Supercritical fluid extraction and convergence chromatographic determination of parthenolide in Tanacetum parthenium L.: Experimental design, modeling and optimization

J. of Supercritical Fluids 95 (2014) 84–91

Contents lists available at ScienceDirect

The Journal of Supercritical Fluids journal homepage: www.elsevier.com/locate/supflu

Supercritical fluid extraction and convergence chromatographic determination of parthenolide in Tanacetum parthenium L.: Experimental design, modeling and optimization Krisztina Végh ∗ , Ágnes Alberti, Eszter Riethmüller, Anita Tóth, Szabolcs Béni, Ágnes Kéry Semmelweis University, Department of Pharmacognosy, Ülloi ˝ str. 26, 1085 Budapest, Hungary

a r t i c l e

i n f o

Article history: Received 2 June 2014 Received in revised form 29 July 2014 Accepted 31 July 2014 Available online 8 August 2014 Keywords: Supercritical fluid extraction Experimental design Response surface methodology Tanacetum parthenium L. Convergence chromatography

a b s t r a c t Feverfew (Tanacetum parthenium L., Asteraceae) is a perennial medicinal plant which has been used to alleviate the symptoms of migraines, headaches and rheumatoid arthritis. The herb contains various potentially active constituents such as sesquiterpene-␥-lactones, flavonoids and volatile oil. The main sesquiterpene-lactone in feverfew is parthenolide which is considered to be responsible for the therapeutical effects. Supercritical CO2 extraction was carried out at different pressures (10–30 MPa), temperatures (40–80 ◦ C) and co-solvent contents (0–10% ethanol) in order to study the extraction yield and the parthenolide recovery of the extracts. Leaves collected before and during flowering and flower heads were investigated. A factorial experiment using a full 33 design was followed during the experiments and response surface methodology was implemented to analyze the influence of the variables and optimize the extraction. The critical values of parthenolide content were found to be 7% EtOH, 22 MPa and 64 ◦ C in case of all three samples. It was determined, that the optimal conditions of the extraction, where the maximum parthenolide content and extract yield can be reached, do not coincide. The highest yield of parthenolide was obtained in the flower heads (0.604 wt.%). © 2014 Elsevier B.V. All rights reserved.

1. Introduction Feverfew (Tanacetum parthenium (L.) Schultz-Bip., Asteraceae) has been known for centuries as a medicinal and ornamental plant. Traditionally, the herb has been used as an antipyretic, from which its common name is derived. In the 1980s the pharmacological effect was proved in migraine prophylaxis, headache and rheumatoid arthritis [1–4]. According to our current knowledge, feverfew has a COX-2 and NF-kB inhibitory effect, suppresses the production of prostaglandin, thromboxane and leukotriene and inhibits the 5-HT release from platelets [5–9]. Sesquiterpene lactones, particularly, parthenolide (Fig. 1) being the main constituent with a proportion of 85%, have been associated with the antimigraine properties of the plant [1,3,4]. There is a wide variation of conventional methods for the extraction of parthenolide from feverfew, using an initial organic or aqueous phase. Several solvent systems were examined applying bottle stirring or Soxhlet extraction [10,11]. The highest content of

∗ Corresponding author. Tel.: +36 20 9562934. E-mail address: [email protected] (K. Végh). http://dx.doi.org/10.1016/j.supflu.2014.07.029 0896-8446/© 2014 Elsevier B.V. All rights reserved.

parthenolide was found in flower heads (1.38%), followed by leaves (0.95%) and with only 0.08% in stalks and 0.01% in roots [12]. Nowadays the novel solvents like supercritical carbon dioxide (SC-CO2 ) are highly desirable. SC-CO2 has been successfully employed for the extraction of medicinally active components from herbs and other vegetable matrices [13–19]. Several advantages of supercritical fluid extraction (SFE) are known, among those the selectivity which changes with the operating conditions and organic solvent free extracts of high quality. Several studies have been performed on feverfew with SFE technique. Smith and Burford [20] performed the extraction at 250 bar and 40 ◦ C on the aerial parts of the plant without separating it. The parthenolide content was found to be 0.8%. As co-solvents 4% acetonitrile and methanol were used in their model experiments, which doubled the yield of parthenolide. Kaplan et al. [21] compared the parthenolide yields obtained with CO2 at different pressures (100–300 atm, 100 ◦ C and no added co-solvent) and with conventional solvents. They found that the seeds contain very high level of parthenolide (1.134%) and supercritical fluid extraction is much more effective, than solvent extraction with chloroform. Kery et al. [22] optimized the extraction conditions on a larger scale equipment with respect to total extraction yield and the yield of parthenolide. They used a 32 fullfactorial experimental design, in order to find the optimum point of

K. Végh et al. / J. of Supercritical Fluids 95 (2014) 84–91

85

The seedlings were raised in Érd and collected before and during flowering. The blooming samples were manually separated to flowers and leaves. After grinding the size of the particles were between 40 and 50 mesh. Voucher specimens are deposited in the Department of Pharmacognosy, Semmelweis University, Budapest. Purity of the CO2 employed in the experiments was 99.5% and it was supplied by Linde (Hungary). Chemical solvents of both reagent grade and HPLC grade and the parthenolide standard (≥98%, HPLC purity) were purchased from Sigma-Aldrich (Germany). Fig. 1. Structure of parthenolide.

