Optimization of ultrasonic-assisted extraction

conditions for bioactive components from coffee

leaves using the Taguchi design and response
surface methodology
Xiumin Chen , Jian Ding, Dayi Ji, Suqun He, and Haile Ma

Abstract: Coffee leaves contain various bioactive compounds that are beneficial for human health. However, there
are very limited researches related to the extraction of the bioactive phytochemicals from coffee leaves. In the present
study, the extraction conditions for bioactive components from coffee leaves were optimized using Taguchi design and
response surface methodology (RSM). Taguchi design was used to screen significant factors that affected the yield of
phytochemicals including trigonelline, caffeine, chlorogenic acids, mangiferin, and rutin, total phenolic content (TPC)
and antioxidant activity. Sequentially, a Box-Behnken design (BBD) was used to optimize the extraction conditions.
Three factors including Liquid-to-solid (L:S) ratio, ethanol concentration, and extraction temperature that significantly
affected most of the phytochemical yields and antioxidant activity were selected from the six variables using Taguchi
design. The optimal extraction conditions obtained from RSM were 30.3:1 L:S ratio, 54.5% ethanol, and 80 °C when
simultaneously considered four responses, including TPC, the yields of mangiferin and 5-CQA and DPPH scavenging

Food Chemistry
capacity. Under the optimal conditions, the experimental results for the above four responses were 62.1 mg gallic acid/g,
4.1 mg/g, 11.4 mg/g, and 356.9 µmol Trolox/g, respectively, which were close to the predicted values. About 97% of
phytochemicals can be extracted in the first two times of extraction. In conclusion, the combination of Taguchi design
and response surface methodology can be successfully used to screen and optimize the significant factors that affected the
bioactive components extracted from coffee leaves.

Keywords: Coffee leaves, ultrasonic-assisted extraction, Box-Behnken design, Taguchi design, antioxidant

Practical Application: Coffee leaves, the byproducts of coffee plants, are considered no- or low-value although it has
a long history for using them as tea-like beverage and ethnomedicine by locals in the coffee plant growing countries.
Bioactive components extracted from coffee leaves can be used as ingredients in functional beverages, functional food,
and natural health products. These applications will add values to coffee leaves as well as increase the incomes of coffee
farmers and workers.

1. INTRODUCTION diabetics, anti-cardiovascular, and anti-cancer capacities (Campa

Coffee plants are widely grown in the developing countries such et al., 2012; Chen, 2019; Chen, Mu, & Kitts, 2019). Coffee leaves
as Brazil, Vietnam, Columbia, Indonesia, Ethiopia, and others that have been used as ethnomedicine and tea-like beverage in coffee
are located in tropical or subtropical regions (FAOSTAT, 2017). plant growing countries such as South Sudan, Ethiopia, Indonesia,
According to the statistics of the Food and Agriculture Organiza- et al. (Patay, Bencsik, & Papp, 2016; Ross, 2005). Recently,
tion of the United Nations, the global coffee harvest area reaches industry applications of coffee leaves such as therapeutic agents for
10.8Mha in 2017 (FAOSTAT, 2017). Coffee leaves are coffee hypertension, gastrointestinal inflammation, and diabetes; tobacco
plant by-products that are generally considered non- or low-value substitute; animal feed; packaging material; personal hygienic
compared with the highly valuable coffee beans. However, coffee products; vehicle perfume; and antimicrobial agents et al. have
leaves contain various phytochemicals such as chlorogenic acids, been discovered (Chen, 2019; Kenconojati, Ulkhaq, Budi, &
mangiferin, rutin, caffeine, and trigonelline that possess different Azhar, 2019; Segheto et al., 2018). Due to the rich resources and
bioactivities including antioxidant, anti-inflammatory, anti- various applications, extraction of bioactive components from
coffee leaves is considered a promising way to produce value-
added products such as functional food, natural health products,
JFDS-2019-1972 Submitted 12/6/2019, Accepted 2/19/2020. Authors are with
School of Food and Biological Engineering, Jiangsu University, 301 Xuefu Road,
and beverage. However, to date, the information regarding the
Jingkou District, Zhenjiang, Jiangsu, 212013, P.R. China. Authors Chen and Ma extraction of bioactive components from coffee leaves are scant. In
are also with Institute of Food Physical Processing, Jiangsu University, 301 Xuefu our previous researches, hot water (Chen, Ma, & Kitts, 2018) and
Road, Jingkou District, Zhenjiang, Jiangsu, 212013, P.R. China. Author Chen 80% methanol (Chen, Kitt, Ji, & Ding, 2019) were used to extract
is also with International Research Center for Food Nutrition and Safety, Jiangsu phytochemicals from coffee leaves, however, both studies did not
University, Zhenjiang, 212013, China. Direct inquiries to author Chen (Email:
xmchen@ujs.edu.cn). focus on the extraction methods for phytochemicals. Marcheafave
et al. (2019) investigated the solvent effects on the pigments and

