Hindawi Publishing Corporation

e Scientific World Journal

Volume 2014, Article ID 523120, 10 pages
http://dx.doi.org/10.1155/2014/523120

Research Article
Optimization of Reflux Conditions for Total
Flavonoid and Total Phenolic Extraction and Enhanced
Antioxidant Capacity in Pandan (Pandanus amaryllifolius
Roxb.) Using Response Surface Methodology

Ali Ghasemzadeh and Hawa Z. E. Jaafar

Department of Crop Science, Faculty of Agriculture, University Putra Malaysia, 43400 Serdang, Selangor, Malaysia

Correspondence should be addressed to Ali Ghasemzadeh; alighasemzadeh@upm.edu.my

Received 11 April 2014; Revised 6 June 2014; Accepted 3 July 2014; Published 23 July 2014

Academic Editor: Fred Stevens

Copyright © 2014 A. Ghasemzadeh and H. Z. E. Jaafar. This is an open access article distributed under the Creative Commons
Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is
properly cited.

Response surface methodology was applied to optimization of the conditions for reflux extraction of Pandan (Pandanus
amaryllifolius Roxb.) in order to achieve a high content of total flavonoids (TF), total phenolics (TP), and high antioxidant capacity
(AC) in the extracts. Central composite experimental design with three factors and three levels was employed to consider the
effects of the operation parameters, including the methanol concentration (MC, 40%–80%), extraction temperature (ET, 40–70∘ C),
and liquid-to-solid ratio (LS ratio, 20–40 mL/g) on the properties of the extracts. Response surface plots showed that increasing
these operation parameters induced the responses significantly. The TF content and AC could be maximized when the extraction
conditions (MC, ET, and LS ratio) were 78.8%, 69.5∘ C, and 32.4 mL/g, respectively, whereas the TP content was optimal when these
variables were 75.1%, 70∘ C, and 31.8 mL/g, respectively. Under these optimum conditions, the experimental TF and TP content and
AC were 1.78, 6.601 mg/g DW, and 87.38%, respectively. The optimized model was validated by a comparison of the predicted and
experimental values. The experimental values were found to be in agreement with the predicted values, indicating the suitability of
the model for optimizing the conditions for the reflux extraction of Pandan.

1. Introduction most suitable protocols for a specific group of phytochemicals

or plant species. Traditionally, the extraction of phenolic
Phytochemicals are important compounds found in medic- acids and flavonoid compounds is accomplished by reflux
inal plants that exert positive effects on human health or or Soxhlet extraction [4]. However, prolonged extraction at
in amelioration of diseases. Though many phytochemicals high temperature may degrade flavonoid and phenolic acid
have been identified, a great many are yet to be identified compounds [5] and involves a high energy cost. A model
[1]. According to a report by the World Health Organization, for optimizing the most relevant operational parameters is
80% of the population in developing countries depends required in order to achieve higher extraction yield. Response
on traditional medicine for primary health care and 85% surface methodology (RSM) is a collection of statistical and
of traditional medicine is derived from plant extracts [2]. mathematical techniques that is used to optimize the range
Extraction and analysis of plant matrices are primary and of variables in various experimental processes with reducing
important processes for the quality control, modernization, the number of experimental runs, cost, and time compared
and development of herbal formulations [3]. In general, to other methods [6, 7]. Pandan (Pandanus amaryllifolius
the first step of complete extraction is the selection of Roxb.) is a tropical plant of the family Pandanaceae. Pandan
plant parts and careful preparation of plant extracts and a has narrow and strap-shaped green leaves with a spiral
thorough review of the existing literature to determine the arrangement [8]. Pandan is characterized by a sweet and
2 The Scientific World Journal

Table 1: Independent variables and their coded and actual values used for optimization.

Coded level
Independent variable
Unit Symbol −1 0 +1 Axial (−𝛼) Axial (+𝛼)
MC % 𝑋1 10.52 50 89.48 20 80
∘
ET C 𝑋2 23.68 50 76.32 30 70
LS ratio % 𝑋3 16.84 30 43.16 20 40
MC: methanol concentration; ET: extraction temperature; LS ratio: liquid to solid ratio.

Table 2: The result of experimental and predicted values of reflux extraction of TF, TP, and AC.

TF (mg/g DW) TP (mg/g DW) AC%

Run 𝑋1 𝑋2 𝑋3
Experimental Predicted Experimental Predicted Experimental Predicted
values values values values values values
1 10.52 50 30 0.55 0.56 3.12 3.18 44.70 44.80
2 80 70 20 1.52 1.50 6.24 6.20 83.20 83.37
3 50 50 16.84 1.16 1.15 5.55 5.50 70.40 70.42
4 (C) 50 50 30 1.33 1.26 5.94 5.90 75.80 74.48
5 80 70 40 1.74 1.70 6.58 6.55 87.50 87.36
6 20 30 20 0.67 0.68 3.79 3.82 52.00 52.19
7 50 76.32 30 1.40 1.39 6.06 6.06 80.00 80.10
8 (C) 50 50 30 1.22 1.26 5.84 5.90 73.00 74.48
9 80 30 40 1.44 1.43 6.10 6.16 79.40 79.81
10 20 30 40 0.71 0.71 3.82 3.85 53.10 52.98
11 (C) 50 50 30 1.31 1.26 5.89 5.90 75.30 74.48
12 50 50 43.16 1.35 1.33 5.95 5.91 75.80 75.60
13 (C) 50 50 30 1.27 1.26 6.02 5.97 74.20 74.48
14 50 23.68 30 1.15 1.17 5.89 5.83 72.70 73.10
15 20 70 20 0.81 0.79 3.99 3.93 53.80 53.43
16 (C) 50 50 30 1.28 1.26 5.94 5.90 74.50 74.48
17 89.48 50 30 1.38 1.40 6.11 6.14 80.50 80.61
18 (C) 50 50 30 1.21 1.26 5.85 5.90 73.80 74.48
19 20 70 40 0.88 0.90 4.10 4.12 54.90 55.82
20 80 30 20 1.37 1.32 6.00 5.98 78.80 77.49
(C): central point; 𝑋1 : MC; 𝑋2 : ET; 𝑋3 : LS ratio.