2.2. Analytical scale supercritical fluid extraction the extraction. Supercritical CO2 was used at pressures from 100 to 400 bar and at temperatures from 40 to 60 ◦ C. They found, that the optimal conditions for the extraction of parthenolide are 400 bar and 60 ◦ C (5.2%), which are the highest parameters of the factors. Cretnik et al. [23] investigated the extraction of parthenolide from feverfew flower heads on a laboratory scale apparatus at 200 and 300 bar and at temperatures 40, 60 and 80 ◦ C. SFE with two separators was employed to partially separate the extracted parthenolide from waxy substances. They also tested the thermal stability of parthenolide, which decreased by 13%, however this was a result of a 1 h manipulation of a reference compound solution at constant 80 ◦ C. Several other articles have been published about SFE modeling and experimental designs, which facilitate optimization of the conditions of the extraction. In recent years there is an uplift in using sub- or supercritical fluids in chromatography. Ultra-performance convergence chromatography integrates supercritical fluid chromatography and ultra performance liquid chromatography with all the advantages of both techniques, including reduced run time, less solvent consumption, drastically increased peak capacities and sensitivity. UPC2 technology is capable of analyzing a diverse range of compounds, including all constituents being present in a supercritical fluid extract, ranging from non-polar to polar, volatile or nonvolatile components. The aim of our work was to carry out a detailed experimental program, in order to evaluate the individual and combined effects of the SFE conditions (temperature, pressure, co-solvent content) on the extraction yield and parthenolide content of T. parthenium L. extracts. Although there have been several studies about supercritical fluid extraction of feverfew, this is the first time when all three factors are analyzed on three levels in a full factorial design on an analytical scale equipment. Ethanol was chosen as co-solvent, due to its non-toxic and environment friendly properties. Till now, the effect of ethanol as modifier of the supercritical CO2 on the parthenolide content has not been investigated. The aerial parts of the plant were separated into leaves collected before and during flowering and flower heads before extraction. The effect of the vegetation period on the accumulation of parthenolide in different parts of the plant was analyzed for the first time. For this purpose, a 33 full factorial experimental design was followed, in order to define the main factors and interactions influencing the responses. To quantify the parthenolide content of the extracts, an ultra-performance convergence chromatographic method was developed and validated. Up to now, no analysis has been performed with ultra-performance convergence chromatography neither on supercritical fluid extracts, nor on T. parthenium L. 2. Materials and methods 2.1. Materials T. parthenium L. seeds were collected from the medicinal herb collection of the National Botanical Garden in Vácrátót (Hungary).

Supercritical extractions were carried out using a system consisting of a Jasco PU-2080 CO2 pump with a Peltier cooling system for the delivery of carbon dioxide and a Jasco PU-2080 pump, linked to a Jasco MX-2080-32 Dynamic mixer for the addition of modifier. A 1 ml Jasco extraction vessel was mounted in the column position in a CO-2060 plus intelligent column thermostat. A Jasco BD-2080 back-pressure regulator was applied to adjust the pressure in the system. The 1 ml extraction vessel was packed with (ca. 0.5 g) plant material and the optimal parameters were investigated in regard to the amount and composition (parthenolide content) of the extract. The solute leaving the extractor was introduced through a pressurereducing valve, where the product was collected. The extractions were carried out for over 1 h, the solvent flow rate was 0.4 ml/min. The experimental conditions [temperature (T), pressure (P), and modifier (ethanol, EtOH)] are listed in Table 1. Each variable was studied on three levels.