C 2020 Institute of Food Technologists

 R

doi: 10.1111/1750-3841.15111 Vol. 0, Iss. 0, 2020 r Journal of Food Science 1

Further reproduction without permission is prohibited
 Optimizing extraction of coffee leaf . . .

antioxidant activities of coffee leave extract using experimental standards comprising 3-CQA, 4-CQA, 5-CQA, 3,4-diCQA,
mixture design. However, this study only focused on the solvent 3,5-diCQA, 4,5-diCQA were obtained from Chengdu Must
effects and pigments instead of other extraction conditions and Bio-Technology Co. Ltd. (Chengdu, China). DPPH and
beneficial phytochemicals such as chlorogenic acids, mangiferin, (±)-6-hydroxy-2,5,7,8-tetramethylchromane-2-carboxylic acid
rutin, et al. (Trolox) were purchased from Shanghai Macklin Biochemical Co.
Extraction conditions such as type of solvent, liquid-to-solid Ltd. (Shanghai, China). Folin-Ciocalteu reagent was purchased
(L:S) ratio, extraction temperature and time, and particle size from Beijing Solarbio Science & Technology Co. Ltd. (Beijing,
are known to affect the yield and bioactivities of plant extracts, China). Trifluoroacetic acid (TFA) and ethanol were purchased
however, their impacts varied. Traditional one factor at a time from Sinopharm Chemical Reagent Co. Ltd. (Shanghai, China).
experiment is time-consuming and does not provide interaction HPLC grade acetonitrile (ACN) were purchased from Tedia
information. Taguchi design is very effective in the screening pro- Company Inc. (Fairfield, OH, USA).
cedure to identify the significant factors while reduces the number
of experiments (Ashengroph, Nahvi, & Amini, 2013). Response 2.2 Extraction procedure
surface methodology (RSM) is an efficient mathematical and Approximately 1.0 g of coffee leaves were placed into a 50 mL
statistical method that is generally used to improve parameters centrifuge tube that contained an appropriate volume of solvent.
in different procedures and to explore their probable interactions After extraction, the resulting solution was filtered, centrifuged
using a mathematical model to predict the optimal conditions and (5,000 × g), and stored in the refrigerator for further study. All
responses. RSM has been applied in the optimization of extraction experiments were performed in triplicate.
conditions (Ayim, Ma, & Alenyorege, 2018; Zhang & Ma, 2017).
The integration of Taguchi design and RSM approaches has been 2.3 Analysis of total phenolic contents
widely applied in process optimization (Dash, Mohammed, & Total phenolic contents of coffee leaf extracts were determined
Humaira, 2016; Thakur, Das, & Das, 2016). Ashengroph et al. using Folin–Ciocalteu method described by Chen, Tait, and Kitts
Food Chemistry

