isoflavones the solvent with higher polarity (58% MeCN) either extracted significantly higher amounts or did not significantly differ from the assayed solvents with lower polarity (80% MeOH and 83% MeCN). The differences between 58% MeCN (most polar) and 83% MeCN (least polar) were important in terms of extraction efficiency of individual and total isoflavones. However, differences between 58% MeCN and 80% MeOH or between 80% MeOH and 83% MeCN were not always relevant. On average, 58% MeCN extracted significantly higher amounts of malonyl glucosides than 80% MeOH and 83% MeCN. Recoveries of aglycones, Ge and De with 80% MeOH resulted significantly lower than those obtained with the other evaluated solvents. The differences in measured isoflavones between solvents with various polarities reflected the differences in malonyl glucosides, because malonyl glucosides was the major form of isoflavones in the soybeans and it was most affected by solvent polarity. Therefore, solvents with relatively higher polarity and no acid were more efficient in general for extracting isoflavones. Among the six examined solvents, 58% MeCN without acidification was the best solvent for the extraction of isoflavones from soybeans, since it yielded the highest total amounts and best maintained the intact structures. With regard to the two most widely used solvent systems, 80% MeOH had a higher extraction efficiency and better protection against chemical transformation than acidified 83% MeCN. These results are in agreement with those reported by Rostagno et al. [106] who compared different solvents for the extraction of isoflavones glucosides and MGi from soybeans. These authors observed that when using pure solvents, low extraction efficiency was obtained and that the maximum amount extracted was obtained using solvents with 40–60% of water. They also observed that temperature has a great impact on the extraction efficiency of isoflavones. Rostagno et al. [106] also reported that most isoflavones present in the sample (80–90%) were extracted in the first 10 min of extraction at 60 ◦ C using 50% EtOH, corroborating similar observations reported by Griffith and Collison [104].

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

Another approach of solvent selection was given by Achouri et al. [107]. These authors compared three solvents (80% MeCN–HCl 0.1N, 80% MeOH and 80% EtOH) for the extraction of isoflavones from defatted soybean meal (DSM) and from soy protein isolate (SPI). In the case of the DSM, the conclusion was that acidified 80% MeCN is more efficient for the extraction of malonyl isoflavones and aglycones, while 80% MeOH is more efficient for the extraction of glucosides (using one extraction). In the case of the SPI, 80% EtOH extracted the highest amount of aglycones, no significant difference was observed between 80% EtOH and 80% MeOH for the extraction of glucosides and that acidified 80% MeCN extracted the highest amount of malonyl glucosides (using one extraction). However, 80% MeOH extracted the highest amount of total isoflavones, followed by 80% EtOH and by acidified 80% MeCN in this particular sample. These results indicate that the extraction efficiency of the solvent will depend of the sample from which the isoflavones are extracted. One of the most interesting remarks in this report was that individual amount of isoflavones extracted after the first extraction increased significantly after 5 consecutive extractions (42–100% depending of the isoflavone) in soy meal, and in soy protein isolate (89–153% depending of the isoflavone). For the different solvents used, the yield of total isoflavones after 5 extractions (compared to only one extraction) increased between 65% and 74% for the DSM sample, and increased by between 107% and 147% for ISP sample, depending of the solvent. The most important observation in this report was that no significant difference in terms of total isoflavones was observed between the assayed solvent after 5 sequential extractions in the DSM sample indicating that it is possible to achieve quantitative extraction with any of the most commonly solvents used for isoflavones extraction, given that conditions are optimized enough. Moreover, these results strongly suggest that one time extraction of isoflavone using conventional methods markedly under-estimates the concentration of isoflavones in these products. However, sample characteristics are likely to play an important role in the ability of a given solvent to extract isoflavone from soybeans and soy foods. It is very interesting the observed variation in the extraction yield of isoflavones between DSM and SPI. For the high protein sample (SPI), a unique extraction extracted only 41% of total isoflavone compared to 58% of lower protein sample (DSM) using MeCN–HCl solvent. This difference was attributed to stronger protein–polyphenol interaction in the SPI sample since a variety of interactions including hydrogen bonding, ionic and covalent binding, and mainly hydrophobic interactions are involved in the formation of protein–polyphenol complex [108]. These interactions are strongly influenced by factors such as temperature, pH and salt, which occur during acidic precipitation of soy proteins. This outcome may also indicate that grinding and the resulting particle size might, due to the effect in the matrix, can influence the ability of the different solvents to extract isoflavones. The same principle can be extended to freeze-drying, which more severely affect sample matrix structures. Another extraction method using 80% MeOH for the analysis of isoflavones from soybean flour was later proposed by Tsai et al. [103]. The proposed method was compared with a modified Murphy method (using different sample to solvent ratio). They observed that, except De and Ge, contents of detected isoflavones (Gi, Di, Gly, MGi, MDi and MGly) extracted by the proposed method were higher than those extracted by the modified Murphy method. These findings imply that that several reports of isoflavone distribution in foods using the method by Murphy are underestimating isoflavone concentration. On the other hand, Zhang et al. [20] evaluated several extraction conditions for the extraction and purification of isoflavones from soybeans. Extraction conditions included EtOH percentage

11

(40–99.99%), solvent volume to sample ratio (1:1 to 10:1 mL g−1 ), temperature (40–90 ◦ C) and extraction time (2–24 h). In this report, the influence of some extraction parameters was different than those obtained by other authors. Pure EtOH extracted the highest amount of isoflavones, while in most studies is clear that a certain amount of water (40–60%) in the solvent is necessary to improve extraction. Also, increasing solvent volume to sample ratio from 3:1 to 8:1 (mL g−1 ) negatively affected yield. The objective of this study however, is the key to understand the differences on the effect of the sample amount observed by other authors. In this case, the objective was to extract the highest amount of aglycones and to obtain a concentrated extract, not to quantify all chemical forms. Using a higher amount of sample with lower amounts of solvent volume, it is logical that the concentration of isoflavone on the extract tends to increase although lower relative extraction efficiency is achieved. Summing up, the differences among the extraction methods reported in this review are most probably related with the amount of water used in the extraction solvent, the sample matrix, the extraction technique, sample to solvent ratio and more importantly, the isoflavone forms that were quantified. In some cases comparison of extraction solvents were carried out for only a few isoflavones present. In this context, it is important to note that some chemical forms are responsible for the greater part of the total amount of isoflavones present in soybeans and soy foods, especially some malonyl and glucoside forms. Moreover, differences in analytical methods and reporting of isomeric conversions can also contribute significantly to variation on the results found in the literature. In some studies, total isoflavone is expressed as the sum of all 12 isomers. In other studies, only aglycone and/or conjugated forms are tested and expressed. Furthermore, in other studies isoflavones are hydrolyzed to their aglycone forms or the amount is normalized by molecular weight to the aglycone forms. Also, some methods were developed before the malonyl glucoside isoflavones were identified [38,39] and therefore the results needed to be revised for the extraction of all 12 isoflavone forms. More importantly, the effect of extraction conditions on stability was not considered in many cases. In general terms, the choice of the most appropriate solvent will depend of the isoflavone in highest amount present in the sample, since the most effective solvent for this particular isoflavone will strongly influence the total amount extracted. For comparison purposes it is important to evaluate different solvents without achieving quantitative recoveries otherwise it will be impossible to determine the magnitude of effect of the solvents. However, the recent trend is to avoid toxic and use environmental friendly solvents such as EtOH. EtOH can be highly effective for the extraction of isoflavone from soy samples with the advantage of lower cost, lower toxicity and environmental compatibility. It also appears clear that to obtain quantitative extraction for the analysis of the isoflavone content of foods is necessary to adjust extraction conditions for each sample and some research is still needed to optimize other extraction variables, especially sample to solvent ratio and extraction time. One of the important conclusions when reviewing information available is that it is possible to achieve quantitative extraction using most commonly used solvents and that it very likely that sequential extractions are required, as previously mentioned. Finally, more research is needed to evaluate and explain the influence of the sample, since it may be the answer to achieve standard methods for the extraction and analysis of isoflavones in foods. 5.1.1.1. Stability during extraction using conventional techniques. Apart of optimizing extraction variables such as solvent, sample amount, temperature and duration, the assessment of stability

12

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

during extraction is essential. Frequently, authors tend to overextend extraction duration in order to achieve higher extraction yields. This strategy, however, may cause not only degradation of some chemical forms, but also the generation of other isoflavones forms and isomers that can drastically modify isoflavone profile of the sample and influence the results obtained. Thus, although long extraction times have been extensively used for the extraction of isoflavone from soy and other matrixes, there are still several issues that should be addressed such as the stability during extraction. Extraction of isoflavones from foods or dietary supplements is a critical process since isoflavone profile can be altered during sample preparation since mild heat and acid are frequently involved in the extraction, which could cause degradation of malonyl isoflavones and the hydrolysis of glucosides. Therefore, when choosing the extraction conditions it is important not only consider extraction efficiency, but also avoid, or at least minimize, the artificial transformations. Thus, temperature conditions during the extraction procedures as well as extraction duration have to be carefully adjusted because of possible degradation of the glucoside derivatives. Also, stability may be related to the solvent used, specially acidified solvents. One of the earlier observations of the influence of the extraction temperature on the isoflavone profile was reported by Kudou et al. [39]. They observed that malonyl isoflavone glucosides in 70% alcohol extracts from both soybean hypocotyls and cotyledons decreased significantly as their respective glucosides increased when the samples were extracted at 80 ◦ C instead of room temperature. The effects of extraction temperature on isoflavone profile were later confirmed by Barnes et al. [78]. They observed that extractions performed at 60 ◦ C caused heat induced de-esterifying reaction of malonyl and acetyl glucosides to their respective glucosides and that increasing temperature to 80 ◦ C led to higher conversion rate. Moreover, the changes on isoflavones profile were not only due to temperature variations, but also time dependent. Even at room temperature malonyl glucosides were gradually converted to their respective glucosides. The conversion rate at room temperature was later reported to be between 0.2 mol% and 0.3 mol% per hour [42]. Obviously, extraction methods using long extraction times can significantly underestimate malonyl glucoside concentration and overestimate glucoside concentration. Coward et al. [71] evaluated the effect of the temperature on the extraction of isoflavones from soy foods. Isoflavone ␤-glucosides conjugates were extracted with 80% MeOH from soybeans at room temperature, at 4 ◦ C and at 80 ◦ C, for 2–72 h by tumbling or shaking. Quantitative and reproducible recovery of the isoflavone glucosides was achieved after 2 h. Extraction at 4 ◦ C gave the highest concentration of malonyl glucosides and the lowest concentration of ␤-glucosides conjugates. Extraction at 80 ◦ C caused extensive conversion of the malonyl glucosides conjugates to the ␤-glucoside conjugates but not to the acetyl conjugates or aglycones. Although the composition of the individual ␤-glucosides was drastically altered by temperature, the total amount of isoflavones extracted was constant. On another study, Franke et al. [93] evaluated the stability of De, Ge, coumestrol, formononetin, biochanin A and flavone under refluxing for 4 h using acidified 77% EtOH (2.0 M HCl) and observed that only flavone was entirely stable. Therefore, it is clear that refluxing is not recommendable for extraction of isoflavones from soybeans and soy foods, since it can cause losses, even if hydrolytic methods are used. However, this may be related to the use of acidified solvent as reported by Lin and Giusti [105], who later observed the transformation of ␤-glucosides to their corresponding aglycones and transformation of acetylglucosides to their corresponding ␤-glucosides when subjected to extraction by

stirring for 2 h at room temperature. Another important remark was that acidification of the extraction solvent favored isoflavone transformations during the extraction and should be avoided for quantification of intact isoflavones. Therefore, when evaluating an extraction method it is of crucial importance to know the stability of target compounds, in order to maintain the isoflavone profile in the sample, unless hydrolytic methods are used (see Section 4). Submitting an extract obtained with optimal conditions to the extraction protocol and comparing concentrations is a simple way to perform such stability tests [109,110]. It cannot guarantee that target compounds are entirely stable since other sample matrix components may influence stability, but may give clues about the possible degradation under extraction conditions. The use of extracts is preferable to the use of standards since extracts contain other components and are more close to real samples. Another method is to control concentration of malonyl isoflavones trying to identify degradation patterns or use hydrolytic methods, quantifying aglycone equivalents. The use of an internal standard may also prove useful in this case. The most recent trend regarding stability during extraction is to use of ␤-glucosidase inhibitors. Toebes et al. [111] identified Tris as a suitable ␤-glucosidase inhibitor in red clover extracts, which was optimized at 350 mM in 80% EtOH at pH 7.2. Extractions using Tris yielded much higher amounts (13–24 times) of malonyl isoflavones as opposed to extractions without Tris. Although it was evaluated for the extraction of isoflavones from red clover, the same principle may be applicable for the extraction of soy isoflavones. In this case, however, concentration of Tris might need adjustment and further investigation, but unveils a strategy to avoid degradation and, therefore, increase the reliability of results obtained in the future. Other possible candidates for this role are HgCl2 , AgNO3 and d-glucono-ı-lactone, which have been reported to inhibit soybean ␤-glucosidase, being the later the most potent inhibitor [112]. Although important advances have been made regarding the stability of isoflavones during extraction using conventional techniques, it is clear that more studies are necessary, especially with the aim of avoiding degradation in order to provide reliable information about the concentration of these compounds in foods. Also, sample and solvent characteristics have to be further examined in detail as well as other factors such as temperature and extraction technique. 5.1.2. Modern extraction techniques and methods The development and application of “modern” samplepreparation techniques with significant advantages over conventional methods (e.g. reduction in extraction time, organic solvent consumption and in sample degradation, elimination of additional sample clean-up and concentration steps before chromatographic analysis, improvement in extraction efficiency, selectivity, and/or kinetics, ease of automation, etc.) for the extraction and determination of isoflavones from soybeans and derived foods is playing an important role in the overall effort of ensuring and providing high quality data for researches worldwide. With this in mind, newer extraction methods have been developed using modern extraction techniques, including supercritical fluid extraction, ultrasound-assisted extraction (UAE), pressurized liquid extraction (PLE), microwave-assisted extraction and solid phase extraction. When selecting the appropriate solvent for the extraction of isoflavones using conventional extraction techniques, solubility is one of the most important factors. However, the selection of an appropriate solvent using “modern” extraction techniques is much more complex, since it will depend of other factors besides of the solubility of target compounds, such as the ability of the solvent to absorb microwave energy (MAE), how it propagates ultrasonic

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

waves (UAE) and the physical–chemical changes in the solvent that take place under elevated temperature and pressure, which will also affect solubility of target compounds (PLE/SFE). 5.1.2.1. Ultrasound-assisted extraction. The enhancement of extraction efficiency of organic compounds by ultrasound is attributed to the phenomenon of cavitation produced in the solvent by the passage of an ultrasonic wave. Cavitation bubbles are produced and compressed during the application of ultrasounds. The increase in the pressure and temperature caused by the compression leads to the collapse of the bubble, resulting on a “shock wave” that passes through the solvent enhancing the mixing. Ultrasound also exerts a mechanical effect. When a bubble collapses near a solid surface it occurs asymmetrically and generates high-speed jets of solvent towards the cell walls, therefore increasing the solvent penetration into the cell and increasing the contact surface area between solid and liquid phase. This effect coupled with the enhanced mass transfer and significant disruption of cells, via cavitation bubble collapse, increases the release of intracellular product into the bulk medium. The use of higher temperatures in UAE can increase the efficiency of the extraction process due to the increase in the number of cavitation bubbles formed. Several extraction parameters, similar to conventional extraction methods, can influence the extraction of organic compounds using ultrasounds, such as polarity and amount of the solvent, the mass and kind of sample and extraction time among others. Also, parameters regarding the ultrasound source such as frequency and intensity as well as the number of pulses applied can have great impact on extraction dynamics [113–117]. Ultrasound-assisted extraction has been used in several occasions to extract isoflavones from soybeans, soy foods and from different matrixes, such as Peanuts, Trifolium pretense, Puerariae radix, Pueraria lobata, Radix astragali and Glycyrrhizae radix [80,82,106,118–125]. However, optimization of UAE based methods has not been conducted with a few exceptions. An overview of the developed methods using ultrasounds for the extraction of soy isoflavones and evaluated parameters is presented in Table 2. One of the first methods where extraction conditions were systematically assessed to achieve quantitative extractions of soy isoflavones was published by Rostagno et al. [106]. For the method development, several extraction parameters were studied including solvent, extraction temperature, sample amount and extraction time. The most important parameters affecting the extraction efficiency were the extraction solvent (and the amount of water), extraction temperature and extraction time. The extraction efficiency was improved by using ultrasounds but was dependent of the solvent employed. 50% EtOH, 50% MeOH and 40% MeCN were the solvent that extracted the highest amount of total isoflavones with similar extraction efficiency. The best extracting solvent for each isoflavone form depended of the chemical form itself. For all chemical forms best extraction efficiency was achieved using solvents with 40–60% of water. Extraction temperature had a great impact on the extraction efficiency while using higher temperature significantly increased the amount of all tested isoflavones. In general, the method was found to be fast and reliable achieving quantitative extractions in 20 min. To be sure that quantitative recovery was achieved; results were compared with 5 sequential extractions with no significant difference. Most isoflavones (80–90%) occurring in the soy flour sample were extracted in 10 min; corroborating the results obtained by Griffith and Collison [104] (see Section 5.1.1). Extending the extraction length to 30 min decreased the yield of some isoflavones. Also, no significant difference was observed between ultrasonic probe and ultrasonic bath and therefore, it can be used as an alternative with the advantage of allowing the extraction of multiple samples. The method developed by Rostagno et al. [106] has been used, with