delightful flavor and is widely used as a natural flavor in 2. Material and Methods
East Asian countries including Indonesia, Thailand, India,
and Malaysia. In Malaysia, Pandan leaves are also used in the 2.1. Plant Material. Plant samples (P. amaryllifolius) were
production of coconut jam, sweets, desserts, and ice cream. A collected from the North of Malaysia, Bachok, Kelantan
number of studies have demonstrated that Pandan leaves are a province. The samples were identified by the Malaysian
valuable source of phenolic compounds [9–11]. However, far Agriculture Research and Development Institute (MARDI)
too little attention has been paid to optimization of Pandan with voucher specimens of MTP008/1. The leaves were shade
extract in folk medicine. To the best of our knowledge, dried and were powdered using a mechanical grinder. This
there are no studies and reports undertaken to optimize powered material was used for further analysis.
the flavonoid and phenolic extraction from Pandan leaf,
following that improving of antioxidant activity using RSM. 2.2. Extraction. The optimization procedure for the extrac-
The aim of this study is to optimize the extraction conditions tion process focusing on the MC (20–80%), ET (30–70∘ C),
of a Malaysian Pandanus amaryllifolius variety to achieve and LS ratio (20–40 mL/g) was devised based on three-
high TF and TP content and thus to enhance the AC based factor central composite design, as summarized in Table 1.
on the use of RSM method with a central composite design Dissolving of different LS ratio in different MC was designed
(CCD) for optimization of the reflux extraction conditions. by RSM software (Table 2) with a total of 20 extraction
The Scientific World Journal 3