2.3. Experimental design In this work the relationship between the measured responses and the individual and combined effects was analyzed with response surface methodology. In the extraction process a factorial experiment using a full 33 design was followed. The influence of temperature (from 40 to 80 ◦ C in 20 ◦ C increments), pressure (at three levels between 10 and 30 MPa) and modifier addition (0%, 5% and 10% ethanol) on the extraction yield and parthenolide content in the extracts of the different blooming stages and parts was studied. Two other independent variables, solvent flow rate (0.4 ml/min) and extraction time (1 h) were kept constant. Twenty seven experiments were performed for the Design of Experiment (DOE) and additional runs were performed to check the trueness and the reproducibility of the experiments and the model. Two additional runs were performed at the extraction optimum point, in case of the parthenolide content. The values of the factors, the extraction yield and parthenolide content of the extracts are shown in Table 1. The mathematical relationship between the three significant independent variables was formulated by the general quadratic polynomial. 2 Y = ˇ0 + ˇ1 XP + ˇ2 XT + ˇ3 XEtOH + ˇ11 XP2 + ˇ22 XT2 + ˇ33 XEtOH

+ ˇ12 XP XT + ˇ13 XP XEtOH + ˇ23 XT XEtOH ,

(1)

where Y is the response, ˇ0 is constant, ˇ1 (EtOH), ˇ2 (P) and ˇ3 (T) are the linear, ˇ11 , ˇ22 and ˇ33 are the quadratic coefficients and ˇ12 , ˇ13 and ˇ23 are the crossed coefficients. For the statistical analyses of the results Statistica software was used. The goodness of fit of the regression models was evaluated by using the coefficients of determination, R2 and their adjusted values. To assess the significant factors and interactions, analyses of variance was utilized.

86

K. Végh et al. / J. of Supercritical Fluids 95 (2014) 84–91

Table 1 Results of the SFE experimental design of Tanacetum parthenium L. samples. The lines marked with “valid” are the additional runsperformed to check the trueness and reproducibility. EtOH (%)

Pressure (MPa)

Temperature (◦ C)

Leaves before flowering Exp.

Extraction yield (g/100 g)

Parthenolide Exp. (g/100 g)

Extraction yield (g/100 g)

Parthenolide Exp. (g/100 g)

Extraction yield (g/100 g)

Parthenolide (g/100 g)

0 0 0 0 0 0 0 0 0 0 0 5 5 5 5 5 5 5 5 5 5 5 10 10 10 10 10 10 10 10 10 10 10

10 20 30 10 20 30 30 20 30 20 20 30 20 30 30 20 30 30 20 30 20 20 30 20 30 30 20 30 30 20 30 20 20

40 40 40 60 60 60 80 80 80 60 60 40 40 40 60 60 60 80 80 80 60 60 40 40 40 60 60 60 80 80 80 60 60

1 2 3 4 5 6 7 8 9 10 valid 11 valid 12 13 14 15 16 17 18 19 20 21 valid 22 valid 23 24 25 26 27 28 29 30 31 32 valid 33 valid

1.926 2.039 2.121 2.119 2.379 2.724 2.515 3.226 3.412 2.402 2.368 2.174 2.291 2.402 2.390 2.710 3.089 2.712 3.652 3.751 2.694 2.726 2.370 2.505 2.610 2.610 2.940 3.360 2.970 3.975 4.080 3.010 2.890

0.433 0.444 0.415 0.448 0.549 0.504 0.442 0.462 0.475 0.541 0.553 0.344 0.352 0.334 0.359 0.412 0.391 0.352 0.375 0.382 0.404 0.421 0.354 0.365 0.345 0.368 0.423 0.397 0.371 0.391 0.395 0.411 0.421

1.292 1.624 2.286 1.844 2.022 2.488 2.284 2.384 2.724 2.050 2.006 1.504 1.782 2.554 2.082 2.310 2.804 2.578 2.750 3.084 2.344 2.302 1.590 1.950 2.820 2.280 2.490 3.060 2.820 2.940 3.360 2.502 2.486

0.407 0.416 0.389 0.421 0.515 0.473 0.414 0.433 0.445 0.508 0.519 0.323 0.330 0.317 0.340 0.387 0.367 0.330 0.351 0.359 0.379 0.396 0.332 0.341 0.325 0.346 0.397 0.371 0.347 0.367 0.370 0.385 0.396

1.889 1.996 2.142 1.985 2.235 2.278 2.138 2.412 2.497 2.231 2.240 2.142 2.260 2.343 2.246 2.518 2.579 2.415 2.701 2.811 2.501 2.529 2.325 2.460 2.580 2.445 2.745 2.805 2.625 2.940 3.075 2.751 2.739

0.476 0.489 0.456 0.494 0.604 0.554 0.485 0.509 0.523 0.596 0.603 0.376 0.388 0.373 0.398 0.455 0.431 0.388 0.413 0.422 0.445 0.464 0.389 0.397 0.381 0.407 0.465 0.435 0.408 0.431 0.435 0.451 0.464

Leaves during flowering

1 2 3 4 5 6 7 8 9 10 valid 11 valid 12 13 14 15 16 17 18 19 20 21 valid 22 valid 23 24 25 26 27 28 29 30 31 32 valid 33 valid