13

and without modifications, for routine analysis, to obtain isoflavone extracts for other studies and as reference method for comparison of other extraction methods [80,82,109,110,119,126–138]. Regarding the extraction solvent, Achouri et al. [107] compared three solvents (80% MeCN + HCl 0.1N, 80% MeOH, 80% EtOH) for the ultrasound-assisted extraction of isoflavones from different matrixes (defatted soybean meal and from soy protein isolate) and observed that 80% MeOH and 80% EtOH extracted the highest amount of isoflavones from both samples. They also observed that sonication for 15 min extracted as much as the total of 5 sequential extractions (with ordinary shaking for a total of 10 h), except for acidified MeCN. This is an important observation, since acidified MeCN is one of the most used solvent with conventional extraction techniques and points that it is not recommendable to use this solvent when using ultrasounds, since it can seriously underestimate isoflavone content of foods. It was also observed that extending the time of sonication from 15 min to 30 and 60 min, did not increase the total amount of isoflavone extracted, and in some cases the total amount decreased, corroborating the observations made by Rostagno et al. [106]. More recently, Bajer et al. [129] compared pure MeOH, MeCN and ACE for the extraction of De and Ge from soy flour. MeCN gave the highest yields and was further studied adding different amounts of water (0–50%) and 60% MeCN gave the best results. Temperature was also evaluated between 25 ◦ C and 80 ◦ C as well as extraction time between 10 min and 50 min. Highest amounts of isoflavones were obtained at 50 ◦ C for 40 min using the ultrasonic bath. Also, using an ultrasonic homogenizer pulse generator was evaluated in the range of 45–98 W (100%) the use of ultrasonic pulses during extraction and the extraction time in the range of 10–50 min. Best extraction yields were obtained using 60% of ultrasonic amplitude for 30 min. These results were obtained at room temperature. In this report, unfortunately, information of the influence of studied extraction conditions and their respective data was not given, only a few isoflavones were studied and was limited in terms of the types of samples evaluated. The influence of ultrasound on the solid–liquid extraction process as regards yields or selectivity is very difficult to predict because of the interaction of many factors, either relative to the phase system (solid, liquid/solute) or to the ultrasonic reactor itself. The differences observed on the amount of water used in the solvent by the different reports may be related with the type of sample used and its characteristics. It is very likely that the amount of water needed to achieve maximum extraction efficiency might need some adjustment depending of the sample type, as reported by Murphy et al. [42] using conventional stirring. Other factors may be influencing the extraction dynamics, since UAE is affected by the ultrasonic wave distribution inside the extractor. Maximum ultrasound power is obtained at the vicinity of the radiating surface of the ultrasonic source and an abrupt decrease of the ultrasonic intensity increases as the distance from the radiating surface increases. Furthermore, the presence of solid particles can affect the ultrasonic intensity profile, which can be are attenuated depending of nature of the sample such as hardness, compactness, particle size and solute distribution [130]. Also, the influence of other important extraction variables such as frequency, intensity and the use of ultrasonic pulses have not been extensively studied in detail and for all isoflavones and therefore future investigations should focus on these issues as well as on the influence of the sample on the extraction. In general, UAE seems a potent technique for the extraction of isoflavones from soybeans and soy foods. This technique can achieve high extraction yields in less than 30 min from different sample types using the commonly used solvent in conventional methods. It clear though, that high temperatures can be used since extractions are short and that

14

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

Table 2 Developed methods using ultrasounds for the extraction of soy isoflavones and evaluated parameters. Sample used for evaluation of the method

Freeze-dried soybeans

Defatted soybean meal and soy protein

Isoflavones

Di, Gi, Gly and MGi

Di, Gi, Gly, De, Ge, Gle, MDia , MGia and MGlya

Fixed extraction conditions

Solvent: 25 mL Vibration amplitude: 100%

Sample: 2 g Solvent: 10 mL Temperature: 22 ◦ C

Ultrasound source: ultrasonic bath

Evaluated parameters Solvent: EtOH (30–70%) MeOH (30–70%) CH3 CN (30–70%) Temperature: 10 and 60 ◦ C Sample amount: 0.5–0.1 g Extraction time: 5–30 min Ultrasound source: ultrasonic probe and ultrasonic bath Solvent: 80% EtOH 80% MeOH

Selected conditions

Reference

50% EtOH, 60 ◦ C, 0.1 g, 20 min

[106]

80% MeOH and 80% EtOH, 15 min

[107]

80% CH3 CN (0.1N HCl) Extraction time: 15–60 min

Soy flour

De and Ge

Sample: 1 g (ultrasonic bath) and 2 g (ultrasonic homogenizer) Solvent: 25 mL (ultrasonic bath) and 45 mL (ultrasonic homogenizer) Temperature: RT (ultrasonic homogenizer)

Solvent: EtOH MeOH CH3 CN (50–100%)

60% CH3 CN

Temperature: 25–80 ◦ C (ultrasonic bath)

Ultrasonic bath: 50 ◦ C, 40 min

Extraction time: 10–50 min

Ultrasonic homogenizer: 30 min, and 60% vibration amplitude, pulse generator (not specified)

[129]

Ultrasound source: ultrasonic bath and ultrasonic homogenizer Pulse generator: 45–98 W Vibration amplitude: range not specified De: daidzein, Ge: genistein, Gle: glycitein, Di: daidzin, Gi: genistin, Gly: glycitin, MDi: malonyl daidzin, MGi: malonyl genistin, MGly: malonyl glycitin, ADi: acetyl daidzin, AGi: acetyl genistin, AGly: acetyl glycitin, MeOH: methanol, EtOH: ethanol, CH3 CN: acetonitrile, RT: room temperature. a Tentatively identified by literature.

intermediate to high amounts of water in the extraction solvent (40–80%) are needed to efficiently extract isoflavones. The influence of ultrasounds on isoflavone distribution during extraction should not be neglected. The same principles of isoflavone stability during extraction using conventional techniques apply when using ultrasounds. However, there are other factors which can affect stability of these compounds such as the production of radicals from the ultrasound dissociation of water. In the presence of these high energy species, oxidative reactions can take place simultaneously with the extraction reactions when water is higher than 50% [106]. This is particularly important, since as previously mentioned intermediate to high amounts of water in the extraction solvent (40–80%) are needed to efficiently extract isoflavones. Rostagno et al. [106], for example, observed a reduction of the extraction efficiency common to solvents with high amounts of water (>60%) which was attributed to an increased production of radicals from the ultrasound dissociation of water. In such report slightly lower yields of total isoflavones were obtained using extractions of 30 min with 50% EtOH at 60 ◦ C than those obtained with extractions of 20 min. Similar results were obtained by Achouri et al. [107], who observed that, in some cases, the total amount of isoflavones decreased if extraction were extended from 15 min to 30 and 60 min. In view of this evidence it seems advisable to use short extractions rather than long extractions when using ultrasounds.

Stability of isoflavones during UAE has not been apparently studied to the moment. With the available evidence that relatively short extraction times can affect isoflavone profile and content, assessment of the influence of ultrasounds on isoflavone degradation is of one of the most urging needs in future research in this field. The effect of extraction solvent, temperature, ultrasound intensity and frequency on stability of soy isoflavones during UAE and the search of effective ways to avoid degradation need to be further examined in detail in future investigations. 5.1.2.2. Pressurized liquid extraction. Pressurized liquid extraction is a procedure that combines elevated temperature (50–200 ◦ C) and pressure (100–140 atm) with liquid solvents (without their critical point being reached) to achieve fast and efficient extraction of the analytes from the solid and semi-solid samples matrix. This technique has received different names, such as accelerated solvent extraction (ASE), pressurized liquid extraction (PLE) and pressurized solvent extraction (PSE). When water is employed as the extraction solvent, the authors tend to use a different name, such as superheated water extraction (SWE), to highlight the use of this environmental-friendly solvent. For rapid and efficient extraction of analytes from solid matrices, extraction temperature is an important experimental factor. Elevated temperatures can lead to significant improvements in extraction efficiency, since it may increase solubility of target

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

compounds, diffusion rates and mass transfer of analytes to the solvent. Moreover, temperature can dramatically modify the relative permittivity of the extracting solvent, increasing selectivity. High pressure allows maintaining the solvent in a liquid state at high temperature and may increase the penetration of the solvent in the sample matrix. Extractions performed under elevated temperature and pressure results in adequate kinetics of dissolution processes and favors desorption of analytes from the surface and active sites of solid sample matrices [67,113,114,131,132]. PLE has been used in several occasions to extract isoflavones from soybeans, soy foods and other different matrixes such as Radix puerariae, Matricaria recutita, Rosmarinus officinalis, Foeniculum vulgare and Agrimonia eupatoria L. [109,121,129,133–140]. An overview of the developed methods using pressurized liquids for the extraction of soy isoflavones and evaluated parameters is presented in Table 3. The same principles of isoflavone stability during extraction using conventional techniques also apply when using PLE. The use of high temperatures can strongly affect isoflavone content and profile as previously discussed in Section 3. In the case of PLE however, temperatures used are much higher than those used in conventional methods and thus it can expected that the extend of degradation and transformations taking place during extraction are much more important. Rostagno et al. [109] evaluated the influence of several extraction parameters, such as solvent, temperature, pressure, sample size, static extraction cycle length and number of static extraction cycles in order to optimize extraction conditions to achieve quantitative recoveries of isoflavone from freeze-dried soybeans. They observed that using EtOH/water mixtures, extraction efficiency increased when increasing the water percentage in the extraction solvent from 0% to 30%, and that higher amount of water in the extraction solvent resulted in a lower extraction efficiency. Similar results were obtained for MeOH/water mixtures, and the highest extraction efficiency was achieved using 60% MeOH. Water extracted the lowest amount of isoflavones between assayed solvents. They also reported that increasing the extraction temperature from 60 ◦ C to 150 ◦ C increased the yield of most isoflavones (except the malonyl forms) and identified a degradation pattern. The increase in the extraction temperature from 60 ◦ C to 100 ◦ C increased the total amount of isoflavones extracted in approximately 20%. The increase of the yield of isoflavone glucosides with the increase of temperature between 100 ◦ C and 150 ◦ C was very pronounced (approximately 30%) while the yield of malonyl isoflavones decreased (approximately 50%), when it was expected to follow the same trend as the glucoside forms and increase. Searching for answers, a stability evaluation of extraction conditions was performed which confirmed that degradation starts above 100 ◦ C for the malonyl forms and above 150 ◦ C for the isoflavone glucosides. Above 100 ◦ C, with the decrease of MGi, a correspondent increase in Gi concentration was observed. Concentration of other glucosides also increased at this temperature. Aglycone levels remained constant below 150 ◦ C indicating that degradation of glucosides was not taking place below this temperature. Above 150 ◦ C, aglycone levels showed a small increase with the decrease in their respective glucoside levels, indicating the conversion between these chemical forms. The stability study confirmed the observations made during the extraction temperature optimization, indicating that 100 ◦ C is the maximum temperature for PLE of isoflavones. It was also reported that the increase of pressure from 100 atm to 200 atm did not have a significant impact on the extraction of isoflavones from freeze-dried soybeans, and that reducing sample size (from 0.5 g to 0.05 g) increased the yield of isoflavones in approximately 13%. However, relative standard devi-

15

ation increased proportionally. The extension of the three static extraction cycle used from 5 min to 7 min increased the extraction yield in approximately 10% and no significant effect was observed between 7 min and 10 min. To ensure that quantitative extraction was obtained, the authors performed four re-extractions of the sample achieving similar recoveries. With the evidence provided by this report, it is clear that caution should be used when increasing the extraction temperature and that more research is needed to evaluate the stability of isoflavones during PLE. Among the main factors that should be studied in detail in future researches are the influence of the sample, solvent and the duration of the procedure. An interesting method for the extraction of isoflavones from soybean foods was developed by Klejdus et al. [135] using PLE with modified extraction cell content. The modification in the extraction cell content was made by using 5 mL of a commercial matrix (SPE-edTM matrix). For the development of the method, similar extraction parameters were evaluated, like solvents, number of extraction cycles, sample amount, pressure and temperature. Additionally, another innovation of the method was the evaluation of the effect of sonication time before PLE. The extraction yield dramatically increased by using sonication before PLE extraction (performed with 90% MeOH). The amounts of extracted individual isoflavones rapidly increased with the sonication time up to 1 min, and using longer sonication times the increase was lowered and it was nearly constant after 5 min. However, extraction yield of aglycone (De and Ge) continuously increased with increasing sonication time until 5 min. The increase in the extraction efficiency was attributed to the disruption of cell walls by ultrasonic waves. Regarding extraction cycles (performed with 90% MeOH), highest isoflavone concentrations in the extracts were obtained after three extraction cycles. However, differences in the yield between two and three cycles were only about 5%. Most often used solvents (i.e. MeCN, EtOH and MeOH (50 or 60–90% in water)), were evaluated for the extraction under PLE. The extraction yields obtained for the extraction of Di and Gi with MeCN were about 60–80% (depending of the water percentage) of the yields of the extraction with MeOH (90%). The extraction efficiency rapidly decreased with the increasing content of MeCN in the case of Di. Extraction yields between 60% and 75% of the amount extracted with MeOH (90%) were obtained using EtOH with different amounts of water (60–90%). Highest yields of both isoflavones were obtained using 90% MeOH, and linear decrease of extraction yields were obtained with decreasing content of MeOH in the extraction agent. Using different amounts of sample, the authors observed decreasing yields with the increasing amounts of sample, obtaining similar results as those obtained by Rostagno et al. [109]. This finding was attributed to the thicker layer of sample in the extraction cell. In such study, the influence of pressure was also evaluated. The decrease in the sample size from 0.5 to 0.1 increased approximately 40% the amount of Di and Gi extracted. With the increase of pressure from 13 kPa to 14 kPa of pressure the amount extracted of both isoflavones increased, and no significant difference was observed between 14 kPa and 15 kPa. Increasing extraction temperature from 70 ◦ C to 110 ◦ C produced an increase of the extraction yield of Di and Gi (15%), and a much more dramatic increase was observed between 110 ◦ C and 145 ◦ C (60%). The authors claimed that temperature of about 145 ◦ C was most suitable for obtaining maximal efficiency. The optimized method was used for the analysis of several isoflavones different soy samples. However, the greatest concern regarding the proposed method is that the malonyl forms were not measured and stability was not evaluated. Using as reference the stability study made by Rostagno et al. [109] where most malonyl isoflavones are not stable

16

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

Table 3 Developed methods using pressurized liquids for the extraction of soy isoflavones and evaluated parameters. Sample used for evaluation of the method

Freeze-dried soybeans

Soy bits

Isoflavones

Di, Gi, Gly and MGi

Di, Gi, De and Ge

Fixed extraction conditions

Evaluated parameters

Extraction cell: 11 mL Inert material: sea sand

Solvent: EtOH (30–80%) MeOH (30–80%) Water Temperature: 60 and 200 ◦ C

Sample amount: 0.2 g Extraction cell: 10 mL Cell content: 5 mL of a commercial matrix (SPE-edTM matrix) Static cycle length: 5 min

Selected conditions

Reference

0.1 g, 100 ◦ C, 70% EtOH, 3× 7 min static cycles (∼22 mL)

[109]

1 min sonication time, 0.1 g, 90% MeOH, 14 kPa, 2 static cycles (∼20 mL)

[135]

70% EtOH + 5% DMSO

[134]

CH3 CN, 2 static cycles, of 15 min

[129]

110 ◦ C, 641 psig, 2.3 h

[139]

80% EtOH, 383 K, 551 kPa, 25 mL/min, 80 g

[140]

Pressure: 100–200 atm Sample amount: 0.5–0.05 g Static cycle length: 5–10 min Number of static cycles: 1–3 (7 min) and 1–2 (10 min) Solvent: CH3 CN EtOH (50–90%) MeOH (50–90%)

Pressure: 13–15 kPa Sonication time: 1–5 min Number of static cycles: 1–3

Soybeans

Soybean flour

Di, Gi, Gly, De, Ge, Gle, MDi, MGi, MGly, ADi, AGi and AGly

De and Ge

Sample amount: 0.5 g Extraction cell: 11 mL Inert material: Ottawa sand Pressure: 1000 psi

Solvent: 58% CH3 CN, 58% CH3 CN + 5% DMSO 70% EtOH, 70% EtOH + 5% DMSO 90% MeOH

Temperature: 100 ◦ C Static cycle length: 5 min Number of static cycles: 3

Water 95% Genapol

Sample amount: 2 g Temperature: 100 ◦ C

Solvent: MeOH, ACE, CH3 CN Pressure: 5–15 MPa

Inert material: quartz wool/glass beds

Number of static cycles: n.e. Extraction time: n.e.