running sets. Solutions were refluxed at various temperatures were dissolved in HPLC grade methanol. The linear regres-
(30–70∘ C) for 2 h. After reflux, the solutions were cooled at sion equation was calculated with 𝑌 = 𝑎𝑋 ± 𝑏, where 𝑋
room temperature and were filtered with a Whatman number was concentration of flavonoid and 𝑌 was the peak area of
1 filter paper and used for future analysis. flavonoids obtained from UHPLC [15]. Compounds were
tentatively identified by comparison of retention times of
2.3. Determination of Total Flavonoids. After extraction, 1 mL standards. All flavonoids and phenolic acids standards were
of extracts was diluted with distilled water (4 mL). Initially, purchased from Sigma-Aldrich (Malaysia).
NaNO2 solution (5%, 0.3 mL) was added to each sample.
After 5 min AlCl3 solution (10%) and at 6 min NaOH (1.0 M, 2.7. Experimental Design. RSM was used to determine the
2 mL) were added. Absorbance of the solutions was read at optimal extraction conditions for maximizing the TF (𝑌1 ), TP
430 nm [12]. (𝑌2 ), and the AC (𝑌3 ). The central composite experimental
design with 3 levels and 3 factors was used to examine the
2.4. Determination of Total Phenolic Content. Leaf extracts extraction variables. Design-Expert software (Version 7.0.0)
(1 mL) were diluted with distilled water (10 mL), and 1 mL was used for data analysis, model building, and experimental
of Folin-Ciocalteu reagent was then added. Solutions were design. The statistical significance of the model and model
allowed to stand for 5 min. Sodium carbonate (2 mL, 20%) variables was determined at the 5% probability level (𝑃 <
was added to the solutions, which were then stored under 0.05).
completely dark conditions at room temperature (25∘ C) for
90 min. The absorbance of the mixtures was read at 750 nm 2.8. Statistical Analysis. The Design-Expert software (Version
[13]. 7.0.0) was used for data analysis, model building, and exper-
imental design. Analysis of variance and response surface
analysis were employed to determine the regression coeffi-
2.5. Determination of Antioxidant Capacity cients and statistical significance of the model terms and to fit
2.5.1. 1,1-Diphenyl-2-picrylhydrazyl (DPPH) Assay. The free the mathematical models of the experimental data that aimed
radical scavenging activity of extracts was determined to optimize the overall region for both response variables. A
according to the Mensor et al. [14] with some modifications. model was applied to predict the response variables as given
DPPH was dissolved in methanol to give final concentration below:
of 2 mM. Following that, 1 mL of DPPH solution was added 𝑌 = 𝑏0 + 𝑏1 𝑋1 + 𝑏2 𝑋2 + 𝑏3 𝑋3 + 𝑏1 2 𝑋1 2 + 𝑏2 2 𝑋2 2
to different concentration of Pandan extracts (20, 40, 60, (2)
80, and 100 mg/mL). The mixture was shaken gently and + 𝑏3 2 𝑋3 2 + 𝑏1 𝑏2 𝑋1 𝑋2 + 𝑏1 𝑏3 𝑋1 𝑋3 + 𝑏2 𝑏3 𝑋2 𝑋3 ,
incubated at 28∘ C in a dark room for 40 min. For the
control, methanol was used as a blank. The absorbance of where 𝑌 is the predicted dependent variable, 𝑏0 is a constant
the samples was read at 517 nm using spectrophotometer. that fixes the response at the central point of the experiment,
BHT (butylhydroxytoluene) and 𝛼-tocopherol were used 𝑏1 , 𝑏2 , and 𝑏3 are the regression coefficients for the linear effect
as positive controls. The scavenging activity was calculated terms, 𝑏1 𝑏2 , 𝑏1 𝑏3 , and 𝑏2 𝑏3 are the interaction effect terms, and
using the following formula: 𝑏1 2 , 𝑏2 2 , and 𝑏3 2 are the quadratic effect terms, respectively.
The relationship between the independent variables (MC:
% inhibition 𝑋1 ; temperature: 𝑋2 , and LS ratio: 𝑋3 ) and the response
variables (TF: 𝑌1 ; TP: 𝑌2 , and AC: 𝑌3 ) was demonstrated
(absorbance of control − absorbance of sample) by the response surface plots. Table 1 shows information
=[ ]
absorbance of control about extraction temperature, MC, and LS ratio of the 20
experiments.
× 100.
(1)
3. Result and Discussion
2.6. Separation and Analysis of Flavonoids by Ultrahigh 3.1. Model Fitting, Statistical Significance Analysis, and
Performance Liquid Chromatography (UHPLC). The UHPLC Response Surface of Reflux Extraction of Total Flavonoid and
system (Agilent, Model 1200) with Agilent C18 (4.6 × Phenolics. In this study, extraction temperatures below 80∘ C
250 mm, 5 𝜇m) column was used for flavonoid separation and were used for the extraction to minimize the possibility
identification. In this system two mobile phases including of degradation of the flavonoid and phenolic compounds,
0.03 M orthophosphoric acid (A) and methanol HPLC grade which has been observed to occur with the application of
(B) were used. The column temperature, flow rate, and high temperatures [5, 16]. In addition, the Maillard reaction
injection volume were adjusted at 35∘ C, 20 𝜇L, and 1 mL/min, may occur at high temperatures, resulting in undesired
respectively. The range of detecting wavelength was between compounds [17]. The results of the experiment and the
260 and 360 nm. Gradient elution was performed as follows: extraction conditions are shown in Table 2. Significant (𝑃 <
0–10 min 40–100% B, 10–15 min 100% B, and 15–20 min 100– 0.05) regression relationships between the response and inde-
40% B and finally washing of the column. To prepare the pendent variables were observed. The TF and TP contents
standard solution all flavonoid and phenolic acid standards of the extract were more significantly affected by the MC
4 The Scientific World Journal

Table 3: Predicted models and statistical parameters calculated after implementation of three-factor central composite design.

Measured 𝑅2 Regression Lack of fit

Predicted models 𝑅2
parameters (adjusted) (𝑃 value) (𝑃 value)
+1.26 + 0.36𝑋1 + 0.095𝑋2 + 0.057𝑋3 + 0.018𝑋1 𝑋2 + 0.023𝑋1 𝑋3
TF 0.98 0.97 0.0001 0.366
+ 0.023𝑋2 𝑋3 − 0.16𝑋1 2 + 0.021𝑋2 2 + 0.00938𝑋3 2
+5.92 + 1.15𝑋1 + 0.12𝑋2 + 0.097𝑋3 + 0.030𝑋1 𝑋2 + 0.038𝑋1 𝑋3
TP 0.99 0.99 0.0001 0.295
+ 0.040𝑋2 𝑋3 − 0.76𝑋1 2 + 0.026𝑋2 2 − 0.10𝑋3 2
+74.46 + 14.19𝑋1 + 2.20𝑋2 + 1.20𝑋3 + 1.18𝑋1 𝑋2 + 0.40𝑋1 𝑋3
DPPH 0.99 0.99 0.0001 0.266
+ 0.40𝑋2 𝑋3 − 6.89𝑋1 2 + 1.05𝑋2 2 − 0.82𝑋3 2