(2013) used Taguchi design to screen the significant influencing (2017) and the data were expressed as mg gallic acid (GA)/g
components from ten ingredients that affect the conversion coffee leaves. Briefly, 20 µL appropriately diluted coffee leaf
of isoeugenol to vanillin by Psychrobacter sp. CSW4, followed extracts or gallic acid (0.1 mg/mL) were incubated with 100 µL,
by optimization of the selected important factors using RSM. ten times diluted Folin–Ciocalteu solution for 1 min followed by
Notwithstanding, to the best of our knowledge, there was no pre- reacting with 80 µL, 75 mg/mL Na2 CO3 for 30 min in a 96-well
vious study regarding the extraction of bioactive compounds from plate. The absorbance was recorded at 765 nm with Spark  R
10M
coffee leaves utilizing ultrasound-assisted extraction with Taguchi multimode microplate reader (Tecan, MA, USA). Each sample
design and RSM. Therefore, in the present study, the combination was measured three times.
of Taguchi design and RSM was used to screen the significant
factors and to the optimization of extraction conditions. 2.4 Analysis of antioxidant activity
In the present study, a two-levels, six-factors, Taguchi design The antioxidant activity of coffee leaf extracts was determined
was first used to screen the significant factors from six variables by the DPPH method, as described previously by Chen et al.
including ethanol concentration, extraction temperature and (2018). Briefly, coffee leaf extracts were diluted to 4 to 20 times
time, L:S ratio, age of coffee leaves, and ultrasound power. Twelve using methanol, then 5 to 50 µL diluted samples or Trolox
responses that are the yield of trigonelline, caffeine, mangiferin, (0.2 mM) were added into 96-well plate that contained 20 µL,
rutin, chlorogenic acids including 3-caffeoylquinic acid (3-CQA), 1 mM DPPH. An appropriate amount of methanol was added
4-caffeoylquinic acid (4-CQA), 5-caffeoylquinic acid (5-CQA), into each well to make up the final volume to 200 µL. The
3,4-dicaffeoylquinic acid (3,4-diCQA), 3,5-dicaffeoylquinic acid mixtures were incubated at room temperature in darkness for
(3,5-diCQA), 4,5-dicaffeoylquinic acid (4,5-diCQA), total phe- 10 min followed by reading at 519 nm with Spark R
10M
nolic content (TPC), and 2,2-diphenyl-1-picrylhydrazyl (DPPH) multimode microplate reader (Tecan, MA, USA). The data were
free radical scavenging capacity were measured. Sequentially, expressed as µmol Trolox/g coffee leaves.
the optimal extraction conditions for the three most significant
factors including L:S ratio, ethanol concentration, and extraction 2.5 HPLC analysis of phytochemical content
temperature were optimized using Box-Behnken design (BBD) The phytochemical contents of coffee leaves were analyzed and
when simultaneously optimized mangiferin, 5-CQA, TPC, and quantified by the method that was described by Chen et al. (2018)
DPPH scavenging capacity under ultrasonic condition. The with some modification. Waters HPLC system including Waters
impacts of extraction times on the phytochemical content and 2707 autosampler, Waters 2998 photodiode array detector and
antioxidant activities of coffee leaf extracted with or without Waters 1525 binary HPLC Pump, equipped with a Phenomenex
ultrasound treatment were also compared. Kinetex C18 (100 × 4.8 mm, 5 µm) column (Phenomenex,
Torrance, CA, USA) was used to quantify phytochemical contents
2. MATERIALS AND METHODS in the coffee leaf extract. The temperature was set at 25 °C
and the flow rate was 1.5 mL/min. Appropriately diluted coffee
2.1 Coffee leaves and chemicals leaf extracts (5 µl) were injected into the column and eluted
Dry coffee leaves were donated by Wize Monkey Inc. (Vancou- with solvent A (0.1% TFA in water) and solvent B (ACN). The
ver, BC, Canada). Young and mature Coffea Arabica var. Catimor mobile phase gradient was as follows: 0 to 10 min, solvent B
leaves were collected from Finca La Aurora coffee farm located in linearly increased from 5% to 20%; 10 to 12 min, solvent B was
the Matagalpa area of Nicaragua (13°03 35.5 N 85°54 19.0 W), maintained at 20%; 12 to 15 min, solvent B decreased from 20%
between August 18 and 25, 2016 and prepared with Chinese-style to 5%; 15.1 to 16.5 min, solvent B was maintained at 95%; 16.6
green tea processed method (Chen et al., 2018). Trigonelline, to 19 min, solvent B was maintained at 5%. The compounds in-
caffeine, mangiferin, rutin, and chlorogenic acids (CGAs) isoform cluding trigonelline, caffeine, CGAs (3-CQA, 4-CQA, 5-CQA,

2 Journal of Food Science r Vol. 0, Iss. 0, 2020

Optimizing extraction of coffee leaf . . .