Defatted soybean flakes

Defatted soybean flakes

Di, Gi, Gly, De and Ge

Di, Gi, Gly, De and Ge

Sample amount: 180 g Extraction cell: 2 L

Temperature: 60–130 ◦ C Pressure: 300–735 psig

Solvent: 1800 mL of water

Extraction time: 1–3 h

Extraction cell: 2 L

Temperature: 333–393 K Pressure: 413–4410 kPa Solvent flow rate: 10–25 mL/min

Solvent: EtOH:water ratio (0–95%) Sample amount: 80–450 g De: daidzein, Ge: genistein, Gle: glycitein, Di: daidzin, Gi: genistin, Gly: glycitin, MDi: malonyl daidzin, MGi: malonyl genistin, MGly: malonyl glycitin, ADi: acetyl daidzin, AGi: acetyl genistin, AGly: acetyl glycitin, MeOH: methanol, EtOH: ethanol, CH3 CN: acetonitrile, n.e: not specified.

under PLE above 100 ◦ C (using 70% EtOH), and a similar dramatic increase on the yield of glucosides was observed at 150 ◦ C, part of the effect of increasing the temperature in the increase of the extraction yield of glucosides may be attributed to degradation of malonyl isoflavones. Corroborating evidence is that yield of glucoside decreased when extractions were performed above 145 ◦ C, as reported by Rostagno et al. [109]. Therefore, the proposed method may not be able to extract all isoflavones and more importantly, without changing the isoflavone profile of the sample. The same method with slight modification (i.e. sonication time of 5 min instead of 1 min) was used for evaluation of isoflavone aglycone and glucoside distribution in soy plants and soybeans by Klejdus et al. [136].

Also, Klejdus et al. [137] improved the previous method and used a two phase PLE extraction program combined with UAE to extract isoflavones from soy bits. In the first PLE phase, the sample was extracted with 2 cycles of 5 min each with hexane at 145 ◦ C using 145 bar of pressure, followed by a second phase of 2 cycles of 5 min each with 90% MeOH at 145 ◦ C using 145 bar of pressure. This is an interesting approach since it allows “cleaning” the sample and performing the extraction of target compounds without manipulation of sample, avoiding the associated errors. However, the same stability issues of the original method [136] persist in the improved method. Later, Luthria et al. [134] using the method developed by Rostagno et al. [109], compared several extraction solvents (58%

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

MeCN, 70% EtOH, 90% MeOH, Water and 95% Genapol) and evaluated the influence of the addition of 5% DMSO to the extraction solvent for the extraction of isoflavones from soybeans. They observed great differences between assayed solvent. Both, the total isoflavone content and the isoflavone HPLC profile varied significantly with different extraction solvents, achieving highest total isoflavone recoveries from soybean samples with DMSO:EtOH:water. 58% ACN extracted only 30.5% of the isoflavones extracted with DMSO:EtOH:water. With the addition of DMSO to 58% ACN improved extraction to 52.3%. The addition of DMSO to 70% EtOH also improved extraction efficiency, while 90% MeOH achieved intermediate yields (83.7%). Very low efficiency was obtained with genapol or water (18.2% and 13.7%, respectively). However, since extraction conditions used were optimized by Rostagno et al. [109] for 70% EtOH, it was expected that the maximum efficiency was obtained using DMSO:EtOH:water (5:70:25) and EtOH:water (70:30). On the other hand, useful information is provided by the improvement of extraction by DMSO. A possible explanation to the improvement of the extraction efficiency was attributed to the solubility of isoflavones in DMSO reported by Sigma–Aldrich web site (http://www.sigma-aldrich.com). The authors also observed important differences among assayed solvents. 90% MeOH extracted the highest amount of glucosides (Gi, Di and Gly) and while for the other nine isoflavones the best solvent was DMSO:EtOH:water (5:70:25). This may explain the results obtained by Klejdus et al. [135], achieving best yields with MeOH than with EtOH, since only glucosides and aglycones were quantified. An interesting observation was the detection of all 12 isoflavones by only two extraction solvent mixtures (DMSO:EtOH:water (5:70:25) and (DMSO:MeCN:water (5:70:25)). De and MGly were not extracted at detectable levels by the other solvents. Similarly, Klejdus et al. [136] extracted only trace amounts of De (1.2% of total isoflavones) and did not detect MGly when using 90% MeOH. In contrast, Bajer et al. [129] observed that out of three solvents tested (MeOH, MeCN and ACE) MeCN gave the highest yields at 100 ◦ C. However, extraction was evaluated for some aglycones (De and Ge only) and the use of a certain amount of water in the extraction solvent was very likely to have influenced the results obtained. After the pressure was optimized in the range of 5–15 MPa, the number of cycles and extraction time were also optimized using MeCN. Unfortunately, data regarding the method optimization and of the influence of the extracting variables were not given. With a different optimization strategy, Li-Hsun et al. [139] used a steepest ascent design to examine the effect of several independent variables (temperature, pressure and duration) on the extraction of isoflavones from defatted soybean flakes by superheated water at elevated pressures. They observed that temperature has a greater impact than pressure and then time, in the extraction of isoflavones using water. The experimental design revealed that the optimal condition for the extraction of isoflavones was 110 ◦ C and 641 psig (4520 kPa) for 2.3 h using 180 g of sample and 1800 mL of water. When extractions were carried out at higher or lower temperature, or with lower pressure, the total amount of isoflavones decreased. The authors concluded that the decreasing dielectric constant (ε) of water at elevated temperature and pressure might play an important role for the enhanced extraction of isoflavones. This is indication that the dramatic changes in the physical–chemical properties of water, especially in its dielectric constant, at elevated temperatures and pressures enhance its usefulness as extraction solvent. The low extraction efficiency of water observed by Rostagno et al. [109] when compared to the above results can be explained by the use of a lower temperature (60 ◦ C), which is not high enough to change the dielectric constant of water and increase its effective-

17

ness. An important remark is that not all isoflavone chemical forms were quantified. Also, the large sample amount and solvent volume and the inability to extract some isoflavones limit its usefulness as an analytical method. However, the reported results provides an important evidence that isoflavones can be extracted using pressurized water if conditions are optimized enough which could be exploited in future investigations. Following the same direction, Chang and Chang [140] examined the effect of pressure, temperature, solvent flow rate, EtOH:water ratio, and the feed loading on the PLE of isoflavones from defatted soybean flakes using water as solvent. Initially, the effect of solvent flow and extraction temperature were evaluated and was observed that using hot pressurized water, increasing the solvent flow rate increased the extraction efficiency as the extraction time increased from 40 min to 200 min. The effect of the increasing temperature was noticeable between 333 K and 383 K but not between 383 K and 393 K (all extraction were performed at 4410 kPa) independently of the solvent flow rate (5 and 10 mL/min). These results are similar to those obtained by Li-Hsun et al. [139]. Using pure water or EtOH (at 383 K and 4410 kPa) the later procedure extracted more than 50% more isoflavones than water. Decreasing feed loading from 480 g to 180 g increased isoflavone extraction efficiency in approximately 20% after 6 h of extraction, being in agreement with the trends observed by Rostagno et al. [109] and Klejdus et al. [135]. Pressure, however, did not significantly affect the PLE using EtOH with 360 g of feed loading as reported by Rostagno et al. [109]. In contrast, solvent flow have had an important effect on extraction efficiency, and increasing flow rate from 10 mL/min to 25 mL/min increased the total amount of isoflavones extracted in approximately 15% after 360 min. The optimization of EtOH:water ratio, feed loading, pressure and solvent flow rate on the recovery of isoflavones of the PLE at 383 K and 2.3 h was made by means of a four factor Taguchi experimental design. EtOH:water ratio and flow rate were the parameters with the highest influence on the extraction efficiency, while small differences were observed while evaluating feed loading and pressure. In general, the higher the EtOH:water ratio and the flow rate, the higher was the recovery. Among the variables affecting PLE, the nature of the extraction solvent and temperature generally have profound effects on the PLE process. When dealing with pressurized solvents, temperature will have different impact depending of the solvent used since physical–chemical properties of each solvent are different. Therefore, the best extraction conditions will depend of the solvent. The dramatically changes in the physical–chemical properties of water, especially in its dielectric constant (ε) at elevated temperatures and pressures enhance its usefulness as an extraction solvent. The dielectric constant is a key parameter in determining solute–solvent interactions, and increasing the temperature under moderate pressure can significantly decrease this constant. At ambient pressure and temperature, water is a polar solvent with a dielectric constant (ε = 78) but at 300 ◦ C and P = 23 MPa this value decreases to 21, which is similar to the value of EtOH (ε = 24 at 25 ◦ C) or acetone (ε = 20.7 at 25 ◦ C). This means that at elevated temperatures and moderate pressures the polarity of solvent can be reduced considerably and water can be used instead of another organic solvent to extract medium-low polarity compounds, or reduce the amount of the organic solvent used to achieve effective extraction rates [67]. Due to the advantages of lower cost, environmental compatibility and toxicity, water can be used as extracting solvent if high efficiency is not required. However, in view of the available literature, pure water, even at elevated temperature and moderate pressure, is not as efficient as other solvents for the extraction of isoflavones, despite that the addition of certain amount of water to the organic solvent is necessary to improve extraction efficiency.

18

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

In general, it is clear that at higher temperatures extraction efficiency tend to increase, independently of the solvent employed and pressure is usually a minor variable (except when using water as solvent) for the resulting efficiency and that it is only required to maintain the solvent in the liquid phase. Another important aspect of PLE is the stability of isoflavones under extraction conditions. Since some isoflavone are very sensitive to temperature and PLE is performed at elevated temperatures, stability may be the limiting factor when using PLE for the extraction without changing the isoflavones profile of the sample. However, this may be considered as positive in the case of hydrolytic methods (see Section 4). To date no hydrolytic method using PLE has been reported and the potential changes in isoflavone profile using high temperatures can be explored in future investigations. Also, the influence of the sample properties, such as particle size, protein content and enzymatic activity and its relation with extraction efficiency and isoflavone stability should be investigated in more detail. 5.1.2.3. Supercritical fluid extraction. Supercritical fluid extraction is the process of separating one component (the extractant) from another (the matrix) using supercritical fluids as the extracting solvent. A supercritical fluid is any substance at a temperature and pressure above its thermodynamic critical point. They can penetrate samples of plant material almost as well as gases, due to their high diffusion coefficients and low viscosity. At the same time, their dissolving power is similar to liquids. Additionally, close to the critical point, small changes in pressure or temperature result in large changes in density, allowing many properties to be modified and to obtain selective extraction. The most commonly used extracting agent is carbon dioxide (CO2 ), because of its low cost, low toxicity, and easily reachable critical parameters (31.1 ◦ C/74.8 atm). Furthermore, CO2 as a non-polar substance is able of dissolving non-polar or moderately polar compounds. The addition of a polar modifier (e.g. MeOH) to supercritical CO2 (SC-CO2 ) is the simplest and most effective way to modify the polarity of CO2 -based fluids in order to increase the solubility of analytes. Modifiers can also overcome interactions between the analyte and the matrix, increasing the extraction efficiency of polar organic compounds [113,114,132,141–143]. Although SFE is one of the most complex technique for the extraction of isoflavones due to the high number of possible variables and interactions, which can effect effectiveness, several researchers successfully applied SFE to extract isoflavones from different soy matrixes such as soy flour, soy hypocotyls and soy cake, as well as from other different matrices, like R. puerariae, M. recutita, R. officinalis, F. vulgare and A. eupatoria L. [91,92,118,129,144,145]. An overview of the developed methods using supercritical fluids for the extraction of soy isoflavones and evaluated parameters is presented in Table 4. As in most modern techniques and methods, stability of isoflavones under extraction conditions has not been studied so far. This is important since relatively high temperatures are frequently used. The same stability principles of the previously discussed techniques may apply to SFE and thus it is feasible to consider that changes in isoflavone profiles can take place during extraction. Therefore, evaluation of stability of isoflavones using different SFE conditions, such as temperature, duration and amount and type of modifier is urgently needed. Regarding the methods developed so far, Chandra et al. [145] tested a limited number of conditions with different pressures and amount and type of modifier for the extraction of some isoflavones (De and Ge) from various soy matrixes. The evaluation of the extraction conditions revealed that at 50 ◦ C, 600 atm and 20% EtOH extracted the highest amount of tested isoflavones (nearly 93%). It is worth noting that the development of the method was per-

formed spiking reference standards onto a filter paper strip which was later extracted by SC-CO2 . The best-evaluated conditions were used for the extraction of De and Ge from miso, tofu, and soy meal and soy flour using sample sizes ranging from 2 g to 10 g. Although high recoveries were achieved, the method was limited in terms of isoflavones quantification. Later, Rostagno et al. [118] evaluated the use of supercritical carbon dioxide for the extraction of soybeans isoflavones (Gi, Ge and De) using different temperatures, pressures and modifier concentration. Maximum yield of Gi and Ge was obtained at 70 ◦ C/200 bar/10 mol%, while maximum yield of De was obtained at 50 ◦ C/360 bar/10 mol%. For the extraction of Gi and Ge, a predominant effect of temperature was observed while for De, a predominant effect of pressure was observed. Also a strong interaction between temperature and pressure was observed in the extraction of the tested isoflavones. The decrease in extraction efficiency with the increase in the temperature can be explained by the decrease in the supercritical fluid density, while the decrease in extraction efficiency with the increase in pressure can be attributed to a decreased fluid diffusivity, which may affect interaction with the sample. However, it is important to note that stability of isoflavones was not accessed, that only one glucoside (Gi) and aglycones were measured and that malonyl glucosides were not quantified. Since relatively high temperatures were used, it is plausible that degradation might have taken place during extraction influencing the obtained results. The authors suggested that enzymatic hydrolysis of Di might have occurred during extraction and influenced the results, since the best extraction temperature for De was 50 ◦ C, close to the optimal temperature for the activity of ␤-glucosidases. More recently, Kao et al. [91] modified the experimental conditions optimized by Rostagno et al. [118] and used 70% EtOH as modifier, instead of 70% MeOH and studied a similar range of temperature and pressure for the SFE of isoflavones (all 12 main chemical forms present in soybeans) from soybean cake. The most important aspect of this method, besides the high recovery (87.3% when compared to stirring extraction), was the quantification of all isoflavone chemical forms, since it is the first report of the use of supercritical fluids for the extraction of malonyl and acetyl isoflavones. The results showed that a maximum yield of malonyl glucoside and glucoside was obtained at 60 ◦ C and 350 bar, while a high level of acetyl glucoside and aglycone was produced at 80 ◦ C and 350 bar. The highest yield of total isoflavones was obtained using 60 ◦ C/350 bar, possibly due to predominant concentration of malonyl and glucosides in the sample. Although a different modifier was used, similar temperature and pressure interaction as reported by Rostagno et al. [118] was observed. However, as in most studies, stability was not evaluated and results might be influenced by degradation of malonyl and glucoside isoflavones to their respective acetyl and aglycone forms. The authors observed that, although using lower temperatures than cooking and toasting, conversion or degradation can still occur when in combination with pressure. The amount of malonyl glucosides declined after following a rise in the extraction temperature, which was suggested to be related with solubility of these chemical forms or to conversion to acetyl glucoside, glucoside or aglycone, which may explain highest yields obtained at 80 ◦ C. Araújo et al. [92] also tested different temperatures, pressures, modifiers, and modifier concentration for the SFE of De and Ge from soybean hypocotyls after hydrolysis. The highest yields of these isoflavones were obtained at 60 ◦ C, 380 bar using 3 cycles of static and dynamic extraction of 15 min each with the addition of 10 mol% of 80% MeCN. Moreover, it was observed that modifiers and pressure variations have significant effects in the extraction efficiency. No isoflavones were extracted without modifiers and the