(20–80%), ET (30–70∘ C), and LS ratio (20–40 mL/g). High of 40 : 1 (mL/g). The results of another study showed that a
TF and TP contents of 1.74 and 6.58 mg/g DW, respectively, high total phenolics value was obtained from Parkia speciosa
were observed in the Pandan extracts using treatment run 5 at the LS ratio of 20 mL/g, whereas the total phenolics content
(Table 2). The predicted TF and TP contents for treatment 5 was unaffected by the ET [20]. In current study, the predicted
were 1.7 and 6.55 mg/g DW, which were consistent with the model obtained for TP (𝑌2 ) extraction was as follows:
experimental values. The most striking observation from the
data is that when the MC (𝑋1 ) increased from 20% to 80% 𝑌2 = + 5.92 + 1.15𝑋1 + 0.12𝑋2 + 0.097𝑋3
(at 𝑋2 : 30 and 𝑋3 : 40), the TF and TP content increased
from 0.71 to 1.44 mg/g DW and from 3.82 to 6.10 mg/g DW, + 0.030𝑋1 𝑋2 + 0.038𝑋1 𝑋3 + 0.040𝑋2 𝑋3 (4)
respectively. Increasing the LS ratio resulted in an increment
− 0.76𝑋1 2 + 0.026𝑋2 2 − 0.10𝑋3 2 .
of the TF content of the extracts. In addition, a lower TF
content was recorded at lower LS ratio (20 : 1). Increasing the The model 𝑃 value of 0.0001 implies that the model is
LS ratio from 20 to 40% increased the TF content by 14.4% significant (Table 3). There is only a 0.01% chance that such
(at 𝑋1 : 80% and 𝑋2 : 70∘ C). In the analysis of the TF content, a as large “model 𝑃 value” could be due to noise. The “lack of fit
good coefficient of determination (𝑅2 = 0.98) was obtained, 𝑃 value” of 0.295 implies that the lack of fit is not significant,
where the model explained most of the observed variations relative to the pure error. The effect of the variables and the
(Table 3). Significant (𝑃 < 0.05) linear and quadratic effects interaction of these variables on the responses can be seen in
of the ET, MC, and LS ratio on the TF content (𝑌1 ) were Figures 1 and 2. Figure 1(a) shows the effect of interaction of
observed. The model 𝑃 value of 0.0001 implies that the model the MC and the ET on the TF content of the extract at a fixed
is significant, with only a 0.01% chance that a “model 𝑃 LS ratio of 30%. The minimum TF value was obtained at the
value” this large could be due to noise. The lack of fit test lowest MC and the maximum TF value was obtained at 80%
for the model describes the variation in the data around methanol at the fixed ET of 70∘ C. Figure 1(b) shows the effect
the fitted model. If the model does not fit the data well, the of the interaction of MC and the LS ratio on the TF content at
lack of fit value will be significant; consequently, proceeding a fixed ET of 50∘ C. The minimum TF value was also obtained
with investigation and optimization of the fitted response at the lowest MC and the maximum TF value was obtained at
surface is likely to give misleading results. The “lack of fit 80% methanol at the fixed LS ratio of 40%. Figure 1(c) shows
𝑃 value” of 0.366 obtained herein implies that the lack of the effect of interaction of the ET and the LS ratio on the TF
fit is not significant relative to the pure error. However, no content at a fixed MC of 50%. The minimum TF content was
interactive effect of the independent variables was observed. observed at the lowest LS ratio (20%) and the maximum TF
The predicted model obtained for TF (𝑌1 ) was as follows: value was obtained at a LS ratio of 40% using the fixed ET
𝑌1 = + 1.26 + 0.36𝑋1 + 0.095𝑋2 + 0.057𝑋3 + 0.018𝑋1 𝑋2 of 70∘ C. Moreover, the results indicated that the MC was the
most significant factor affecting the responses at the 𝑃 < 0.01
+ 0.023𝑋1 𝑋3 + 0.023𝑋2 𝑋3 − 0.16𝑋1 2 + 0.021𝑋2 2 level. Figure 2(a) shows the effect of interaction of the MC and
the ET on the TP content at a fixed LS ratio of 30%. As shown
+ 0.00938𝑋3 2 . in Figure 2, the TP content increased significantly to ca.
(3) 84.5% as the MC increased (from 20 to 80%). Furthermore,
the TP content increased slightly as the LS ratio increased
The data in Table 2 demonstrate that when the LS ratio from 20 to 40% (Figure 2(b)). Figure 2(c) shows the effect of
increased from 20 : 1 to 40 : 1, the TP content also increased by interaction of the ET and the LS ratio on the TP content at
about 5.4% (at 𝑋1 : 80% and 𝑋2 : 70∘ C). It is plausible that these 50% methanol. The TP content increased significantly as the
results are due to the fact that more solvent could enter the LS ratio increased up to 32%, but beyond a LS ratio of ∼32%
cells while more phenolic compounds could permeate into the TP content decreased. These findings suggest that the
the solvent at higher LS ratios [18]. The results of previous extraction yield of TF and TP was influenced primarily by the
study showed that as the liquid/solid ratio increased, the MC rather than the ET. It is difficult to explain this result, but
content of phenolic compounds in the extract of E. oleracea this trend might be related to the increased solubility of the
was enhanced [19]. The highest content of phenolics was flavonoid compounds in the mixture of methanol and water
reportedly obtained from fruits of E. oleracea at a LS ratio [21]. The findings of the current study are consistent with
The Scientific World Journal 5

1.6
1.6
1.325
TF (mg/g DW)

1.325

TF (mg/g DW)
1.05
1.05
0.775
0.775
0.5
0.5

70.00 80.00
40.00 80.00
60.00 65.00 35.00 65.00
B: ex 50.00 50.00 n
trac ratio C: li
quid 30.00 50.00 ratio
n
tion 40.00 35.00
oncent to so 25.00 35.00 on cent
tem 30.00 20.00 c c
nol nol
atio 20.00 20.00 : metha
p era lid r
ture etha
A: m A
(a) (b)