Flower heads

2.4. Ultra performance convergence chromatography

3. Results and discussion

The chromatographic analysis was performed using a Waters ACQUITY Ultra-Performance Convergence Chromatography (UPC2 ) system, equipped with a binary solvent delivery pump, an autosampler, a column oven, a diode array detector and a back-pressure regulator. The quantitative analysis of parthenolide was performed using an Acquity UPC2 HSS C18 SB column (100 mm × 3 mm, 1.8 ␮m, Waters, MA, USA). The eluents were supercritical carbon dioxide (A) and acetonitrile (B). An isocratic program was applied: 0–5 min 18% B with a flow rate of 2 ml/min. The column temperature was 45 ◦ C and the detection wavelength was 210 nm during the analysis, UV spectra were recorded from 200 nm to 400 nm. The back pressure was set to 2000 psi. Empower 3 software was used for the data processing. The UPC2 method was validated for accuracy, linearity, limit of detection, limit of quantification, repeatability and intermediate precision in accordance with ICH guideline. The parthenolide standard was dissolved in acetonitrile to yield stock solutions containing 0.0313–0.5000 mg/ml reference substance. Standard solutions at six concentrations were analyzed in triplicate to establish calibration plots. Linearity was determined by linear regression analysis of mean peak area against the corresponding concentration. Limit of detection (LOD) and limit of quantitation (LOQ) values of the method were determined as the ratio between the parthenolide peak height and the baseline noise. The repeatability was calculated according to 9 determinations (3 concentrations/3 replicates) covering the range of the procedure. The intermediate precision was determined in inter-day evaluations (low, mid and high concentration of the standard in three parallel runs on three successive days).

3.1. UPC2 validation

1 2 3 4 5 6 7 8 9 10 valid 11 valid 12 13 14 15 16 17 18 19 20 21 valid 22 valid 23 24 25 26 27 28 29 30 31 32 valid 33 valid

3.1.1. Calibration and linearity Parthenolide standard solutions were prepared as described earlier to determine the linearity of UPC2 response. Standard solutions were analyzed between the range of 0.0313 mg/ml and 0.5000 mg/ml. The relationship between the peak areas and the concentrations was evaluated with linear regression analysis (y = 271.08x + 0.917) for parthenolide. The obtained correlation coefficient of the determination (R2 = 0.99997) indicated high linearity over the investigated concentration range. The statistical significance of the intercept was calculated by Student’s t-test. The calculated test statistic for the intercept was 0.302, which was not greater than the t0.05 value (3.038), thus the absence of interferences was confirmed. The correlation coefficient, slope and y-intercept of the calibration curve are shown in Table 2. 3.1.2. Sensitivity The limit of detection and the limit of quantitation were determined as LOD = 3 × S/N and LOQ = 10 × S/N. In case of the LOD this value was 3.677 ␮g/ml and it was found to be 11.142 ␮g/ml for LOQ. This result suggests, that the UPC2 method allows the detection of very small quantities of parthenolide. LOD and LOQ values are given in Table 2. 3.1.3. Precision Inter-day precision of the method was evaluated by measuring freshly prepared standard solutions in triplicate on three consecutive days. Intra-day precision was evaluated at three different

K. Végh et al. / J. of Supercritical Fluids 95 (2014) 84–91

87

Table 2 Calibration, linearity and sensitivity of the developed convergence chromatographic method. Regression equation (y = ax + b) a (slope)

b (intercept)

271.080

0.917

R2

Linear range (␮g/ml)

Rt repeatability n = 6

LOD (␮g/ml)

LOQ (␮g/ml)

0.99997

31.25–500

RSD% = 0.075

3.677

11.142

Table 3 Precision of the developed convergence chromatographic method. Inter-day

Intra-day

Nominal concentration (mg/ml)

0.5000 0.2500 0.1250

Recovery (%)

RSD (%)

Recovery (%)

RSD (%)