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

19

Table 4 Developed methods using supercritical fluids for the extraction of soy isoflavones and evaluated parameters. Sample used for evaluation of the method

Standards

Isoflavones

Fixed extraction conditions

Evaluated parameters

Selected conditions

Reference

De and Ge

Extraction cell: n.e. Temperature: 50 ◦ C Flow rate: 950–1000 mL/min Extraction time: 60 min Restrictor temperature: 175 ◦ C Rinse solvent: none

Extraction conditions: 400 atm and no modifier 400 atm and 5% chloroform 400 atm and 5% MeOH 600 atm and 20% MeOH 600 atm and 20% EtOH

600 atm and 20% EtOH

[145]

Sample amount: 1 g Extraction cell: 7.0 mL (reduced to 5.46 mL) Inert material: glass stick Modifier: 70% MeOH Static cycle length: 10 min Freeze-dried soybeans

Gi, Ge and De

Dynamic cycle length: 20 min CO2 flow rate: 1.0 mL/min Extraction time: 90 min (3× 30 min) Trap: ODS Rinse solvent: 1.5 mL MeOH Rinse flow rate: 0.5 mL/min

Modifier concentration: 0.5 and 10 mol%a Temperature: 40–70 ◦ C

Soybean hypocotyls

Di, Gi, Gly, De, Ge, Gle, MDi, ADi, MGi, AGi and MGly

De and Ge

Temperature: 50–80 C

Dynamic cycle length: 20 min CO2 flow rate: 1.0 mL/min Extraction time: 90 min (3× 30 min) Fluxing solvent: 5 mL 50% EtOH

Pressure: 300–400 bar

Static cycle length: 15 min Dynamic cycle length: 15 min CO2 flow rate: 1.5 mL/min Extraction time: 90 min (3× 30 min) Rinse solvent: 1.5 mL 80% MeOH Rinse flow rate: 0.5 mL/min

Soybean flour

De and Ge

Sample amount: 0.3 g Inert material: glass beads

Sample amount: 100 g Soybean meal

Di, Gi, De and Ge

◦

Static cycle length: 10 min

Sample amount: 0.08 g Extraction cell: 7.0 mL (reduced to 5.46 mL) Inert material: glass stick Modifier: 70% MeOH

Extraction cell: 1 L

Separator 1: 8 ± 0.3 MPa (40 ◦ C) Separator 2: 6 ± 0.3 MPa (30 ◦ C)

[118]

Pressure: 200–360 bar

Sample amount: 1 g Extraction cell: 10 mL Modifier: 70% EtOH Modifier concentration: 10 mol%a

Soybean cake

50 ◦ C/360 bar, 10 mol% (TIS and De) 70 ◦ C/200 bar, 10 mol% (Gi and Ge)

Modifier: MeOH, EtOH and CH3 CN Modifier concentration: 0.5 and 10 mol%a Temperature: 50–70 ◦ C Pressure: 176–360 bar

Pressure: 15–40 MPa Temperature: 10–100 ◦ C Extraction time: 10–50 min Restrictor diameter: 25 and 50 ␮m Temperature: 40–70 ◦ C Pressure: 30–60 MPa Modifier composition: MeOH (60–100%) Modifier concentration: 5.4, 7.8 and 10.2 mass%a

Malonyl glucosides, glucosides and TIS: 60 ◦ C/350 bar Acetyl glucosides and aglucones: 80 ◦ C/350 bar

[92]

60 ◦ C/380 bar, 10 mol% 80% CH3 CN

[93]

35 MPa, 70 ◦ C, 5% MeOH, 30 min, 50 ␮m

[129]

40 ◦ C, 50 MPa, 9.80 kg/h, 80% MeOH at 7.8% mass, 20–30 mesh, 200 min

[147]

CO2 flow rate: 3.92–9.80 kg/h Sample particle size: 10–60 mesh Extraction time: 0–200 min

De: daidzein, Ge: genistein, Gle: glycitein, Di: daidzin, Gi: genistin, Gly: glycitin, MDi: malonyl daidzin, MGi: malonyl genistin, MGly: malonyl glycitin, ADi: acetyl daidzin, AGi: acetyl genistin, AGly: acetyl glycitin, TIS: total isoflavones, MeOH: methanol, EtOH: ethanol, CH3 CN: acetonitrile, n.e: not specified. a mol% of the CO2 mass passed through the system during the dynamic extraction.

20

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

predominant effect of the pressure in the amount of these two isoflavones (De and Ge) extracted was attributed to the likely decrease of the steam pressure and increase in the density of fluid and a higher kinetics of desorption of the compounds from the sample matrix. As pressures increases, desorption is faster and more solute is available for extraction. They also observed similar trend as observed by Rostagno et al. [118], where an enhancement of the extraction yield by the increase of pressure was dependent of the temperature with correlation of pressure and temperature. Also, major differences were observed for the assayed modifiers. Using 80% MeOH, 80% EtOH and 80% MeCN as modifier, the relative amount of aglycones extracted were 9.61%, 11.27% and 25.65% respectively when compared to stirring. It is clear that using the proposed method 80% MeCN is much more effective than 80% EtOH and 80% MeOH. Another approach is to previously use SC-CO2 to enrich, or “clean”, the sample matrix with isoflavones by removing other components from the matrix, as reported by Yu et al. [146], where soy hypocotyls were defatted by SC-CO2 and used to produce isoflavone enriched soy protein. Bajer et al. [129] optimized the extraction of De and Ge from soy flour using different pressures (15–40 MPa), temperatures (10–100 ◦ C) and extraction times (10–50 min). Optimal conditions were 35 MPa, 70 ◦ C, modifier MeOH (5%, v/v) and 30 min. The authors reported clogging using restrictors of both 25 and 50 ␮m of internal diameter and placing the sample between glass beds avoided the problem. Unfortunately, data obtained during the method optimization was not provided, which could have been useful for future investigations on the use of supercritical fluids for the extraction of isoflavones. One of the latest applications of supercritical fluid for the extraction of isoflavones from soy matrixes was recently published by Zuo et al. [147]. In this report the influence of several extraction parameters, such as pressure, temperature, flow rate, modifier composition and concentration as well as sample particle size, was evaluated for the extraction of some isoflavones (De, Ge, Gi and Di) from soybean meal. They observed that using specific conditions (40 ◦ C/50 MPa, 5.88 kg/h, modifier flow rate of 0.6 L/h) higher or lower amounts of water than 20% in MeOH (i.e. 80% MeOH) extraction yield decreased, especially when using high MeOH concentrations (i.e. 90–100%). This effect may be related with the polarity of the supercritical CO2 which is excessively changed by high or low amounts of modifier. However, it does not exclude the possibility that if extraction conditions were different (especially temperature and pressure), the highest yield modifier could have been achieved using a different modifier composition since extraction conditions may influence CO2 polarity. The authors also observed that using 80% MeOH, higher modifier concentrations (i.e. 10.2 mass%) resulted in highest extraction yields, which can also be explained by the changes in CO2 polarity. Here, extraction conditions can also influence the effect of the modifier concentration. However, these results confirm the observations made by Rostagno et al. [118], who also reported that higher modifier concentration (10 mol% of 70% MeOH) resulted in higher extraction efficiency. In most extraction techniques the increase of temperature (which may be limited by stability of some isoflavones) usually increases extraction efficiency due to several factors. This may not be the case for SFE, since higher temperatures decreases CO2 density (maintaining pressure constant) and thus solvent power of the fluid. This issue has been demonstrated by Rostagno et al. [118] and reflects the results obtained by Zuo et al. [147], who observed that increasing temperature from 40 ◦ C to 70 ◦ C decreased extraction yield. Increased extraction temperature favors several processes such as mass transfer, solubility and volatility of isoflavones and increase penetration power of SC-CO2 , but the positive influence of

these factors may not be sufficient to counteract the reduced CO2 density. Also, the higher temperature may increase the solubility of other sample components in the SC-CO2 and reduce the extraction efficiency of isoflavones. Pressure (at 40 ◦ C) was also evaluated by Zuo et al. [147] and a straightforward trend was observed, achieving highest yields using higher pressures, possibly due to increased CO2 density. Increasing CO2 flow rate increased yields, obviously due to increased mass of CO2 used. In this report, the most interesting extraction parameter evaluated (which has not been assessed so far) was the influence of sample particle size. Reducing particle size from 1.19 mm (10–20 mesh) to 0.68 mm (20–30 mesh) improved extraction yields and smaller particle size decreased extraction yields. Smaller particles provide greater contact surface and better allow the penetration of the SC-CO2 and consequently extraction efficiency increase. The explanation given by the authors to the decrease of yields observed when using smaller particle sizes than 0.68 mm was the aggregate formation, which can cause the fluid to channel or short circuit. One of the limiting aspects of this study was the use of one-attime strategy to optimize extraction conditions, which is not the most recommended to be used for supercritical fluids, since there are much more variables and interactions that can more severely influence effectiveness than in other extraction techniques. The best approach when using SC-CO2 seems to use experimental designs or, preferably, a full screening of extraction conditions. Also, the quantification of only glucosides and aglycones provide only limited information on the effect of evaluated parameters on the extraction of malonyl and acetyl glucosides, which have been demonstrated to be the most difficult isoflavone forms to be extracted by supercritical fluids. There are a large number of variables that can affect the extraction using SFE, including not only pressure, temperature and type and amount of modifier, but also others such as the duration of dynamic and static cycles, CO2 flow rate, thimble times swept, restrictor temperature, trap composition, rinse solvent and rinse flow rate. Trap composition and elution conditions (solvent, flow rate and temperature) have not been studied so far and may strongly influence isoflavone recovery and need to be investigated in the future. Also the sample is an important factor since the particle size and interactions with the matrix (i.e. protein content) may greatly influence the ability of the supercritical fluid to extract target compounds. Other sample characteristics may influence stability during extraction (i.e. glucosidase activity). The presence of other sample components, such as oil, may interfere with the amount of target compounds that can be extracted depending of the fluid density. Moreover, extraction conditions can have a great impact on the effectiveness of a certain extraction parameter, and selection of extraction variables should be carefully studied in order to achieve maximum effectiveness. Therefore, much research is still needed to fully determine the potential of supercritical fluids for the extraction of isoflavones from soybeans and soy foods. 5.1.2.4. Microwave-assisted extraction. Microwaves are electromagnetic waves with wavelengths ranging from 1 mm to 1 m, or frequencies between 300 MHz and 300 GHz. Microwave-assisted extractions are based on absorption of microwave energy by molecules of polar chemical compounds. The energy absorbed is proportional to dielectric constant of the medium, resulting in rotation of dipoles in an electric field (usually 2.45 GHz). The efficiency of MAE depends on several factors, such as solvent properties, sample material, the components being extracted, and, specifically, on the respective dielectric constants. The higher dielectric constant, more energy is absorbed by the molecules and the faster the solvent reaches the extraction temperature. In most cases, the extracting solvent has a high dielectric constant and strongly absorbs

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

microwave radiation. Microwaves may also promote selective and rapid localized heating of moisture in the sample by microwaves. Due to the localized heating, pressure builds up within the cells of the sample, leading to a fast transfer of the compounds from the cells into the extracting solvent. Additionally, by using closed vessels the extraction can be performed at elevated temperatures (above boiling point of the solvent) accelerating the mass transfer of target compounds from the sample matrix [113,114,132,148]. This technique has been used only recently and on a few occasions to extract isoflavones from soybeans and from different matrixes, such as R. astragali, R. puerariae and peanuts [86,110,124,149,150]. An overview of the developed methods using microwaves for the extraction of soy isoflavones and evaluated parameters is presented in Table 5. Since high temperatures are usually used by MAE is convenient to discuss the stability of isoflavones during extraction before detailing developed methods. There is apparently only one study of soy isoflavone stability during MAE, which was reported by Rostagno et al. [110]. They evaluated stability of soy isoflavone extracts (50%EtOH) at different temperatures (50–150 ◦ C) for 30 min and observed that temperatures above 50 ◦ C significantly changed isoflavone profile of the extracts mainly due to conversions between malonyl and their respective glucosides. Malonyl isoflavones were not detected above 100 ◦ C and temperatures higher than 125 ◦ C promoted hydrolysis of glucosides to their respective aglycones. Regarding the developed methods, several extraction solvents, temperatures, and extraction solvent volume, as well as the sample size and extraction time were studied by Rostagno et al. [110] for the optimization of an extraction protocol for all main soy isoflavones. In the first selection of the most adequate extraction solvent, pure solvents (MeOH, EtOH and water) extracted lower amounts than 50% EtOH and 50% MeOH. Since 50% EtOH extracted the highest amounts of total isoflavones, it was further studied in terms of water percentage (30–70%) and results indicated that using higher or lower amounts of water than 50% reduced extraction efficiency. A similar result was obtained with solvent volume; using high or low solvent volumes resulted in lower extraction efficiency than intermediate volumes (20–30 mL). After solvent and volume were optimized, sample amount was evaluated revealing that using low sample size reduce extraction efficiency and sample size greater than 0.5 g does not improve yield. Therefore, the authors concluded that a sample:solvent ratio of 0.5:25 (g/mL) results in maximum efficiency using the optimized conditions so far. The effect of sample size is different when using MAE than with other techniques, such as PLE, which increases extraction efficiencies using smaller samples. Examining the effect of extraction time the authors obtained a straightforward response: increasing extraction time increased extraction yield, and quantitative extractions were obtained in 20 min. They also observed that most isoflavones present in the sample (approximately 75%) were extracted in the first 10 min of extraction. No isoflavone degradation was observed using the developed extraction protocol and a high reproducibility was achieved (>95%). The most interesting point of this report was the proposal of an effectiveness factor in order to evaluate extraction conditions. Most authors simply use total isoflavones as reference of extraction efficiency. However, sometimes there is no significant difference in the total isoflavone yield although there are significant differences between chemical forms, and selection of a certain solvent can be complicated procedure. In the case of the proposed effectiveness factor, it balances the extraction effectiveness for all isoflavones and more accurately identify the most adequate solvent. Also recently, Careri et al. [86] adopted a fractional factorial design to develop a hydrolytic method for the extraction of isoflavonoid aglycones (Di, De, Gi, Ge, Biochanin A and Formonotin)