1.47

1.3875
TF (mg/g DW)

1.305

1.2225

1.14

40.00 70.00
35.00 60.00
C: li
quid 30.00 50.00 re
ratu
to so 25.00 40.00 te mpe
lid r n
atio 20.00 30.00 c tio
xtra
B: e
(c)

Figure 1: Response surface plots for the effects of MC (20–80%), ET (30–70∘ C), and LS ratio (20–40 mL/g) on the TF content of Pandan
extract. MC and ET (a), MC and LS ratio (b), and ET and LS ratio (c).

those of Liyana-Pathirana and Shahidi [22], who found that Gan and Latiff [20] reported that, in the extraction of Parkia
the TF content of wheat increased with increasing ethanol speciosa, the highest phenolic concentration was achieved at
concentration. an ET of 35∘ C and the extraction rate could be increased by
The result of our study showed that TF and TP con- reducing the extraction time as well as increasing the ET [28].
tent increased with increasing of the temperature till 70∘ C,
but we cannot say that this trend will continue even for 3.2. Model Fitting, Statistical Significance Analysis, and
high temperatures. Previous studies have shown that the Response Surface of Reflux Extraction of Antioxidant Capacity
application of very high temperatures (⩾95∘ C) may alter (AC). The AC of the extract was significantly affected by the
the concentration and composition of phenolic compounds temperature, solvent concentration, and LS ratio (𝑃 < 0.05)
[23]. Some of the authors reported that critical temperature with three linear effects (𝑋1 , 𝑋2 , and 𝑋3 ), two quadratic
for flavonoids is below 80∘ C [24]. In different plants and effects (𝑋1 2 and 𝑋2 2 ), and three interactive effects (𝑋1 𝑋2 ,
organs this temperature will be variable and according to the 𝑋1 𝑋3 , and 𝑋2 𝑋3 ). The DPPH capacity of the extract ranged
previous studies this changes in phenolic acids and flavonoids from 44.7 to 87.5% when treatments 1 and 5 were, respectively,
at high temperature could be related to PAL or CHS enzymes employed. The predicted AC values for treatments 1 and 5
activity at high or low temperature [25]. At high temperature, were, respectively, 44.8 and 87.36%, which were close to the
the flavonoid and phenolic content can be increased as a experimental values. The effect of the variables and their
result of enhancement of their solubility, extraction rate, interaction on the AC of the Pandan extracts is shown in
diffusion rate, and the reduced surface tension and solvent Figure 3. The AC increased in positive proportion to the MC
viscosity [26]. However, further increment of the ET may in the range of 20–80% for the extraction medium. Thus, the
degrade flavonoids and phenolics due to destabilization of MC of the extraction medium had a significant influence on
the compounds by reaction with other plant components the antioxidant properties of the Pandan extracts. Liyana-
or enzymatic and chemical degradation, thus reducing the Pathirana and Shahidi [22] reported that a higher AC was
extraction efficiency [27]. In contrast with the current results, obtained in the extraction of wheat by using 50% ethanol
6 The Scientific World Journal

6.6 6.4
5.725
TP (mg/g DW)

TP (mg/g DW)
5.575
4.85 4.75
3.975 3.925
3.1 3.1

70.00 80.00 40.00 80.00

60.00 65.00 35.00 65.00
B: ex 50.00 50.00 n C: li
trac ratio quid 30.00 50.00 trati
on
tion
tem 40.00 35.00 cent 35.00 ncen
lid r 25.00 20.00 20.00
con to so o
p era 30.00 20.00 nol no l c
ture etha atio etha
A: m A: m
(a) (b)

6.12
TP (mg/g DW)

5.9775

5.835

5.6925

5.55

40.00 70.00
35.00 60.00
C: li 30.00 50.00
quid ture
to so 25.00 40.00 e mp era
lid r n t
atio 20.00 30.00 c tio
xtra
B: e
(c)

Figure 2: Response surface plots for the effects of MC (20–80%), ET (30–70∘ C), and LS ratio (20–40 mL/g) on the TP content of Pandan
extract. MC and ET (a), MC and LS ratio (b), and ET and LS ratio (c).