99.97 101.22 101.03

0.047 0.044 0.062

98.87 100.89 100.96

0.054 0.052 0.063

concentrations of the freshly prepared standard solutions in triplicate during the same day. Results were expressed as relative standard deviation (RSD%) (Table 3). The RSD% for the repeatability ranged from 0.044% to 0.062% and the intermediate precision ranged from 0.052% to 0.063%. These values indicate good precision. 3.1.4. Accuracy The accuracy of the method was determined by adding known amounts of parthenolide standard in a known T. parthenium extract at three concentrations. The recovery ranged from 94.77% to 100.72% and RSD% values were less than 0.20% (Table 4). 3.1.5. Application The convergence chromatographic method developed was applied to determine the parthenolide content of the supercritical fluid extracts of T. parthenium L. The target compound was identified by comparison of its retention time and UV spectrum to authentic standard. 3.2. Supercritical fluid extraction The aim of our work was to maximize the parthenolide content of T. parthenium extracts obtained by supercritical fluid extraction. Leaves collected before flowering, leaves collected during flowering and flower heads were studied based on a full factorial 33 experimental design. The total extraction yield and parthenolide content of the 33 SFE experiments of the three samples are shown in Table 1. Accordingly, there are 27 experiments for the DOE and 6 for the validation of the model. These additional runs showed, that the extraction was accurate and reproducible. The total extraction yield in the case of all three samples reached the lowest value in run 1 (10 MPa, 40 ◦ C, 0% EtOH) and the highest value in run 31 (30 MPa, 80 ◦ C, 10% EtOH), while the parthenolide content of samples led to a minimum value in run 14 (30 MPa, 40 ◦ C, 5%EtOH) and a maximum value in run 5 (20 MPa, 60 ◦ C, 0%EtOH). Polynomial regressions were fitted to investigate the main factors influencing the extraction yield and parthenolide content, and to evaluate their trends. The influencing effects coded and un-coded and the coefficients of the responses are presented in Table 5. In case of the extraction yield, the coefficients of determination have shown a value from 0.973 to 0.996, indicating good fitting of the

model. The R2 coefficients of the parthenolide content showed a lower value (0.931–0.934). The effects which were not significant at 95% confidence level were removed from the equations. The final, reduced equations and the regression coefficients are shown in Table 6. After the removal, the adjustable R2 values raised, consequently the model didn’t include any non-significant effect. All the individual factors (ethanol content, pressure, temperature) have a statistically significant positive or negative impact on the responses. The effect of each factor can be evaluated from the regression coefficients of the reduced experimental models. 3.2.1. Total extraction yield The fitted response surface of the total extraction yield in the case of the leaves collected before flowering, the leaves collected during flowering and the flower heads is shown in Fig. 2, for 5% ethanol content, based upon the equations in Table 6. The determination coefficients were 0.972; 0.996 and 0.993, respectively, proving that the models represented the experimental data well. This ascertainment is also confirmed by the diagrams, where the predicted values are depicted against the observed values. The points are in all cases near to the diagonal, which means that the equation describing the surface fits well. The R2 adj did not differ significantly. The Pareto chart of the effects showed that all the individual factors have a significant effect on the extraction yield. In the case of the leaves collected before and during flowering the temperature had a higher influence on the extraction, than the pressure. Positive effect of pressure on the extraction is consistent with the increasing solubility of compounds by the increasing extraction pressure at constant temperature. At high pressures, the solubility of compounds increases by increasing temperature. This crossover effect, usually occurring from 20 to 35 MPa, is caused by the competing effects of the reduction in solvent density and the increase of the vapor pressure. The ethanol content of the supercritical fluid is also among the three most significant factors. The interaction between pressure and temperature and the quadratic effect of these two variables are also notable. The flower heads differed from the other two samples, because the most significant factor in this case was the ethanol content of the supercritical fluid. It is well known that the modifier changes the polarity of the supercritical solvent, thus solubility of both valuable

Table 4 Accuracy of the developed convergence chromatographic method. Parthenolide content (mg/ml)

Added amount (mg/ml)

Theoretical amount (mg/ml)

Recorded amount (mg/ml)

Recovery (%)

RSD (%)

0.147 0.147 0.147

0.0625 0.1250 0.1875

0.2095 0.2720 0.3345

0.2110 0.2650 0.3170

100.72 97.43 94.77

0.121 0.175 0.142

88

K. Végh et al. / J. of Supercritical Fluids 95 (2014) 84–91

Table 5 Regression coefficients of the quadratic model, their significance at 95% confidence level, and the determination coefficients for the full model. Leaves before flowering

ˇ0 (ˇ1 ) EtOH (%) (L) (ˇ11 ) EtOH (%) (Q) (ˇ2 ) Pressure (Mpa) (L) (ˇ22 ) Pressure (Mpa) (Q) (ˇ3 ) Temp. (◦ C) (L) (ˇ33 ) Temp. (◦ C) (Q) (ˇ12 ) 1 L by 2 L (ˇ13 ) 1 L by 3 L (ˇ23 ) 2 L by 3 L R2 R2 adj.