21

from yellow soybeans using microwave-assisted extraction. For development of the method, several extraction parameters were evaluated such as microwave power, extraction time, solvent, extraction volume, acid concentration and re-hydratation time. Several interactions among extraction variables were found and the most relevant parameters resulted to be the microwave power, the extraction time and the acid concentration. It is important to note that sample was submitted to sonication for 15 min before extraction using MAE, which may have extracted most isoflavones in the sample, as previously discussed, and the proposed method can almost be considered a hydrolysis method rather an extraction method. It is clear that although highly efficient extractions can be achieved with MAE, its potential is limited since only short extractions can be used due to isoflavones stability. However, this may be very useful when proposing hydrolytic methods, which can be an interesting application for this extraction technique. Also, more research is needed to determine the influence of other parameters not only on extraction but also on stability, such as pressure. Sample characteristic such as humidity, ␤-glucosidase activity, and particle size also need to be investigated in future researches. 5.2. Liquid samples Apart of solid samples, there are liquid soy samples that also contain isoflavones. Most common liquid soy foods are soy milk and soy beverages blended with fruit juices. Usually, these samples are freeze-dried and treated as solid samples, using the same methods and techniques [41,44,46–48,87,102,151,152]. The problem with the freeze-drying procedure is that it can take days and may as well, increase variations on the determination of isoflavones, due to increased errors and degradation of the sample. Moreover, it goes in the opposite direction of the recent trend of sample preparation that is to use fast methods and reduce to a minimum the necessary steps from sample to analysis. It is not logical to have at hand an extraction method that can be accomplished in 20 min and use a sample pretreatment of days. Liquid samples are similar to a solid sample in that most isoflavones are in the suspended solids or already in the liquid phase and therefore an extraction step can be used to extract the isoflavones from the solids and avoid freeze-drying the sample. Indeed some authors successfully direct extraction of liquid samples without freeze-drying. Most authors used methanol (MeOH) or ethanol (EtOH) with a sample:solvent ratio ranging from 4:1 to 1.6:1 (v/v) and extraction by refluxing or shaking for 1–4 h [51,77,153,154]. These methods were adapted from protocols used for solid samples and more importantly, were not evaluated, very long extraction times were used or only a few isoflavones were studied. Also, the use of refluxing causes malonyl isoflavones to undergo degradation to the respective glucosides and aglycones, changing the isoflavone profile of the samples and limiting the information obtained. On the other hand, they indicate that an extraction step can used instead of freeze-drying the sample. This was demonstrated by Rostagno et al. [151], who recently developed a new method for the fast determination of isoflavones from soy beverages blended with fruit juices using UAE. During the method development, several parameters were studied such as solvent (methanol and ethanol), sample:solvent ratio (5:1 to 0.2:1), temperature (10–60 ◦ C) and extraction time (5–30 min). The most important parameter for the extraction of isoflavones from soy drinks was the sample:solvent ratio. Also, samples were freeze-dried, extracted using a reference method and compared with the optimized method and no significant difference was observed on total and individual isoflavone concentration.

22

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

Table 5 Developed methods using microwaves for the extraction of soy isoflavones and evaluated parameters. Sample used for evaluation of the method

Freeze-dried soybeans

Isoflavones

Di, Gi, Gly, De, Ge, Gle, MDi, MGi, MGly, ADi, AGi and AGly

Fixed extraction conditions

Solvent: 25 mL Microwave power: 500 W Magnetic stirring: 50% nominal power

Evaluated parameters Solvent: Water EtOH, 50% EtOH MeOH, 50% MeOH EtOH (30–70%)

Selected conditions

Reference

0.5 g, 50 ◦ C, 20 min and 50% EtOH

[110]

Temperature: 50 and 150 ◦ C Solvent volume: 15–35 mL Sample amount: 0.1–5.0 g Extraction time: 10–30 min

Soybeans

Di, De, Gi, Ge, Biochanin A and Formononetin

Sample amount: 0.1 g Sonication before extraction: 15 min

Microwave power: 10 and 600 W Extraction time: 1 and 8 min Solvent: 80% CH3 CN and 80% MeOH Solvent volume: 3 and 8 mL

600 W, 1 min, 3 mL of 80% CH3 CN, 12 M HCl and no re-hydratation time

[87]

Acid concentration: 1 and 12 M Re-hydratation time: 0 and 10 h De: daidzein, Ge: genistein, Gle: glycitein, Di: daidzin, Gi: genistin, Gly: glycitin, MDi: malonyl daidzin, MGi: malonyl genistin, MGly: malonyl glycitin, ADi: acetyl daidzin, AGi: acetyl genistin, AGly: acetyl glycitin, MeOH: methanol, EtOH: ethanol, CH3 CN: acetonitrile.

The novelty of this work resides in its simplicity and rapidity when treating a troublesome liquid sample without the need of freezedrying the sample before extraction. This report provides valuable information although further evaluation of the influence of other extraction parameters, such as sample characteristics, ultrasound frequency and power and the use of ultrasonic pulses is still needed and will likely be explored in future investigations. Another sample preparation technique that can be used for extracting soy isoflavones from liquid foods is solid phase extraction. SPE involves adsorption of sample components on the surface of a solid sorbent, followed by elution with a selected solvent. A variety of sorbents available in the market allows not only the isolation of analytes, but also the removal of interferences. However, the whole potential of this technique for the analysis of isoflavones in foods is yet to be determined. Although SPE applications for the analysis of isoflavones from blood, plasma, urine and serum are relatively common [155–161], only a few works explored the SPE potential for the analysis of isoflavones from liquid samples. Mitani et al. [162] for instance, proposed an automated on-line in-tube solid phase microextraction (SPME) coupled to HPLC for the determination of daidzein and genistein in soy foods. In-tube SPME is a preconcentration technique using an open tubular fusedsilica capillary with an inner surface coating as the SPME device, which can be easily coupled on-line with HPLC. In tube SPME allows for convenient automation of the extraction process, which not only shortens the analysis time, but also provides better accuracy, precision and sensitivity relative to off-line manual techniques. However, a hydrolysis step was required because the isoflavone glucosides present in the sample were difficult to concentrate using such conditions, which limit its usefulness for quantification of all isoflavone chemical forms. Moreover, since most isoflavones are in the suspended solids in liquid samples, it is very likely that an extraction step before the in-tube SPME method will be required; otherwise it will only separate isoflavones present in the liquid phase and may lead to an underestimation of isoflavone concentrations. Therefore, the greatest potential of this technique is the concentration and clean-up of extracts coupled to an extraction technique such as PLE or SFE, or after the extraction procedure with one of the previously discussed methods. It also may be used before extraction to eliminate undesirable components of the sample and allowing a more selective extraction of target compounds.

5.3. Optimization of extraction conditions In view of the fact that chemical modifications of isoflavones may occur during the extraction process, not only isoflavone extraction efficiency for a particular solvent need to be considered when comparing the extraction solvents, but also preservation of original isoflavone composition during extraction, minimizing chemical transformations. Concentrations of ␤-glucosides and acetyl glucosides forms could be increased or decreased during extraction procedure, and aglycones could be increased as a consequence of chemical transformations. Higher amounts obtained of these chemical forms do not necessarily mean higher extraction efficiency since it could be result of the transformations. Thus, it is difficult to determine the most efficient solvent for extracting ␤-glucosides, acetyl glucosides or aglycone isoflavones by simple comparison of yields. Malonyl isoflavones are the chemical forms most susceptible to degradation and therefore the higher amount of malonyl glucosides present in the extracted material indicates either higher extraction efficiency of the solvent, better protection from chemical transformations or both. In the case of not quantifying malonyl isoflavones the best choice is to determine the stability during extraction. Apart of stability issues, extraction conditions should be optimized for each solvent and for each sample. Quantitative recoveries can be achieved with most commonly used solvents for the extraction of isoflavones from soybeans and soy foods, given extraction conditions are optimized enough. Therefore, the discussion of which is the best solvent should be addressed in terms of advantages and disadvantages. For example: MeCN sometimes can extract more isoflavones than MeOH or EtOH. However, MeCN have a higher cost, toxicity and lower environmental compatibility than EtOH and MeOH and if quantitative recoveries are achieved with these solvents there is no justification for the use of MeCN as extractant. Also, the influence of the sample should not be overlooked. It can seriously affect the ability of a given solvent to extract some isoflavones and use a method optimized with a certain sample should not be used without evaluation even for a similar sample. An internal standard can be used to account for losses during extraction for quantification studies. An ideal internal standard should be a compound structurally related to the analyte and with a similar polarity, but with a retention time that does not overlap the other peaks during the chromatographic analysis.

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

Spiking the sample may also be used to estimate if quantitative extractions are achieved, but enough time should be given for the standards interact with the sample. However, interaction with the sample matrix cannot be ensured even if prolonged soaking is used. Also, it may be of interest to use an extract obtained with optimized conditions rather than pure standards, since an extract more closely resemble the sample. Finally, sequential extractions are seriously recommended to ensure that quantitative recoveries are achieved [106,107,109,151]. Overall, to optimize extraction conditions three strategies can be used: one-variable-at–a-time optimization, the use of an experimental design, or a combination of both. The one at a time approach is more objective and can be used to isolate the effect of a given variable and provide easier interpretation. The drawback of this strategy is that it is somewhat limited to observe interactions among extraction variables. For instance, when optimizing an extraction using PLE, MeOH extracted higher amounts of isoflavones than EtOH at 60 ◦ C and 100 bar of pressure. However, if higher temperature or pressure was used, the physical–chemical properties of the solvents change and MeOH may not obtain higher yields than EtOH. To observe interactions, either a screening of all possible extraction conditions or an experimental design can be used. Screening implies a huge number of extractions, which takes time and have high cost. The experimental design allows reducing the number of analysis needed to identify the most important extraction variables. An excellent optimization strategy would be to use an experimental design to identify the most important variables, and further investigate these variables in detail by screening. Also, the use of mathematical models and computer aided optimization may be valuable tools for reducing the time required to develop extraction methods hampered by the existence of a great number of variables influencing the process and will likely explored in the future. 5.4. Critical comparison of extraction methods Serious efforts have been made in the last decade trying to compare techniques, methods and solvents for the extraction of soy isoflavones. A relative comparison of extraction techniques/ methods available in the literature is shown in Table 6. Ascertaining the most suitable extraction technique/method/ solvent for determination of isoflavones in soy samples is relatively difficult. In spite of the fact that the substances investigated are quite close chemically, physically and physiologically, there are important differences on the extractability of each chemical form and isoflavone type, and therefore it is almost impossible to suggest a single extraction solvent that ensures that all isoflavonoids are extracted with maximum yields from all types of soy samples [107,129]. It is also important to remark that, in most cases, what are being compared are extraction methods using different extraction techniques and not the extraction technique itself. Stability may also affect results and provide incorrect relative comparison values. Rostagno et al. [118], for example, developed an extraction method for the extraction of some isoflavones from soybean flour using SFE and compared the results obtained with different techniques (UAE and soxhlet). Highest yield of total isoflavones evaluated were obtained by UAE, followed by soxhlet and SFE methods. Soxhlet extracted 70%, and SFE extracted approximately 30% of the amount obtained by the UAE method, respectively. However, there are several factors that may responsible for the observed differences among extraction methods that may be attributed to the extraction method rather than to the technique itself. In this case, the most important is that stability was not evaluated and only a few isoflavone chemical forms were quantified. This is of

23

uttermost importance since using soxhlet extraction, degradation of isoflavone malonyl, acetyl and glucoside forms is known to take place increasing concentration of glucosides and aglycones and therefore results may not be comparable. Also, performing UAE for 90 min is very likely to increase extraction temperature and on similar basis and promote degradation of isoflavone conjugate forms. Therefore, these results should be taken with caution and the relative efficiency of SFE may be underestimated. The comparison of extraction methods can be very relative, especially regarding the method used as reference. Araújo et al. [92] compared SFE and stirring with for the extraction of some isoflavones from soybean hypocotyls. The proposed SFE method extracted 25.65% of the aglycones extracted by stirring. The relative amount extracted reported is overestimated since the hydrolysis was only applied to samples extracted by the SFE technique and not to the reference method (stirring). Hydrolysis markedly increases the concentration of aglycones present in the sample at the expense of the glucosides and hence increases the available amount of aglycones to be extracted and the actual yield may be much lower. Also, the necessary corrections when using hydrolytic methods were not applied which further influence results obtained. The authors go even further and claim the results obtained with the proposed method were superior to those obtained by Chandra et al. [145] and Rostagno et al. [118]. The yield of aglycones obtained by Araújo et al. [92] was 180 ␮g/g while Chandra et al. [145] extracted between 15 ␮g/g and 103 ␮g/g dry weight (depending of the sample) and Rostagno et al. [118] extracted 32.6 ␮g/g. A key point for the differences between these reports is also the hydrolysis of the extracts used by Araújo et al. [92]. Comparing yields of only aglycones between hydrolytic and non-hydrolytic methods is rather complicated and even more is the non-hydrolytic methods used as reference quantify only a few glucosides [118] or no glucosides at all [145]. If all chemical forms are quantified it is possible to compare hydrolytic and non-hydrolytic methods by making the corrections for the molecular mass. Moreover, different samples and sample types were used. Araujo et al. [92], extracted isoflavones from soy hypocotyls, Chandra et al. [145] from miso, tofu, soy meal and soy flour and Rostagno et al. [118], from soy flour. The concentration of a given compound on the sample directly affects the yields and comparison of different samples should be based on relative recoveries rather than in yield. Also, sample stability may be affecting the recoveries of the earlier reports, thus these results may not be comparable in the same basis. Therefore, authors should be very careful when making assumptions when comparing results with those obtained with other methods available in the literature. Kao et al. [91], compared SFE and shaking for the extraction of isoflavones from soybean cake and observed that shaking extracted higher amounts of total isoflavones (approximately 35%) than the highest amounts obtained with SFE (60 ◦ C/350 bar). However, SFE extracted higher amounts of acetyl glucosides and aglycones. At 60 ◦ C/350 bar SFE extracted approximately 33% and 91% of malonyl and glucosides, respectively than the amount obtained with shaking, while shaking extracted 80% and 87% of the acetyl and aglycones, extracted by SFE. At 80 ◦ C/350 bar, SFE extracted even lesser malonyl and glucosides (17% and 70%, respectively) than shaking, and shaking extracted even less acetyl and aglycones (65% and 58%, respectively) than SFE. These observations may have been influenced by degradation of malonyl isoflavones and without proper evaluation of isoflavone stability results might give a false impression of effectiveness. Also, the relative yield obtained with the developed method (65%) is much higher than those obtained by Rostagno et al. [118] (28%) and Araújo et al. [92] (26%). However, the relative yields reported by Rostagno et al. [118] (28%) and Araújo et al. [92] were

24

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

Table 6 Relative comparison of extraction techniques/methods. Sample

Isoflavones

Compared techniques/methods

Relative yield (%)a

Reference

Soy flour Soybean cake

Gi, Ge and De Di, Gi, Gly, De, Ge, Gle, MDi, MGi, MGly, ADi, AGi and AGly De and Ge Di, Gi, De and Ge Di, Gi, Gly and MGi Di, Gi, Gly, De, Ge, Gle, MDi, MGi, MGly, ADi, AGi and AGly Di, Gi, De and Ge Di, Gi, Gly, Ononin, De, Gle and Ge Gi, MGi, AGi and Ge

SFE/UAE/Soxhlet SFE/Shaking

28/100/68 74/100

[118] [92]

SFE/Stirring SFE/Stirring UAE/Stirring PLE/UAE/Soxhlet/Shaker/Vortex/Stirring

26/100 87/100 100/85–100b 100/93/68/71/66/70

[93] [147] [106] [134]

PLE/UAE/Soxhlet/PLE + UAE UAE/Soxhlet/PLE + UAE PLE/Stirring

49/14/64/100 22/68/100 98–100/88–100c

[135] [137] [133]

UAE/UHOM/SFE/PLE/Soxhlet MAE/UAE

100/93/16/71/69 100/100

[129] [110]

Soybean hypocotyls Soybean meal Soybeans Soybeans Soy bits Soy bits Soy flour, Meat substitute, nuts and protein isolate Soy flour Soy flour

De and Ge Di, Gi, Gly, De, Ge, Gle, MDi, MGi, MGly, ADi, AGi and AGly

De: daidzein, Ge: genistein, Gle: glycitein, Di: daidzin, Gi: genistin, Gly: glycitin, MDi: malonyl daidzin, MGi: malonyl genistin, MGly: malonyl glycitin, ADi: acetyl daidzin, AGi: acetyl genistin, AGly: acetyl glycitin, UHOM, Ultrasonic homogenizer. a Relative to the technique which extracted the highest amount of total isoflavones. b Depending of the solvent used. c Depending of the sample used.