compared to other aqueous solvents. The current finding is AC. In another study, high AC of Parkia speciosa extract
in agreement with those of Pompeu et al. [19] and Kiassos et was achieved using a liquid/solid ratio of 20 mL/g [20]. The
al. [29], who demonstrated that the concentration of ethanol regression equation obtained for the AC (𝑌3 ) as the response
had a significant influence on the AC of onion extract. variable also showed significant (𝑃 < 0.05) dependence of 𝑌3
The ET caused a linear increase in the AC of wheat extract, on the variation of the independent variables. The predicted
and increasing the temperature increased the total AC. It model obtained for 𝑌3 is given below:
was confirmed that the rate of extraction of thermally stable 𝑌3 = + 74.46 + 14.19𝑋1 + 2.20𝑋2 + 1.20𝑋3
antioxidants at elevated temperature was higher than the
rate of decomposition of less soluble antioxidants [22]. The + 1.18𝑋1 𝑋2 + 0.40𝑋1 𝑋3 + 0.40𝑋2 𝑋3 (5)
temperature utilized during extraction generally influences
the compound stability due to chemical and enzymatic degra- − 6.89𝑋1 2 + 1.05𝑋2 2 − 0.82𝑋3 2 .
dation and losses by thermal decomposition; these factors
The model 𝑃 value of 0.0001 obtained for the AC implies that
have been suggested to be the main mechanisms underlying
the model is significant (Table 3). The “lack of fit 𝑃 value” of
reduction of the polyphenol content in the extraction of grape 0.266 implies that the lack of fit is not significant, relative to
[30]. Pompeu et al. [19] obtained high AC in the extraction the pure error. There is a 26.65% chance that a “lack of fit
of Euterpe oleracea at a temperature of 58∘ C. Similarly, high 𝐹 value” this large could occur due to noise. One question
AC of wheat extracts was observed when a temperature of that needs to be asked, however, is why does software take
61∘ C was utilized [22] and high AC was achieved (83.37%) about five or six center points in the design? The reason is
when the LS ratio was low (20 mL/g). The latter is plausibly also related to the variance of a predicted value. When fitting
due to increased probability of the antioxidant components a response surface we want to estimate the response function
coming into contact with the extraction solvent as the amount in this design region where we are trying to find the optimum.
of solvent increased. However, further increase of the LS We want the prediction to be reliable throughout the region
ratio may dilute the extraction solution thereby lowering the and especially near the center since we hope the optimum
The Scientific World Journal 7

87 83
76.25 73.25
DPPH (%)

DPPH (%)
65.5 63.5
54.75 53.75
44 44

70.00 80.00 40.00 80.00

B: ex 60.00 65.00
n
35.00 65.00
trac 50.00 50.00 ratio quid 30.00 ation
C: li 50.00
tion
tem 40.00 35.00 oncent to so 25.00 35.00 ncentr
ture 30.00 20.00
p era c co
nol lid r
atio 20.00 20.00 nol
etha etha
A: m A: m
(a) (b)

80

77.6
DPPH (%)

75.2

72.8

70.4

40.00 70.00
35.00 60.00
C: li 50.00 ture
quid 30.00
to so 25.00 40.00 e mp era
lid r 20.00 30.00 nt
atio c tio
xtra
B: e
(c)

Figure 3: Response surface plots for the effects of MC (20–80%), ET (30–70∘ C), and LS ratio (20–40 mL/g) on the AC of Pandan extract. MC
and ET (a), MC and LS ratio (b), and ET and LS ratio (c).

is in the central region. By picking five to six center points, conditions for the three responses to evaluate the adequacy
the variance in the middle is approximately the same as the of the response surface models for predicting the optimum
variance at the edge. If we only had one or two center points, response values. As shown in Table 4, the observed
then we would have less precision in the middle than we values of the TF and TP content and AC were 1.78,
would have at the edge. As we go farther out beyond a distance 6.601 mg/g DW, and 87.38%, respectively. The response
of 1 in coded units, we get more variance and less precision. surface models of TF, TP, and antioxidant activity were
What we are trying to do is to balance the precision at the verified using the experimental and predicted values. The
edge of the design relative to the middle. obtained results from verification experiment were in
consent with the predicted values, because nonsignificant
3.3. Optimization of Reflux Extraction Condition for TF (𝑃 > 0.05) difference was observed between the verification
and TP Content and AC. The optimum reflux extraction experimental and the predicted values (𝐸TF,TP = 0.069%;
conditions for maximizing the TF and TP and for achiev- 𝐸DPPH = 0.48%).
ing high AC of Pandan extracts were predicted using the
Design-Expert software. Multiple graphical and numerical
optimizations were carried out to determine the optimum 3.5. Identification of Flavonoids and Phenolic Acids. Figure 5
level of independent variables with desirable response goals. shows the UHPLC chromatogram of the identified flavonoid
Two optimal conditions were developed for the responses: and phenolic acids of the Pandan extract. Gallic acid, (+)-
the TF content and AC were maximized using a MC of catechin, caffeic acid, myricetin, luteolin, and quercetin were
78.8%, ET of 69.5∘ C, and LS ratio of 32.4 mL/g, whereas the identified in the Pandan extract at concentrations of 0.489,
corresponding conditions for maximizing TP were 75.1%, 0.594, 0.856, 0.076, 0.08 and 0.112 mg/g DW, respectively.
70∘ C, and 31.8 mL/g, respectively (Table 4, Figure 4). Caffeic acid was the most abundant compound of the identi-
fied compounds, and the concentration of phenolic acids was
3.4. Verification of the Models. The experiment was per- higher than that of the flavonoid compounds in the Pandan
formed using the recommended optimum treatment extract.
8 The Scientific World Journal

20.00 80.00 30.00 70.00 20.00 40.00

Methanol concentration = 78.80 Extraction temperature = 69.50 Liquid to solid ratio = 32.40

0.55 1.80 3.12 6.58 44.7 87.5

TF = 1.78125 TP = 6.52991 DPPH = 87.38

Desirability = 0.949
(a)

20.00 80.00 30.00 70.00 20.00 40.00

Methanol concentration = 75.10 Extraction temperature = 70.00 Liquid to solid ratio = 31.80

0.55 1.74 3.12 6.65 44.7 87.5

TF = 1.59726 TP = 6.6010 DPPH = 86.0711

Desirability = 0.943
(b)

Figure 4: Predicted value of TF, TP, and DPPH activity of Pandan leaf from optimized extraction condition using RSM ((a) optimized
condition for TF and DPPH activity; (b) optimized condition for TP).