Extraction yield

Parthenolide

Effect

Effect

2.769345 0.553545 0.028682 0.640333 0.085535 1.095000 −0.147132 0.067167 0.078833 0.397167 0.973 0.962

p <0.001 <0.001 0.083 <0.001 <0.001 <0.001 <0.001 0.020 0.011 <0.001

0.404526 −0.093273 −0.057000 0.018611 0.028829 0.028833 0.041829 −0.004420 0.001083 0.020667 0.932 0.905

p <0.001 <0.001 <0.001 0.001 <0.001 <0.001 <0.001 0.322 0.800 0.002

Leaves during flowering

Flower heads

Extraction yield

Parthenolide

Extraction yield

Parthenolide

Effect

Effect

Effect

Effect

2.361099 0.481273 0.040273 0.767333 −0.159035 0.835778 0.028965 0.078667 0.095000 −0.298000 0.996 0.994

p <0.001 <0.001 0.001 <0.001 <0.001 <0.001 0.005 <0.001 <0.001 <0.001

0.379719 −0.087545 −0.052682 0.017333 0.026491 0.026222 0.039825 −0.004000 0.001000 0.019000 0.931 0.904

p <0.001 <0.001 <0.001 0.002 <0.001 <0.001 <0.001 0.361 0.813 0.003

2.432023 0.495182 0.025318 0.322222 0.076070 0.386333 0.004237 0.026667 0.042500 0.082667 0.993 0.989

p <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.252 0.002 <0.001 <0.001

The significant coefficients in each case are written in bold.

Fig. 2. Response surfaces, observed and predicted values and Pareto charts over the extraction yield of the three samples.

0.445351 −0.102364 −0.061182 0.021000 0.030886 0.032111 0.046219 −0.005167 0.001833 0.021667 0.934 0.909

p <0.001 <0.001 <0.001 0.001 <0.001 <0.001 <0.001 0.279 0.688 0.003

K. Végh et al. / J. of Supercritical Fluids 95 (2014) 84–91

89

Fig. 3. Response surfaces, observed and predicted values and Pareto charts over the parthenolide content of the three samples.

and non-valuable co-extracts may increase [19]. Pressure and temperature also have a significant positive effect. The positive effect of pressure and temperature interaction on the extraction was also noted by Kery et al. [22]. The optimum values of the models were also investigated. The extraction yield did not stabilize within the studied ranges, but reached a maximum value at 30 MPa, 80 ◦ C and 10% ethanol content. 3.2.2. Parthenolide extraction yield Fig. 3 shows the response surface of the parthenolide content in leaves collected before and during flowering and flower heads for 5% ethanol. The determination coefficients of the equation were lower (0.931; 0.930; 0.934, respectively), than in the case of the total extraction yield. The observed and predicted values were near to diagonal, so despite of the lower R2 values, they show a god fit of the models. The R2 adj. did not differ significantly, indicating that model did not include non-significant terms (Table 6). The Pareto charts have shown that the ethanol content of the fluid individually and also in quadratic form had the most significant, however negative effect on the extraction of parthenolide. This effect was explained earlier. The co-solvent enhanced the solubility of both valuable and non-valuable components. The increasing polarity, caused by the addition of ethanol, also enhanced the co-extraction of the more polar sesquiterpene lactones and flavonoids. All the other linear and quadratic impacts

had a statistically significant positive effect. In all three cases the temperature had more significant influence on the extraction, than the pressure, also in quadratic form. The observed results on these factors influencing the extraction correlate with the previous study [22]. Only one linear interaction, the pressure-temperature was significant, with a positive effect. It could be observed, that the parthenolide content rose until 60 ◦ C, then decreased at 80 ◦ C. This regressive tendency can be attributed either to the solubility of the compound, which is decreasing with increasing temperature close to the critical pressure, or to the thermal sensibility of parthenolide [23]. Since no degradation products were found during the chromatographic analyzes in any of the samples, the protective effect of the complex sample matrix may occur. The flower heads contained the highest amount of parthenolide (0.604%), which is consistent with previous studies [12,23]. Higher parthenolide content was observed in the leaves collected before flowering than in the leaves collected during flowering. The critical values of the extraction of the leaves collected before and during flowering and the flower heads are the same (7% EtOH, 22 MPa, 64 ◦ C). The optimum point of the temperature was 64 ◦ C, which showed, that a higher rate was necessary to achieve the optimum point than in previous studies [22]. Control extractions for all three samples were made to survey the model’s trueness. These extractions resulted in the following outcomes: the parthenolide content of the leaves collected before flowering under the critical

R2 adj.

0.963

0.912 0.994

0.911

0.990

0.916

0.972

0.931 0.996

0.930

0.993

0.934

K. Végh et al. / J. of Supercritical Fluids 95 (2014) 84–91

R2

90

circumstances was 0.399%, the leaves collected during flowering resulted 0.375% and the flower heads 0.453%. The predicted values were 0.406%; 0.381% and 0.447%, respectively, which were not significantly different from the observed ones, thus the equations of the fitted surfaces were acceptable.