influenced by degradation during extraction with the reference method and by the use of hydrolytic methods. Since stability was not accessed by Kao et al. [91] the relative yield, although much higher than those reported earlier, is only speculative. Zuo et al. [147], achieved even higher relative recoveries (87%) than Kao et al. [91] when compared to solvent extraction using magnetic stirring using different extraction conditions. In this case, only a few isoflavones were quantified (glucosides and aglycones) and results were reported in only in terms of total isoflavones and the issues of the previous discussed method apply. Stability during extraction may be one of the most important aspects when comparing extraction techniques/methods since they may change isoflavone profile of the sample and affect yields, which is especially important when all isoflavone chemical forms are not quantified. For the reliable comparison of extraction techniques, extraction conditions should be optimized for each technique, ensure that using the optimized conditions do not affect isoflavone profile and only then, results might be comparable. One option to obtain comparable results between methods/techniques is the use of hydrolytic methods, including the reference method, since it will eliminate variations derived from transformations. The handicap of using hydrolysis is that only limited information is obtained (i.e. total isoflavones) although it may prove useful in some cases. Several other authors tried to compare different extraction techniques/methods. Rostagno et al. [106] compared UAE and magnetic stirring using several different solvents (water, EtOH, MeOH and MeCN with different water percentages) at 10 ◦ C for 10 min and observed that UAE extracted between 0% and 15% more isoflavones than magnetic stirring at 10 ◦ C, depending of the solvent. At 60 ◦ C similar increase in extraction efficiency was observed. Also, a similar solvent response was observed using magnetic stirring and UAE, achieving maximum extraction efficiency using solvents with 40–60% water. Luthria et al. [134] compared several extraction techniques (stirring, shaker, UAE, vortexing, soxhlet and PLE) of 12 main isoflavones from soybeans using the same solvent (DMSO:EtOH:water (5:70:25) as extracting solvent. PLE was the most effective method for the extraction of total isoflavones, extracting between 30% and 35% more isoflavones than the other methods. Total isoflavones extracted by UAE was 93.3% as compared to PLE and shaking extracted 75.6% of amount extracted by UAE. Both, the total isoflavone content and the isoflavone HPLC profile varied significantly with different techniques. MGly and De were detected only

on PLE and UAE extracts. Shaking and stirring extracted the highest amounts of malonyl isoflavones (MDi and MGi) while PLE extracted the highest amount of acetyl glucosides. Extraction conditions (sample size, extraction length, number of extraction cycles and temperature) used in the PLE procedure were point by point optimized by Rostagno et al. [109] while the other extraction methods were not, and therefore is not surprising that PLE revealed to be the most effective extraction technique. Klejdus et al. [135] evaluated different techniques/methods (PLE, UAE, soxhlet and PLE + UAE) for the extraction of isoflavones (De, Ge, Di and Gi) from soybean foods. PLE + UAE (1 min sonication) extracted the highest amount of isoflavones followed by soxhlet, PLE and UAE, in this order. Soxhlet extracted the highest amount of aglycones while PLE + UAE extracted the highest amount of glucosides. However, malonyl isoflavones were not quantified and stability was not accessed and the highest amount of isoflavones extracted by PLE, PLE + UAE and soxhlet than by UAE alone may be partially attributed to degradation of malonyl isoflavones leading to their respective glucoside and aglycone forms. Later, Klejdus et al. [137] evaluated soxhlet, UAE and PLE + UAE for the extraction of isoflavones from soy bits and obtained similar results. In this case, however, sonication time before PLE was 5 min instead of 1 min. The same stability issues of the later report also apply here. Downing et al. [133], compared PLE and the method developed by Barnes et al. [78] using stirring. The method by Barnes et al. [78] required 60 min, while the PLE procedure (performed at 80 ◦ C) required 20 min to extract similar levels of genistein equivalents. They observed significant differences on the extraction of conjugated forms of genistein extracted by these two methods. Heat during PLE caused significantly less acetyl genistin to be present in the extracts when compared with stirring where ambient temperature was used. However, this outcome was dependent of the sample. In some soy flour samples deesterification occurred and in others not. Acetyl genistin was much more susceptible to degradation than malonyl genistin and degradation of the former only occurred in one sample (soy nuts). The change in the forms of genistein was attributed to heat-induced deesterification of the acetylgenistin and malonyl genistin to genistin. Bajer et al. [129] also evaluated different extraction methods (UAE, ultrasonic homogenizer, SFE, PLE and soxhlet) for the extraction of De and Ge from soy flour using optimized conditions. They observed that the different isoflavones present in the assayed

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

samples are extracted in maximum yields by different methods. The highest amount of these two isoflavones was obtained by UAE, followed by ultrasonic homogenizer, PLE, soxhlet and SFE, in this order. No significant difference was observed between PLE and soxhlet. For De, the best extraction methods were UAE and soxhlet followed by ultrasonic homogenizer, PLE and SFE, in this order, while for Ge the extraction methods with highest yields were ultrasonic homogenizer, UAE, PLE, soxhlet and SFE, in this order. However, these results should be taken with care. Stability was not assessed nor was the malonyl and glucoside forms quantified. Soxhlet extraction is known to promote degradation of these forms and nevertheless, extracted low amounts of Ge (more than half of the amount extracted by ultrasonic homogenizer). Moreover, no data on the optimization of each extraction method was provided and therefore these may be rather speculative. There is a fundamental difference when comparing the extraction of a given compound, comparing the extraction technique (UAE, PLE, etc.) and the extraction method. Using the same technique is possible to have two different quantitative extraction methods. In most cases, authors compare different extraction techniques using different extraction methods and therefore to draw conclusions from this kind of reports is difficult. If the aim of the study is to optimize an extraction method using a certain technique and make a comparison with other methods (that use different extraction techniques), it is essential that authors, use optimized conditions for all extraction methods and ensure that they do not affect isoflavone profile of the sample. Comparison with a reference method reported in the literature may also be used. An illustrative example for the comparison of the different extraction methods/techniques can be taken from the work of Rostagno et al. [110]. These authors proposed an MAE method after optimizing several extraction conditions and evaluated isoflavone stability with the optimized method, which did not affect isoflavone profile in the sample. The MAE method was compared with a previous developed method using UAE and no difference was observed in total and individual isoflavone yields. With both methods, quantitative extractions were obtained in 20 min. In this case, not only are the methods comparable but also the extraction technique, which are similarly effective for the extraction of isoflavones from soybeans. Moreover, quantitative recoveries are achieved with both techniques without changing the isoflavone profile of the sample. Altogether, SFE seems to be less efficient for extraction of isoflavones than other techniques, while UAE, PLE and MAE, the most efficient. Apart from extraction efficiency, there are other aspects that are important when determining the most suitable extraction technique. Selection of an appropriate extraction technique entails consideration of not only the recovery but also the cost, time of extraction, and the volume of solvent used, among others. From the point of view of solvent consumption, SFE is without doubt the best extraction technique. In contrast, soxhlet require large amounts of solvent and is a time-consuming procedure and may affect isoflavone profile. UAE and MAE have demonstrated to be fast techniques for the extraction of isoflavones from soybeans, followed closely by PLE. SFE is an intermediate option. Also, the need of post-extraction purification steps is an important issue. PLE and SFE provide sufficiently pure extracts without the need of subsequent filtration, while using UAE the filtration step is required. PLE has been shown to have important advantages over competing techniques as regards time saving, solvent use, automation and efficiency. PLE and SFE have the advantage that no filtration step is needed, since the matrix components that are not dissolved in the extraction solvent may be retained inside the sample cell. Also, it allows to easily performing reextractions of the sample to ensure quantitative extractions are achieved. PLE and SFE are very convenient for the purposes of automation and on-line coupling of the

25

extraction and separation techniques. Also, PLE and SFE offers the possibility of performing the extractions under an inert atmosphere and protected from light, which represents an attractive advantage since many compounds, are sensitive to these two external factors [67,109]. However, some modern extraction techniques (MAE, PLE and SFE) are not always available in the average laboratory, due to the high cost of the equipment. For the analysis of a particular sample with approximate knowledge of concentration and distribution of isoflavones, such as for routine quality control of similar soy flours, UAE can be used due to its low cost and high efficiency. In contrast, for the analysis of different samples with unknown isoflavone concentration and distribution, PLE may be preferred since besides high efficiency it allows to easily perform reextractions of the sample. Thus, the choice of an extraction technique will depend of several factors besides efficiency. Among these factors, implicit characteristic of the techniques are particularly relevant such as instrumental cost, level of automation and possibility of on-line coupling with analysis technique. A good way to select an appropriate extraction technique is to consider practical aspects and establish a multicriteria decision making procedure using desirability function optimization. 6. Post-treatment of extracts After extraction of isoflavones from the sample matrix is performed, the extract can be submitted to a series of post-treatment steps before the analysis. These procedures can be reduced to a minimal depending of extraction technique used. After extraction, insoluble materials are usually removed by filtration or centrifugation and sometimes, the extract are immediately analyzed without further preparation. If extract is obtained using PLE or SFE filtration and centrifugation is not required. Also, several authors simply pass the sample through 0.45 ␮m filters after extraction and avoid the centrifugation step [86,91,109,110,134,144,147,151]. The limitation of not using centrifugation is the difficulty of correcting the sample volume and solvent losses during filtration, especially with small samples. This problem can be prevented by using an internal standard with the specific aim of correcting the sample volume [81,106,109,110,151]. After filtration, liquid–liquid extraction can be used to remove undesired sample components such as the lipophilic components, in order to preserve reverse phase chromatographic columns. Hydrolysis of the extracts can also be used after the extraction using the same methods discussed in Section 4. Another common post-extraction procedure is the partial or complete removal of the solvent by rotary evaporation and redissolution of the sample either on the mobile phase used for the chromatographic analysis or in 80% MeOH [41,42,72,90,102,107,129, 133,135–137,144,152]. This procedure can be used to pre-concentrate the extracts and reduce detection and quantification levels during chromatographic analysis. Another reason for this post-extraction step is to avoid the peak distortion caused by injecting samples containing high concentration of MeCN onto columns equilibrated with low MeCN concentration. The procedure is time-consuming and such handling always increase variability and can be source of losses and degradation. Moreover, the use of this post-extraction step can be avoided by using a compatible solvent for extraction such as EtOH or MeOH, by limiting the sample size to less than 5 ␮L when using conventional C18 columns or by using high flow HPLC methods with monolithic columns [81,104]. Avoiding the use of such cumbersome procedure can greatly decrease the time required for sample preparation. More often, some additional sample preparation are used to isolate analytes of interest from other sample components that can

26

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

interfere with the chromatographic analysis or as an extract enrichment step, wherein the analyte concentration is increase above the determination limit of the final determination technique. SPE is one of the most used enrichment techniques. SFE have been used by some authors [52,129,163,164] to provide a clean concentrated isoflavone extract to be used in the chromatographic analysis. A wide selection of sorbents, ranging from classical C8 and C18 silica based sorbents to new polymeric materials, enables substantial selectivity of the enrichment process. Most traditional solid phase extraction sorbents often result in poor analyte recoveries, insufficient cleanup, or irreproducibility from extraction to extraction. Polymeric sorbents are the latest breakthrough in SPE, since they enable higher recoveries, higher reproducibility, and lower consumption of plant materials for the HPLC analysis than classical sorbents. Polymeric sorbents also have the advantage of remaining “conditioned” even if the sorbents accidentally run dry during the extraction. Divinylbenzene based polymeric sorbents exhibit excellent stability over the whole pH range unlike classical modified silica gel sorbents C18 and C8 [164]. Excellent results were obtained using polymeric sorbents for concentration and clean-up of isoflavones from red clover [164,165] and soy extracts [126]. For soy isoflavones, divinylbenzene based polymeric cartridges showed better retention and much higher breakthrough volume during sample load and washing steps than classical C18 sorbents from different manufactures. Besides of the use of new polymeric sorbents, the recent trend for the use of SPE is automation and coupling on-line with the analysis method. Compared with manual methods, automated SPE is less labour intensive, requires less sample handling providing better recovery, is more reproducible, is performed in a closed system (less chance of sample oxidation or solvent evaporation) and can be performed relatively fast. For instance, Rostagno et al. [126] developed an automated SPE method for soy isoflavones achieving very high recoveries (99.37%) and reproducibility (>98%) with a concentration factor of approximately 6:1 in less than 10 min. Another future prospect for the use of SPE is to be coupled on-line with the extraction (such as PLE and SFE) and analysis methods reducing to a minimum post-treatment of extracts in order reduce sample handling allowing more precise and reliable data to be obtained. 7. Separation approaches/techniques Many different analytical methods can be used for the analysis of isoflavones from soybeans and soy foods. These analytical methods include gas chromatography, liquid chromatography (both with and without mass detectors) capillary electromigration techniques (CE), and immunoassay. In the last few years several reviews about the analysis of isoflavone extracts using these techniques have been published [166–171]. Chromatography and CE are, without doubt, the most relevant techniques applied in this field. The use of CE for the analysis of soy isoflavone samples is very attractive due to the high resolution, efficiency and analysis speed with minimum reagent and sample consumption. There are a variety of versatile CE separation principles which are feasible of adapting to solve different analytical problems. The possibility of coupling CE to different types of detectors, especially to sensitive electrochemical detectors, is one of the main advantages of these techniques and point to a powerful tool for the characterization of isoflavones in soy derived samples. Although the use of CE in the identification process is a potentially appropriate means of rapid screening it has been applied only on a few occasions for the analysis of isoflavones from soy samples, and in most cases only some chemicals forms were identified [168,171–174].

However, CE is characterized by poor quantitative reproducibility, mainly caused by inconsistent flow rate and injection volume or amount. Although significant advances in this aspect has been made, the reproducibility issues of CE, especially when applied to real samples, still needs to be solved before it become a real alternative to more consolidated techniques such as chromatography [171,175]. In this context, of all available analysis techniques, HPLC is the method of choice since it requires simple pre-analysis sample preparation, allows measurement of all isoflavone chemical forms, is highly efficient and reproducible, is widely available and has been extensively studied. HPLC separation of isoflavones is generally carried out on reversed-phase columns with using MeOH or MeCN and water containing a small amount of acid (formic, acetic, phosphoric or trifluoroacidic acids) as mobile phase. Since isoflavones exhibit a weak acidic nature the use of acids can make the analytes to be easily dissociated in a solvent system enhancing chromatographic separation, resolution and improve peak shape [169]. Most often used detectors coupled to HPLC are UV and UV-diode array detection (DAD) monitoring in the range of 230–280 nm, since all isoflavones exhibit an intense absorption in this UV region of the spectrum. Gradient elution is usually necessary in order to separate all main isoflavones since they are very chemically close. As previously mentioned some isoflavones are particularly difficult to separate from each other (i.e. MGi, AGly and De) [81] and isocratic elution has proven to be insufficient. Isocratic elution may be accomplished if a hydrolysis step is used before analysis with the implicit handicap of quantifying only the aglycone forms [51]. Conventional microparticulate 5 ␮m RP-C18 columns are the most used stationary phase and analysis time needed to separate all main soy isoflavones usually reach 60 min. Similarly to sample preparation, the current trend for the analysis of soy isoflavone extracts is toward fast, high sensitive and high-resolution separation of all main chemical forms of these compounds in soybeans and soy foods. One alternative to achieve faster and more sensitive analysis is to reduce the particle size of the stationary phase. The use of smallparticle columns (less than 2 ␮m particle size) can shorten analysis times, while maintaining – or even increasing – high separation efficiencies, since it is very well known from Van Deemter equations that the efficiency of chromatographic processes is proportional to particle size decrease and to the higher allowed linear velocities. The negative aspect of small particle packed columns used in HPLC is the higher column back-pressure generated [176,177]. Hence, to take full advantage of sub-2 ␮m particles stationary phases high pressure liquid chromatography systems that operate at high pressures (>400 bar) are required. Not only is the system capacity of operating at high pressures important but also the ability to accurately and reproducibility integrate an analyte peak and detector sampling rate must be high enough to capture enough data points across the peak. Some applications of this innovative technology for isoflavones from different matrixes can be found in the literature [178–181]. Churchwell et al. [178] compared UPLC–MS conventional HPLC–MS for the determination of isoflavones in waste water and found that in general, UPLC–MS produced significant improvements in method sensitivity, speed, and resolution when compared to conventional HPLC–MS. Improvements in chromatographic resolution with UPLC were apparent from generally narrower peak and from a separation of diastereomers not possible using HPLC. As an example of the enormous potential of these new advances in chromatography, Klejdus et al. [179], developed an analysis method for some selected isoflavones (Di, Gly, Gi, De, Gle, Ge, Ononin, sissotrin, formononetin and biochanin) which takes less than 1 min. The method was successfully applied to soy bits and red