Table 4: Optimum conditions and experimental value of responses at the optimum conditions.

Optimum conditions TF content Optimum conditions TP content Optimum conditions AC (DPPH assay)
(predicted) (predicted) (predicted)
MC (%) 78.8 75.1 78.8
ET (∘ C) 69.5 1.78 70 6.601 69.5 87.38
LS ratio (mL/g) 32.4 31.8 32.4
TF and TP: mg/g DW; AC: %.

3 enhancement of the AC of the extracts. The results can be

600 1 2 easily explained on the basis that both the ET and the MC
(mAU)

400 have a positive effect on the solubility of flavonoids in the

200 4 56 extraction solution. The most efficient set of reflux conditions
0
for Pandan leaf extraction are at MC 78.8% (𝑋1 ), ET 69.5∘ C
0 2 4 6 8 10 12 14 16 18
(𝑋2 ), and LS ratio of 32.4 mL/g (𝑋3 ) for maximizing the TF
(min)
content and AC and 75.1% (𝑋1 ), 70∘ C (𝑋2 ), and 31.8 mL/g
Figure 5: UHPLC chromatogram of Pandan extract. Identified (𝑋3 ) for maximizing the TP content with consequently high
compounds are gallic acid (1), catechin (2), caffeic acid (3), myricetin AC. Moreover, the models used to fit the response variables
(4), luteolin (5), and quercetin (6). were significant (𝑃 < 0.01), and the “lack of fit” was not
significant (𝑃 > 0.05) for all responses, indicating that the
models used to fit the response variables were adequate for
representing the relationship between the response values
4. Conclusion and the independent variables.

The reflux extraction of TF and TP and the AC of Pandan Conflict of Interests

extract were successfully optimized using RSM. The results
indicate that the MC, ET, and LS ratio had a significant The authors declare that they have no conflict of interests
effect on the TF and TP extraction yields with consequent regarding the publication of this paper.
The Scientific World Journal 9