Y = 2.518075 + 0.029744EtOH + 0.003297P − 0.038594T − 0.000855P + 0.000368T + 0.000672P × EtOH + 0.000394T × EtOH + 0.000993P × T Y = 0.025631 − 0.032127EtOH + 0.009362P + 0.012236T + 0.002280EtOH2 − 0.000288P2 − 0.000105T2 + 0.000052P × T Y = −0.298052 + 0.020003EtOH + 0.015519P + 0.042109T − 0.001611EtOH2 + 0.001590P2 − 0.000072T2 + 0.000787P × EtOH + 0.000475T × EtOH − 0.000745P × T Y = 0.021211 − 0.029827EtOH + 0.008613P + 0.011653T + 0.002107EtOH2 − 0.000265P2 − 0.000100T2 + 0.000047P × T Y = 1.355731 + 0.041562EtOH + 0.033258P + 0.004463T − 0.001013EtOH2 − 0.000772P2 + 0.000267P × EtOH + 0.000212T × EtOH − 0.000207P × T Y = 0.024646 − 0.034709EtOH + 0.010154P + 0.013585T + 0.002447EtOH2 − 0.000309P2 − 0.000116T2 + 0.000054P × T Extraction yield (g/100 g)

Parthenolide (g/100 g) Extraction yield (g/100 g)

Leaves before flowering

Leaves during flowering

Parthenolide (g/100 g)

Extraction yield (g/100 g) Flower heads

Parthenolide (g/100 g)

Table 6 The reduced equation fitted on the results of the experimental models.

2

2

Reduced equation Flowering period and response

4. Conclusion The aim of our work was to select the optimum conditions of the supercritical fluid extraction of Tanacetum parthenium L. carried out at different temperatures (40 ◦ C, 60 ◦ C and 80 ◦ C), pressures (10 MPa, 20 MPa and 30 MPa) and modifier contents (0%, 5% and 10% ethanol), in order to maximize the parthenolide recovery of the extracts. The experiments were based on a full factorial 33 design. Three samples were investigated, leaves collected before flowering, leaves collected during flowering and flower heads, according to previous studies that confirmed these two parts of the herb to contain the highest amount of parthenolide [20–23]. The maximum extraction yield was found at 30 MPa, at 80 ◦ C and with 10% ethanol as added co-solvent. The maximum values of parthenolide content for all the three samples, owing to the surface fitting model, were found to be with 7% ethanol, at 22 MPa and at 64 ◦ C. It was confirmed, that the optimum conditions of the extraction, where the maximum parthenolide content and extract yield can be reached, did not coincide. These results may be attributed to the solubility of parthenolide, which is decreasing with increasing temperature close to the critical pressure or to the thermal degradation of the compound. The flower heads contained the highest amount of parthenolide (0.604 wt.%). The leaves collected before flowering contained more parthenolide and other substances compared with the leaves collected during flowering. That means that with the beginning of the flowering period a reduce of contents in the leaves occurs. A simple and rapid isocratic Ultra-Performance Convergence Chromatographic method was developed for the qualitative and quantitative analyses of T. parthenium L. supercritical fluid extracts, which was proved to be reliable and accurate. The method was successfully validated with advantages of high resolution, sensitivity, very good reproducibility and short analysis time. Acknowledgements The authors would like to thank Professor Dr. Sándor Kemény for the help given in the design and evaluation of the experiments. We thank the Waters Hungary Kft., especially Dr. Tamás Juhász for the opportunity to work on the UPC2 convergence chromatographic instrument. References [1] E.S. Johnson, N.P. Kadam, D.M. Hylands, P.J. Hylands, Efficacy of feverfew as prophylactic treatment of migraine, British Medical J. 291 (1985) 569–573. [2] J.J. Murphy, S. Heptinstall, J.R. Mitchell, Randomised double-blind placebocontrolled trial of feverfew in migraine prevention, The Lancet 2 (1988) 189–192. [3] M.H. Pittler, E. Ernst, Feverfew for preventing migraine, Cochrane Database of Systematical Reviews 1 (2004) 2286. [4] R. Pavela, M. Sajfrtová, H. Sovová, M. Bárnet, J. Karban, The insecticidal activity of Tanacetum parthenium (L.) Schultz Bip. extracts obtained by supercritical fluid extraction and hydrodistillation, Industrial Crops and Products 31 (2010) 449–454. [5] N.K. Jain, S.K. Kulkarni, Antinociceptive and anti-inflammatory effects of Tanacetum parthenium L. extract in mice and rats, J. Ethnopharmacology 68 (1999) 251–259. [6] B.H. Kwok, B. Koh, M.I. Ndubuisi, M. Elofsson, C.M. Crews, The antiinflammatory natural product parthenolide from the medicinal herb feverfew directly binds to and inhibits IkappaB kinase, Chemistry and Biology 8 (2001) 759–766. [7] H.O. Collier, N.M. Butt, W.J. McDonald, S.A. Saeed, Extract of feverfew inhibits prostaglandin biosynthesis, The Lancet 2 (1980) 922–923.