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

clover extracts with excellent results. Moreover, Klejdus et al. [180], accomplished simultaneous separation of not only isoflavones but also together with several phenolic acids in less than 2 min. Further investigations however, are still needed to evaluate the use of small particle columns for the separation of malonyl and acetyl isoflavones that are the most troublesome compounds to separate in soy extracts. Another alternative to perform high-speed separations using liquid chromatography is the use of monolithic columns. As compared to particle bed columns, monolithic columns represent a single piece made of porous cross-linked polymer or porous silica. Monoliths are made in different formats as porous rods, generated in thin capillaries or made as thin membranes or disks. The major goals of applying monolithic columns in HPLC were to achieve low column backpressure and fast mass transfer kinetics [182,183]. Major chromatographic features of monolithic silica columns arise from the large through-pore size/skeleton size ratios and high porosities, resulting in high permeability and large number of theoretical plates per unit pressure drop. High permeability and small diffusion path length provided by the presence of large throughpores and relatively small-sized skeletons resulted in the lower plate height and the lower pressure drop with monolithic silica columns compared with a particle-packed column. With lower column backpressure it is possible to increase solvent flow rate making faster separations possible with current instrumentation [182]. Monolithic columns have been used in some occasions for the analysis of isoflavones in soy extracts [81,110,126,151,184,185]. Apers et al. [184], achieved separation of most soy isoflavones (except malonyl glucosides) from soy extracts in less than 18 min. Further, Kim et al. [185] recently proposed another method using monolithic columns for the analysis of four isoflavones (Gi, Ge, Di and De) from soybeans and soybean pastes in 7 min. To date, the fastest separation of all main isoflavones in soy extracts was obtained by Rostagno et al. [81]. After optimization of analysis conditions using monolithic columns all isoflavones were resolved in less than 10 min using acidified MeOH–water at a flow rate of 4 mL/min. Such high solvent flow rate illustrates the low pressure obtained with this type of column. However, the use of MeOH in the mobile phase to perform fast separation has some limitations when compared to MeCN, since it has a higher viscosity resulting in higher pressure, which can reduce maximum solvent flow rate allowed within the chromatographic system maximum pressure and increase separation time required for all isoflavone chemical forms. Therefore, it is feasible to assume than even shorter analysis run times than 10 min can be achieved either with monolithic columns with MeCN or with small particle columns using UPLC systems and that more research is needed in this direction. To perform fast HPLC analysis, aside the chromatographic system, the stationary or mobile phase, the simplest approach consists in operating columns at higher temperatures. Mobile phase viscosities decrease rapidly with increasing temperature; the column efficiency is barely changed but the optimum velocity increases markedly, allowing the same resolution to be obtained much faster but with nearly the same inlet pressure [186]. In this case, some of the new small particle columns have a clear advantage over conventional 5 ␮m C18 and monolithic columns since they can operate at much higher temperatures, reaching 90 ◦ C depending of pH conditions. 8. Conclusions In the last two decades considerable efforts were directed to quantify isoflavones in soybeans and derived foods generating several sample preparation and analysis methods which resulted in

27

huge amounts of scattered information. Although a critical view of these methods is given in this review, is important to point out that they contributed to better understanding of the complex task that is isoflavone determination and that the necessary steps are being given in the direction of achieving reliable and precise information about the distribution of these compounds in foods. Several aspects of sample conservation and sample preparation for the determination of soy isoflavones should be carefully observed since they are an important source of misinformation. Sample conservation is particularly important since some isoflavones are relatively unstable and adequate storage conditions are necessary to preserve the original profile of these in soybeans and soy foods. Recent trends in sample preparation include automation, highthroughput performance, reduction in solvent volume and time, on-line coupling with analytical instruments and more importantly, reduction of sample manipulation. Although significant advances on the extraction of isoflavones using modern extraction techniques have been achieved, the full potential of the new technology available needs to be further explored. There is increasing evidence that a combination of extraction/clean-up techniques such UAE, PLE and SPE is the most promising application of the new available technology for the development of extraction methods for the determination of isoflavones. Using such combinations not only high-throughput performance and on-line coupling with the analysis instrument is possible, but also the reduction of post-extraction steps necessary before analysis. Moreover, the availability of new SPE sorbents may be easily used in the future to improve the performance of developed combined methods using actual available technologies. Another advantage of the use of PLE is the possibility of re-extraction of the same sample without manipulation which reduces analytical errors. This characteristic is very interesting since it may be recommendable to perform sequential extractions (up to 5 extractions of the same sample) in order to ensure that quantitative extractions are achieved. For the extraction procedure, the natural tendency is to use less toxic, expensive and environmental friendly solvents, such as ethanol, under optimized conditions that maximize extraction efficiency achieving fast and quantitative recoveries. Another technical tendency in sample preparation is to minimize pre and post-extraction steps in order to take full advantage of fast extraction and analysis procedures and to reduce analytical errors due to sample manipulation. This trend coupled to the increasing availability of commercial standards and higher sensitivity and resolution of new chromatography technology points toward the use of non-hydrolytic methods for the quantification of soy isoflavones. Furthermore, gathering full information about the distribution and concentration of all main soy isoflavone chemical forms present in the samples may be critical to understand the role of these compounds in preventing diseases. Similarly, for the determination of isoflavones the current trend is toward fast, high sensitive and high-resolution separation of all main chemical forms of these compounds in soybeans and soy foods. The development of new column packing technology and materials as well as of chromatographic systems that can operate at high pressures allows analysis time to be drastically reduced from the usual 60 min to a few minutes with outstanding performances showing that further advances can be made in analytical methodology currently used. Altogether, the optimization of sampling, sample preservation and sample preparation parameters are critical for accurate estimation of isoflavones present in soybeans and soy foods. Accurate estimation of isoflavones in soybeans and soy foods, as well as in other samples, will enable researchers to correctly evaluate the influence of such phytochemicals on health and provide precise

28

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29

dietary and safety guidelines on consumption of these compounds. In addition, accurate quantification of isoflavones in soybeans and soy foods will allow manufacturers, consumers, and marketing professionals to differentiate quality value-added products from the conventional ones. Most importantly, the scientific community, including journal editors and referees, must recognize the necessity of validating sampling, preservation and extraction as critical steps in describing the content of isoflavones in foods, in order to advance in the field of functional foods for both research and industry. Acknowledgment The authors acknowledge funding from the Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria (INIA) project AT07-003. References [1] J. Schwager, M.H. Mohajeri, A. Fowler, P. Weber, Curr. Opin. Biotechnol. 19 (2008) 66. [2] M.B. Roberfroid, Am. J. Clin. Nutr. 71 (2000) 1660S. [3] I. Goldberg, Functional Foods: Designer Foods, Pharmafoods Nutraceuticals, Chapman & Hall, New York, NY, 1994, p. 571. [4] K. Liu, Soybeans as Functional Foods and Ingredients, AOCS Press, Champaign, IL, 2004, p. 331. [5] S. Barnes, G. Peterson, C. Grubbs, K. Setchell, Adv. Exp. Med. Biol. 354 (1994) 135. [6] H. Adlercreutz, W. Mazur, Ann. Med. 29 (1997) 95. [7] F.H. Sarkar, Y. Li, Cancer Invest. 21 (2003) 744. [8] M.M. Lee, S.L. Gomez, J.S. Chang, M. Wey, R.T. Wang, A.W. Hsing, Cancer Epidemiol. Biomar. Prev. 12 (2003) 665. [9] E.H. Moriguchi, Y. Moriguchi, Y. Yamori, Clin. Exp. Pharmacol. Physiol. 31 (2004) S5. [10] R.S. Clair, M. Anthony, Handb. Exp. Pharmacol. 170 (2005) 301. [11] Y. Ikeda, M. Iki, A. Morita, E. Kajita, S. Kagamimori, Y. Kagawa, H. Yoneshima, J. Nutr. 136 (2006) 1323. [12] A.H. Wu, M.C. Yu, C.C. Tseng, M.C. Pike, Br. J. Cancer 98 (2008) 9. [13] Y. Birk, Int. J. Pept. Protein Res. 25 (1985) 113. [14] R.L. Anderson, W.J. Wolf, J. Nutr. 125 (1995) 581S. [15] K.M. Phillips, D.M. Ruggio, J.I. Toivo, M.A. Swank, A.H. Simpkins, J. Food Compos. Anal. 15 (2002) 123. [16] A. Romani, P. Vignolini, C. Galardi, C. Aroldi, C. Vazzana, D. Heimler, J. Agric. Food Chem. 51 (2003) 5301. [17] H. Wang, P.A. Murphy, J. Agric. Food Chem. 42 (1994) 1674. [18] J.A. Hoeck, W.R. Fehr, P.A. Murphy, G.A. Welke, Crop Sci. 40 (2000) 48. [19] S.J. Lee, W. Yan, J.K. Ahn, I.M. Chung, Field Crop Res. 81 (2003) 181. [20] E.J. Zhang, K.M. Ng, K.Q. Luo, J. Agric. Food Chem. 55 (2007) 6940. [21] M. Messina, O. Kucuk, J.W. Lampe, J. AOAC Int. 89 (2006) 1121. [22] T.C. Yeh, P.C. Chiang, T.K. Li, J.L. Hsu, C.J. Lin, S.W. Wang, C.Y. Peng, J.H. Guh, Biochem. Pharmacol. 73 (2007) 782. [23] K.D.R. Setchell, H. Adlercreutz, in: I.R. Rowland (Ed.), Role of Gut Flora in Toxicity and Cancer, Academic Press, San Diego, CA, 1988, p. 315. [24] L.W. Lissin, J.P. Cooke, J. Am. Coll. Cardiol. 35 (2000) 1403. [25] H. Adlercreutz, Environ. Toxicol. Pharm. 7 (1999) 201. [26] D.F. Birt, S. Hendrich, W. Wang, Pharmacol. Ther. 90 (2001) 157. [27] T.H. Kao, W.M. Wu, C.F. Hung, W.B. Wu, B.H. Chen, J. Agric. Food Chem. 55 (2007) 11068. [28] C. Steiner, S. Arnould, A. Scalbert, C. Manach, Br. J. Nutr. 99 (2008) ES78. [29] D. Ma, L. Qin, P. Wang, R. Katoh, Clin. Nutr. 27 (2008) 57. [30] P.S. Williamson-Hughes, B.D. Flickinger, M.J. Messina, M.W. Empie, Menopause 13 (2006) 831. [31] M.J. Messina, Nutr. Rev. 61 (2003) 117. [32] S. Banerjee, Y. Li, Z. Wang, F.H. Sarkar, Cancer Lett. 269 (2008) 226. [33] E. Walz, Justus Liebigs Ann. Chem. 489 (1931) 118. [34] M. Naim, B. Gestetner, I. Kirson, Y. Birk, A. Bondi, Phytochemistry 12 (1973) 169. [35] N. Ohta, G. Kuwata, H. Akahori, T. Watanabe, Agric. Biol. Chem. 44 (1980) 469. [36] N. Ohta, G. Kuwata, H. Akahori, T. Watanabe, Agric. Biol. Chem. 43 (1979) 1415. [37] E. Farmakalidis, P.A. Murphy, J. Agric. Food Chem. 33 (1985) 385. [38] S. Kudou, M. Shimoyamada, T. Imura, T. Uchida, K. Okubo, Agric. Biol. Chem. 55 (1991) 859. [39] S. Kudou, Y. Fleury, D. Welti, D. Magnolato, T. Uchida, K. Kitamura, K. Okubo, Agric. Biol. Chem. 55 (1991) 2227. [40] S.A. Bingham, C. Atkinson, J. Liggins, L. Bluck, A. Coward, Brit. J. Nutr. 79 (1998) 393. [41] P.A. Murphy, T. Song, G. Buseman, K. Barua, J. Agric. Food Chem. 45 (1997) 4635. [42] P.A. Murphy, T. Song, G. Buseman, K. Barua, G.R. Beecher, D. Trainer, J. Holden, J. Agric. Food Chem. 47 (1999) 2697.