Acknowledgments DPPH free radical method,” Phytotherapy Research, vol. 15, pp.
127–130, 2001.
The authors would like to thank the Ministry of Higher [15] H. Bae, G. K. Jayaprakasha, J. Jifon, and B. S. Patil, “Extraction
Education, Malaysia, and the Research Management Cen- efficiency and validation of an HPLC method for flavonoid
tre, University Putra Malaysia (UPM), for sponsoring this analysis in peppers,” Food Chemistry, vol. 130, no. 3, pp. 751–758,
work. Authors also are grateful to the Malaysian Agriculture 2012.
Research and Development Institute for collecting of the [16] Í. Şensoy, R. T. Rosen, C.-T. Ho, and M. V. Karwe, “Effect of
plant samples. The authors would like to acknowledge all processing on buckwheat phenolics and antioxidant activity,”
staff of Laboratory of Nutrition, Department of Nutrition and Food Chemistry, vol. 99, no. 2, pp. 388–393, 2006.
Dietetics, Faculty of Medicine and Health Sciences, for all the [17] J. Shi, M. L. Maguer, Y. Kakuda, A. Liptay, and F. Niekamp,
help and guidance in order to accomplish this project. “Lycopene degradation and isomerization in tomato dehydra-
tion,” Food Research International, vol. 32, no. 1, pp. 15–21, 1999.
[18] K. N. Prasad, E. Yang, C. Yi, M. Zhao, and Y. Jiang, “Effects of
References high pressure extraction on the extraction yield, total phenolic
[1] J. Boyer and R. H. Liu, “Apple phytochemicals and their health content and antioxidant activity of longan fruit pericarp,”
benefits,” Nutrition Journal, vol. 3, article 1, 2004. Innovative Food Science & Emerging Technologies, vol. 10, no. 2,
pp. 155–159, 2009.
[2] WHO Organization, WHO Traditional Medicine Strategy 2002–
2005, WHO, 2002. [19] D. R. Pompeu, E. M. Silva, and H. Rogez, “Optimisation of
the solvent extraction of phenolic antioxidants from fruits
[3] M. J. Bak, S. Ok, M. Jun, and W. S. Jeong, “6-shogaol-rich extract
of Euterpe oleracea using Response Surface Methodology,”
from ginger up-regulates the antioxidant defense systems in
Bioresource Technology, vol. 100, no. 23, pp. 6076–6082, 2009.
cells and mice,” Molecules, vol. 17, no. 7, pp. 8037–8055, 2012.
[20] C. Gan and A. A. Latiff, “Optimisation of the solvent extrac-
[4] A. Khoddami, M. A. Wilkes, and T. H. Roberts, “Techniques for
tion of bioactive compounds from Parkia speciosa pod using
analysis of plant phenolic compounds,” Molecules, vol. 18, no. 2,
response surface methodology,” Food Chemistry, vol. 124, no. 3,
pp. 2328–2375, 2013.
pp. 1277–1283, 2011.
[5] A. M. Nuutila, K. Kammiovirta, and K.-M. Oksman-Caldentey, [21] G. K. Jayaprakasha, B. Girennavar, and B. S. Patil, “Radical
“Comparison of methods for the hydrolysis of flavonoids and scavenging activities of Rio Red grapefruits and Sour orange
phenolic acids from onion and spinach for HPLC analysis,” Food fruit extracts in different in vitro model systems,” Bioresource
Chemistry, vol. 76, no. 4, pp. 519–525, 2002. Technology, vol. 99, no. 10, pp. 4484–4494, 2008.
[6] F. Gu, F. Xu, L. Tan, H. Wu, Z. Chu, and Q. Wang, “Optimization [22] C. Liyana-Pathirana and F. Shahidi, “Optimization of extraction
of enzymatic process for vanillin extraction using response of phenolic compounds from wheat using response surface
surface methodology,” Molecules, vol. 17, no. 8, pp. 8753–8761, methodology,” Food Chemistry, vol. 93, no. 1, pp. 47–56, 2005.
2012.
[23] M. Igual, E. Garcı́a-Martı́nez, M. M. Camacho, and N.
[7] C. H. Kuo, F. W. Hsiao, J. H. Chen, C. W. Hsieh, Y. C. Liu, Martı́nez-Navarrete, “Effect of thermal treatment and storage
and C. J. Shieh, “Kinetic aspects of ultrasound-accelerated lipase on the stability of organic acids and the functional value of
catalyzed acetylation and optimal synthesis of 4󸀠 -acetoxyres- grapefruit juice,” Food Chemistry, vol. 118, no. 2, pp. 291–299,
veratrol,” Ultrasonics Sonochemistry, vol. 20, no. 1, pp. 546–552, 2010.
2013.
[24] G. Xu, X. Ye, J. Chen, and D. Liu, “Effect of heat treatment on
[8] K. V. Peter, Handbook of Herbs and Spices, Elsevier, 2012. the phenolic compounds and antioxidant capacity of citrus peel
[9] S. R. Chilakwad, K. Manjunath, K. S. Akki, R. Savadi, and N. extract,” Journal of Agricultural and Food Chemistry, vol. 55, no.
Deshpande, “Pharmacognostic and Phytochemical investiga- 2, pp. 330–335, 2007.
tion of leaves of Pandanus odoratissimus Linn.f.,” Ancient [25] M. T. Lafuente, A. R. Ballester, J. Calejero, and L. González-
Science of Life, vol. 28, no. 2, pp. 3–6, 2008. Candelas, “Effect of high-temperature-conditioning treatments
[10] A. Ghasemzadeh and H. Z. Jaafar, “Profiling of phenolic on quality, flavonoid composition and vitamin C of cold stored
compounds and their antioxidant and anticancer activity in “Fortune” mandarins,” Food Chemistry, vol. 128, no. 4, pp. 1080–
pandan (Pandanus amaryllifolius Roxb.) extracts from different 1086, 2011.
locations of Malaysia,” BMC Complementary and Alternative [26] Z. Y. Ju and L. R. Howard, “Effects of solvent and temperature
Medicine, vol. 13, article 341, 2013. on pressurized liquid extraction of anthocyanins and total
[11] K. H. Miean and S. Mohamed, “Flavonoid (myricetin, querce- phenolics from dried red grape skin,” Journal of Agricultural and
tin, kaempferol, luteolin, and apigenin) content of edible tropi- Food Chemistry, vol. 51, no. 18, pp. 5207–5213, 2003.
cal plants,” Journal of Agricultural and Food Chemistry, vol. 49, [27] N. E. Durling, O. J. Catchpole, J. B. Grey et al., “Extraction of
no. 6, pp. 3106–3112, 2001. phenolics and essential oil from dried sage (Salvia officinalis)
[12] B. Sultana, F. Anwar, and M. Ashraf, “Effect of extraction sol- using ethanol-water mixtures,” Food Chemistry, vol. 101, no. 4,
vent/technique on the antioxidant activity of selected medicinal pp. 1417–1424, 2007.
plant extracts,” Molecules, vol. 14, no. 6, pp. 2167–2180, 2009. [28] J. E. Cacace and G. Mazza, “Extraction of anthocyanins and
[13] D.-O. Kim, O. I. Padilla-Zakour, and P. D. Griffiths, “Flavonoids other phenolics from black currants with sulfured water,”
and antioxidant capacity of various cabbage genotypes at Journal of Agricultural and Food Chemistry, vol. 50, no. 21, pp.
juvenile stage,” Journal of Food Science, vol. 69, no. 9, pp. C685– 5939–5946, 2002.
C689, 2004. [29] E. Kiassos, S. Mylonaki, D. P. Makris, and P. Kefalas, “Imple-
[14] L. L. Mensor, F. S. Menezes, G. G. Leitão et al., “Screening of mentation of response surface methodology to optimise extrac-
Brazilian plant extracts for antioxidant activity by the use of tion of onion (Allium cepa) solid waste phenolics,” Innovative
10 The Scientific World Journal

Food Science and Emerging Technologies, vol. 10, no. 2, pp. 246–
252, 2009.
[30] M. Corrales, S. Toepfl, P. Butz, D. Knorr, and B. Tauscher,
“Extraction of anthocyanins from grape by-products assisted by
ultrasonics, high hydrostatic pressure or pulsed electric fields: a
comparison,” Innovative Food Science & Emerging Technologies,
vol. 9, no. 1, pp. 85–91, 2008.