K. Végh et al. / J. of Supercritical Fluids 95 (2014) 84–91 [8] A.N. Makheja, J.M. Bailey, A platelet phospholipase inhibitor from the medicinal herb feverfew (Tanacetum parthenium), Prostaglandins, Leukotrienes, and Medicine 8 (1982) 653–660. [9] W.J. Pugh, K. Sambo, Prostaglandin synthetase inhibitors in feverfew, J. Pharmacy and Pharmacology 40 (1988) 743–745. [10] J.Z. Zhou, X. Kou, D. Stevenson, Rapid extraction and high-performance liquid chromatographic determination of parthenolide in feverfew (Tanacetum parthenium), J. Agricultural and Food Chemistry 47 (1999) 1018–1022. [11] J.P. Rey, J. Levesque, J.L. Pousset, Extraction and high-performance liquid chromatographic methods for the gamma lactones parthenolide (Chrysanthemum parthenium Bernh.), marrubiin (Marrubium vulgare L.) and artemisinin (Artemisia annua L.), J. Chromatography A 605 (1992) 124–128. [12] S. Heptinstall, D.V.C. Awang, B.A. Dawson, D. Kindack, D.W. Knight, J. May, Parthenolide content and bioactivity of feverfew (Tanacetum-parthenium (L.) Schultz-Bip.). Estimation of commercial and authenticated feverfew products, J. Pharmacy and Pharmacology 44 (1992) 391–395. [13] H.I. Castro-Vargas, L.I. Rodríguez-Varela, S.R.S. Ferreira, F. Parada-Alfonso, Extraction of phenolic fraction from guava seeds (Psidium guajava L.) using supercritical carbon dioxide and co-solvents, J. Supercritical Fluids 51 (2010) 319–324. [14] C. Yang, Y.R. Xu, W.X. Yao, Extraction of pharmaceutical components from Ginkgo biloba leaves using supercritical carbon dioxide, J. Agricultural and Food Chemistry 50 (2002) 846–849. [15] M.E.M. Braga, P.A.D. Ehlert, L.C. Ming, M.A.A. Meireles, Supercritical fluid extraction from Lippia alba: global yields kinetic data, and extract chemical composition, J. Supercritical Fluids 34 (2005) 149–156.

91

[16] A. Donelian, L.H.C. Carlson, T.J. Lopes, R.A.F. Machado, Comparison of extraction of Patchouli (Pogostemon cablin) essential oil with supercritical CO2 and by steam distillation, J. Supercritical Fluids 48 (2009) 15–20. [17] S.M. Ghoreishi, E. Heidari, Extraction of epigallocatechin gallate from green tea via modified supercritical CO2 : experimental, modeling and optimization, J. Supercritical Fluids 72 (2012) 36–45. [18] K.J. Huang, J.J. Wu, Y.H. Chiu, C.Y. Lai, C.M.J. Chang, Designed polar cosolvent modified supercritical CO2 removing caffeine from and retaining catechins in green tea powder using response surface methodology, J. Agricultural and Food Chemistry 55 (2007) 9014–9020. [19] B. Simándi, Sz.T. Kristo, Á. Kéry, L.K. Selmeczi, I. Kmecz, S. Kemény, Supercritical fluid extraction of dandelion leaves, J. Supercritical Fluids 23 (2002) 135–142. [20] R.M. Smith, M.D. Burford, Supercritical fluid extraction and gas chromatographic determination of the sesquiterpene lactone parthenolide in the medicinal herb feverfew (Tanacetum parthenium), J. Chromatography A 627 (1992) 255–261. [21] M. Kaplan, M.R. Simmonds, G. Davidson, Comparison of supercritical fluid and solvent extraction of feverfew (Tanacetum parthenium), Turkish J. Chemistry 26 (2002) 473–480. [22] A. Kery, E. Ronyai, B. Simandi, E. Lemberkovics, A. Deak, S. Kemeny, Recovery of a bioactive sesquiterpene lactone from Tanacetum parthenium by extraction with supercritical carbon dioxide, Chromatographia 49 (1999) 503–508. [23] L. Cretnik, M. Skerget, Z. Knez, Separation of parthenolide from feverfew: performance of conventional and high-pressure extraction techniques, Separation and Purification Technology 41 (2005) 13–20.