[43] A.A. Franke, J.H. Hankin, M.C. Yu, G. Maskarinec, J. Agric. Food Chem. 47 (1999) 977. [44] S.T. Umphress, S.P. Murphy, A.A. Franke, L.J. Custer, C.L. Blitz, J. Food Compos. Anal. 18 (2005) 533. [45] M. Morton, O. Arisaka, A. Miyake, B. Evans, Environ. Toxicol. Phar. 7 (1999) 221. [46] M. Fukutake, M. Takahashi, K. Ishida, H. Kawamura, T. Sugimura, K. Wakabayashi, Food Chem. Toxicol. 34 (1996) 457. [47] S. Morandi, A. D’Agostina, F. Ferrario, A. Arnoldi, Eur. Food Res. Technol. 221 (2005) 84. [48] H. Wiseman, K. Casey, D.B. Clarke, K.A. Barnes, E. Bowey, J. Agric. Food Chem. 50 (2002) 1404. [49] J. Liggins, A. Mulligan, S. Runswick, S.A. Bingham, Eur. J. Clin. Nutr. 56 (2002) 961. [50] M.P. Prabhakaran, L.S. Hui, C.O. Perera, Food Res. Int. 39 (2006) 730. [51] L.S. Hutabarat, H. Greenfield, M. Mulholland, J. Food Compos. Anal. 14 (2001) 43. [52] M.I. Genovese, M.F. Lajolo, J. Agric. Food Chem. 50 (2002) 5987. [53] L.U. Thompson, B.A. Boucher, Z. Liu, M. Cotterchio, N. Kreiger, Nutr. Cancer 54 (2006) 184. [54] U.S. Department of Agriculture, Agricultural Research Service, USDA-Iowa State University Database on the Isoflavone Content of Foods, Release 1.42007. Nutrient Data Laboratory, 2007, Web site: http://www.ars.usda.gov/ nutrientdata. [55] Venus database, Vegetal Estrogens in Nutrition and the Skeleton (Venus) project, http://www.venus-ca.org. [56] M. Kiely, M. Faughnan, K. Wähälä, H. Brants, A. Mulligan, Br. J. Nutr. 89 (2003) S19. [57] Finland National Food Composition Database (Fineli® ), http://www.ktl.fi/ fineli. [58] L.M. Valsta, A. Kilkkinen, W. Mazur, T. Nurmi, A.M. Lampi, M.L. Ovaskainen, T. Korhonen, H. Adlercreutz, P. Pietinen, Br. J. Nutr. 89 (2003) S31. [59] M.R. Ritchie, J.H. Cummings, M.S. Morton, C. Michael Steel, C. Bolton-Smith, A.C. Riches, Br. J. Nutr. 95 (2006) 204. [60] K. Reinli, G. Block, Nutr. Cancer 26 (1996) 123. [61] P.C. Pillow, C.M. Duphorne, S. Chang, J.H. Contois, S.S. Strom, M.R. Spitz, S.D. Hursting, Nutr. Cancer 33 (1999) 3. [62] W. Mazur, Baillieres Clin. Endocrinol. Metab. 12 (1998) 729. [63] N.M. Saarinen, C. Bingham, S. Lorenzetti, A. Mortensen, S. Mäkelä, P. Penttinen, I.K. Sørensen, L.M. Valsta, F. Virgili, G. Vollmer, A. Wärri, O. Zierau, Genes Nutr. 1 (2006) 143. [64] D.L. Luthria, J. Sci. Food Agric. 86 (2006) 2266. [65] R.M. Smith, J. Chromatogr. A 1000 (2003) 3. [66] J. Pawliszyn, Anal. Chem. 75 (2003) 2543. [67] R. Carabias-Martínez, E. Rodríguez-Gonzalo, P. Revilla-Ruiz, J. HernándezMéndez, J. Chromatogr. A 1089 (2005) 1. [68] S. Nyiredy, J. Chromatogr. B 812 (2004) 35. [69] C.J.C. Jackson, J.P. Dini, C. Lavandier, H.P.V. Rupasinghe, H. Faulkner, V. Poysa, D. Buzzell, S. DeGrandis, Process Biochem. 37 (2002) 1117. [70] A.H. Simonne, M. Smith, D.B. Weaver, T. Vail, S. Barnes, C.I. Wei, J. Agric. Food Chem. 48 (2000) 6061. [71] L. Coward, M. Smith, M. Kirk, S. Barnes, Am. J. Clin. Nutr. 68 (1998) 1486S. [72] H.J. Wang, P.A. Murphy, J. Agric. Food Chem. 44 (1996) 2377. [73] I.U. Grun, K. Adhikari, C. Li, Y. Li, B. Lin, J. Zhang, L.N. Fernando, J. Agric. Food Chem. 49 (2001) 2839. [74] S.J. Lee, I.M. Chung, J.K. Ahn, J.T. Kim, S.H. Kim, S.J. Hahn, J. Agric. Food Chem. 51 (2003) 3382. [75] J.J. Kim, S.H. Kim, S.J. Hahn, I.M. Chung, Food Res. Int. 38 (2005) 435. [76] H.J. Hou, K.C. Chang, J. Food Sci. 67 (2002) 2083. [77] B. Eisen, Y. Ungar, E. Shimoni, J. Agric. Food Chem. 22 (2003) 2212. [78] S. Barnes, M. Kirk, L. Coward, J. Agric. Food Chem. 42 (1994) 2466. [79] M.A. Rostagno, M. Palma, C.G. Barroso, Food Chem. 93 (2005) 557. [80] E. Rijke, A. Zafra-Gómez, F. Ariese, U.A.Th. Brinkman, C. Gooijer, J. Chromatogr. A 932 (2001) 55. [81] M.A. Rostagno, M. Palma, C.G. Barroso, Anal. Chim. Acta 582 (2007) 243. [82] P. Delmonte, J. Perry, J.I. Rader, J. Chromatogr. A 1107 (2006) 59. [83] L.S. Hutabarat, H. Greenfield, M. Mulholland, J. Chromatogr. A 886 (2000) 55. ˜ [84] J.L. Penalvo, T. Nurmi, H. Adlercreutz, Food Chem. 87 (2004) 297. [85] M. Careri, L. Elviri, A. Mangia, Chromatographia 54 (2001) 45. [86] M. Careri, C. Corradini, L. Elviri, A. Mangia, J. Chromatogr. A 1152 (2007) 274. [87] S.P. Klump, M.C. Allred, J.L. MacDonald, J.M. Ballam, J. AOAC Int. 84 (2001) 1865. [88] W.M. Mazur, T. Fotsis, K. Wähälä, S. Ojala, A. Salakka, H. Adlercreutz, Anal. Biochem 233 (1996) 169. [89] W.M. Mazur, J.A. Duke, K. Wahala, S. Rasku, H. Adlercreutz, J. Nutr. Biochem. 9 (1998) 193. [90] J. Liggins, L.J.C. Bluck, W.A. Coward, S.A. Bingham, Anal. Biochem. 264 (1998) 1. [91] T. Kao, J. Chien, B. Chen, Food Chem. 107 (2008) 1728. [92] J.M.A. Araújo, M.V. Silva, J.B.P. Chaves, Food Chem. 105 (2007) 266. [93] A.A. Franke, L.J. Custer, C.M. Cerna, K.K. Narala, J. Agric. Food Chem. 42 (1994) 1905. [94] I. Rowland, M. Faughnan, L. Hoey, K. Wähälä, G. Williamson, A. Cassidy, Br. J. Nutr. 89 (2003) S45. [95] I.L. Nielsen, G. Williamson, Nutr. Cancer 57 (2007) 1.

M.A. Rostagno et al. / J. Chromatogr. A 1216 (2009) 2–29 [96] C.E. Rüfer, A. Bub, J. Möseneder, P. Winterhalter, M. Stürtz, S.E. Kulling, Am. J. Clin. Nutr. 87 (2008) 1314. [97] A.C. Eldridge, J. Agric. Food Chem. 30 (1982) 353. [98] A.C. Eldridge, W.F. Kwolek, J. Agric. Food Chem. 31 (1983) 394. [99] P.A. Murphy, J. Chromatogr. A 211 (1981) 166. [100] P.A. Murphy, E. Farmakalidis, L.D. Johnson, Food Chem. Toxicol. 20 (1982) 315. [101] T. Song, K. Barua, G. Buseman, P.A. Murphy, Am. J. Clin. Nutr. 68 (1998) 1474S. [102] P.A. Murphy, K. Barua, C.C. Hauck, J. Chromatogr. B 777 (2002) 129. [103] H. Tsai, L. Huang, Y. Lai, J. Chang, R. Lee, R.Y. Chiou, J. Agric. Food Chem. 55 (2007) 7712. [104] A.P. Griffith, M.W. Collison, J. Chromatogr. A 913 (2001) 397. [105] F. Lin, M.M. Giusti, J. Agric. Food Chem. 53 (2005) 3795. [106] M.A. Rostagno, M. Palma, C.G. Barroso, J. Chromatogr. A 1012 (2003) 119. [107] A. Achouri, J.I. Boye, D. Belanger, Food Res. Int. 38 (2005) 1199. [108] A. Papadopoulou, R.A. Frazier, Trends Food Sci. Technol. 15 (2004) 186. [109] M.A. Rostagno, M. Palma, C.G. Barroso, Anal. Chim. Acta 522 (2004) 169. [110] M.A. Rostagno, M. Palma, C.G. Barroso, Anal. Chim. Acta 588 (2007) 274. [111] A.H.W. Toebes, V. de Boer, J.A.C. Verkleij, H. Lingeman, W.H. Ernst, J. Agric. Food Chem. 53 (2005) 4660. [112] M. Matsuura, A. Obata, J. Food Sci. 58 (1993) 144. ´ [113] G. Romanik, E. Gilgenast, A. Przyjazny, M. Kaminski, J. Biochem. Biophys. Methods 70 (2007) 253. [114] C.W. Huie, Anal. Bioanal. Chem. 373 (2002) 23. [115] D.S. Junior, F.J. Krug, M.G. Pereira, M. Korn, Appl. Spectrosc. Rev. 41 (2006) 305. [116] F. Priego-Capote, M.D. Luque de Castro, TrAC-Trend Anal. Chem. 23 (2004) 644. [117] K. Vilkhu, R. Mawson, L. Simons, D. Bates, Innov. Food Sci. Emerg. Technol. 9 (2008) 161. [118] M.A. Rostagno, J.M.A. Araújo, D. Sandi, Food Chem. 78 (2002) 111. [119] A. Escarpa, M.C. González, A.J. Blasco, M.C. Rogerio, M. Hervás, Electroanalysis 19 (2007) 952. [120] H. Xu, Y. Zhang, C. He, Chinese J. Chem. Eng. 15 (2007) 861. [121] M. Lee, C. Lin, Food Chem. 105 (2007) 223. [122] J. He, Z. Zhao, Z. Shi, M. Zhao, Y. Li, W. Chang, J. Agric. Food Chem. 53 (2005) 518. [123] H.B. Xiao, M. Krucker, K. Albert, X.M. Liang, J. Chromatogr. A 1032 (2004) 117. [124] Y.C. Chukwumah, L.T. Walker, M. Verghese, M. Bokanga, S. Ogutu, K. Alphonse, J. Agric. Food Chem. 55 (2007) 285. [125] T. Bo, K.A. Li, H. Liu, Anal. Chim. Acta 458 (2002) 345. [126] M.A. Rostagno, M. Palma, C.G. Barroso, J. Chromatogr. A 1076 (2005) 110. [127] D. Yuan, Y. Pan, Y. Chen, T. Uno, S. Zhang, Y. Kano, Chem. Pharm. Bull. 56 (2008) 1. [128] L. Kuo, W. Cheng, R. Wu, C. Huang, K. Lee, Appl. Microbiol. Biotechnol. 73 (2006) 314. [129] T. Bajer, M. Adam, L. Galla, K. Ventura, J. Sep. Sci. 30 (2007) 122. [130] M. Romdhane, C. Gourdon, Chem. Eng. J. 87 (2002) 11. ˜ [131] J.A. Mendiola, M. Herrero, A. Cifuentes, E. Ibanez, J. Chromatogr. A 1152 (2007) 234. [132] O. Sticher, Nat. Prod. Rep. 25 (2008) 517. [133] J.M. Downing, O.K. Chung, P.A. Seib, J.D. Hubbard, Cereal Chem. 84 (2007) 44. [134] D.L. Luthria, R. Biswas, S. Natarajan, Food Chem. 105 (2007) 325. ˇ Anal. [135] B. Klejdus, R. Mikelová, V. Adam, J. Zehnálek, J. Vacek, R. Kizek, Vl. Kubán, Chim. Acta 517 (2004) 1. [136] B. Klejdus, R. Mikelová, J. Petrlová, D. Potˇesˇ il, V. Adam, M. Stiborová, P. Hodek, ˇ J. Agric. Food Chem. 53 (2005) 5848. J. Vacek, R. Kizek, V. Kubán, [137] B. Klejdus, R. Mikelová, J. Petrlová, D. Potˇesˇ il, V. Adam, M. Stiborová, P. Hodek, ˇ J. Chromatogr. A 1084 (2005) 71. J. Vacek, R. Kizek, V. Kubán, ˇ J. Chromatogr. [138] B. Klejdus, J. Vacek, J. Zehnálek, R. Kizek, L. Trnkova, V. Kubán, B 806 (2004) 101. [139] C. Li-Hsun, C. Ya-Chuan, C. Chieh-Ming, Food Chem. 84 (2004) 279. [140] L. Chang, C.J. Chang, J. Chinese Inst. Chem. Eng. 38 (2007) 313. [141] C.S. Kaiser, H. Römpp, P.C. Schmidt, Pharmazie 56 (2001) 907.

29

[142] Q. Lang, C.M. Wai, Talanta 53 (2001) 771. [143] B. Diaz-Reinoso, A. Moure, H. Dominguez, J.C. Parajo, J. Agric. Food Chem. 54 (2006) 2441. ˇ J. Sep. [144] B. Klejdus, L. Lojková, O. Lapcík, R. Koblovská, J. Moravcová, V. Kubán, Sci. 28 (2005) 1334. [145] A. Chandra, M.G. Nair, Phytochem. Anal. 4 (1996) 259. [146] J. Yu, Y. Liu, A. Qiu, X. Wang, LWT - Food Sci. Technol. 40 (2007) 800. [147] Y.B. Zuo, A.W. Zeng, X.G. Yuan, K.T. Yu, J. Food Eng. 89 (2008) 384. [148] C.S. Eskilsson, E. Björklund, J. Chromatogr. A 902 (2000) 227. [149] J. Song, S. Mo, Y. Yip, C. Qiao, Q. Han, H. Xu, J. Sep. Sci. 30 (2007) 819. [150] Z. Guo, Q. Jin, G. Fan, Y. Duan, C. Qin, W. Minjie, Anal. Chim. Acta 436 (2001) 41. [151] M.A. Rostagno, M. Palma, C.G. Barroso, Anal. Chim. Acta 597 (2007) 265. [152] S. Jung, P.A. Murphy, I. Sala, Food Chem. 111 (2008) 592. [153] T. Nguyenle, E. Wang, A.P. Cheung, J. Pharmaceut. Biomed. 14 (1995) 221. [154] K.D.R. Setchell, S.J. Cole, J. Agric. Food Chem. 51 (2003) 4146. [155] N.C. Twaddle, M.I. Churchwell, D.R. Doerge, J. Chromatogr. B 777 (2002) 139. [156] S. Moors, M. Blaszkewicz, H.M. Bolt, G.H. Degen, Mol. Nutr. Food Res. 51 (2007) 787. [157] Q. Wang, X. Li, S. Dai, L. Ou, X. Sun, B. Zhu, F. Chen, M. Shang, H. Song, J. Chromatogr. B 863 (2008) 55. [158] X. Zhang, Y.G. Sun, M.C. Cheng, Y.Q. Wang, H.B. Xiao, X.M. Liang, Anal. Chim. Acta 602 (2007) 252. [159] D.R. Doerge, M.I. Churchwell, K.B. Delclos, Rapid Commun. Mass Spectrom. 14 (2000) 673. [160] P.B. Grace, J.I. Taylor, N.P. Botting, T. Fryatt, M.F. Oldfield, S. Bingham, Anal. Biochem. 315 (2003) 114. [161] S. Chan, S. Lin, K. Yu, T. Liu, M. Fuh, Talanta 69 (2006) 952. [162] K. Mitani, S. Narimatsu, H. Kataoka, J. Chromatogr. A 986 (2003) 169. [163] J. Wang, P. Sporns, J. Agric. Food Chem. 48 (2000) 5887. [164] B. Klejdus, D. Vitamvásová-Sterbová, V. Kubán, Anal. Chim. Acta 450 (2001) 81. [165] B. Klejdus, D. Vitamvásová-Sterbová, V. Kubán, J. Chromatogr. A 839 (1999) 261. [166] H.M. Merken, G.R. Beecher, J. Agric. Food Chem. 48 (2000) 577. [167] P. Delmonte, J.I. Rader, J. AOAC Int. 89 (2006) 1138. [168] J. Vacek, B. Klejdus, L. Lojková, V. Kubán, J. Sep. Sci. 31 (2008) 31. [169] S. Dentith, B. Lockwood, Curr. Opin. Clin. Nutr. 11 (2008) 242. [170] Q. Wu, M. Wang, J.E. Simon, J. Chromatogr. B 812 (2004) 325. ˜ [171] M. Herrero, E. Ibánez, A. Cifuentes, J. Sep. Sci. 28 (2005) 833. [172] Y. Peng, Q. Chu, F. Liu, J. Ye, Food Chem. 87 (2004) 135. [173] O. Mellenthin, R. Galensa, J. Agric. Food Chem. 47 (1999) 594. [174] J.A. Starkey, Y. Mechref, C.K. Byun, R. Steinmetz, J.S. Fuqua, O.H. Pescovitz, M.V. Novotny, Anal. Chem. 74 (2002) 5998. [175] J.P. Schaeper, M.J. Sepaniak, Electrophoresis 21 (2000) 1421. [176] L. Nováková, L. Matysová, P. Solich, Talanta 68 (2006) 908. [177] S.A.C. Wren, P. Tchelitcheff, J. Chromatogr. A 1119 (2006) 140. [178] M.I. Churchwell, N.C. Twaddle, L.R. Meeker, D.R. Doerge, J. Chromatogr. B 825 (2005) 134. ´ O. Lapˇcík, V. Kubán, ˇ Anal. Bioanal. [179] B. Klejdus, J. Vacek, L. Beneˇsová, J. Kopecky, Chem. 389 (2007) 2277. ˇ J. Chromatogr. A 1195 [180] B. Klejdus, J. Vacek, L. Lojková, L. Beneˇsová, V. Kubán, (2008) 52. [181] M. Farré, M. Kuster, R. Brix, F. Rubio, M. López de Alda, D. Barceló, J. Chromatogr. A 1160 (2007) 166. [182] K.K. Unger, R. Skudas, M.M. Schulte, J. Chromatogr. A 1184 (2008) 393. [183] A.M. Siuoffi, J. Chromatogr. A 1000 (2003) 801. [184] S. Apers, T. Naessens, K. Van Den Steen, F. Cuyckens, M. Claeys, L. Pieters, A. Vlietinck, J. Chromatogr. A 1038 (2004) 107. [185] W.C. Kim, S.H. Kwon, I.K. Rhee, J.M. Hur, H.H. Jeong, S.H. Choi, K.B. Lee, Y.H. Kang, K.S. Song, Food Sci. Biotechnol. 15 (2006) 24. [186] G. Guiochon, J. Chromatogr. A 1168 (2007) 101.