Figure 1–HPLC profile of coffee leaf extract

(monitored at 257 nm).

3,4-diCQA, 3,5-diCQA, 4,5-diCQA), rutin, and mangiferin Where d1 . . . dm is the individual desirability for 1 . . . m
were detected at 264, 280, 330, 257, and 257 nm, respectively. number of responses; d is the overall desirability.
The HPLC profile of coffee leaf extract is shown in Figure 1.
2.7 Statistical analysis
2.6 Experimental Design

Food Chemistry
All experiments were carried out in triplicates. Taguchi and
2.6.1 Taguchi design. A two-level Taguchi L8 (26 )
BBD were designed and analyzed using Minitab version 17
orthogonal design (Table 1) was used to screen the significant
software from Minitab Inc. (State College, PA, USA). Significant
factors out of six independent variables including L:S ratio,
differences were evaluated by analysis of variance (ANOVA) at P
ethanol concentration, extraction temperature, extraction time,
< 0.05 and Tukey’s test was used to perform pairwise comparison.
ultrasound power, and age of leaves. The twelve dependent
variables that were evaluated included the yields of caffeine,
trigonelline, chlorogenic acids (3-CQA, 4-CQA, 5-CQA, 3,4- 3. RESULTS AND DISCUSSION
diCQA, 3,5-diCQA, and 4,5-diCQA), mangiferin, and rutin, 3.1 Results from the Taguchi design
TPC, and DPPH scavenging capacity. The factors and levels are
Numerous factors such as solvent type and concentration, L:S
as follows: ethanol concentration (0 & 60%), temperature (30 and
ratio, time, temperature, ultrasonic power, and particle size are
80 °C), ultrasound power (0 and 210 W), time (10 & 40 min),
known to affect the extraction of bioactive components. How-
coffee leaf age (young and mature), and L:S ratio (10:1 and 40:1).
ever, the significance of those factors varied due to different plant
2.6.2 Box–Behnken design. A Box–Behnken design
sources and responses that are evaluated. Therefore, screening of
(BBD) comprised of three factors, three levels, and three center
factors that significantly affect the dependent variables before the
points that includes 15 experiments (Table 2) was utilized to deter-
optimization of the extraction conditions is very critical in con-
mine the optimal conditions of L:S ratio, ethanol concentration,
sideration of reducing time and cost. The drawbacks of the single
and extraction temperature on the responses when mature coffee
factor experiment include that first, only one factor is varied at a
leaves were extracted for 10 min at ultrasound power of 210W
time while the rest of the factors remained the same; second, no in-
(ultrasonic cleaner KQ-300DE, Kunshan Ultrasonic Instrument
teraction information can be obtained, which results in difficulty
Co., Ltd). The 12 responses were analyzed by multiple regression
to correctly compare the relative significance of multi-factors.
equation to fit into a second-order polynomial model shown as
Taguchi design is a robust parameter design that applies the orthog-
follows:
onal array for process optimization when a minimum number of
 k K 
k−1
experiments are used. A two-level multi-factor Taguchi design can
Y = β0 + βi Xi + βi i Xi2i + βi j Xi X j (1) be effectively used to screen the factors that significantly influence
i =1 i =1 i the response variables, thus reduce the time and experimental cost.
where Y is the predicted response; Xi (i = 1, 2, 3) are independent The Taguchi experimental results are presented in Tables 1
variables (L:S ratio, ethanol concentration (%), and temperature), and the main effects of each factor on different responses were
β 0 , βi, βii, and βij are the regression coefficients for the intercept, shown in Figure 2. The extraction yields of phytochemicals,
linear, quadratic, and interaction coefficients of the model, respec- TPC, and DPPH free radical scavenging capacity of coffee leaf
tively. The optimal conditions for each response were predicted extracts were increased as the extraction temperature, ethanol
based on the regression equation. In addition, the extraction con- concentration, L:S ratio, extraction time and ultrasound power
ditions when simultaneously optimized multi-responses including increased, whereas, in case of the age of leaves, the results were
TPC, DPPH scavenging capacity, 5-CQA, and mangiferin were vice versa. Our previous studies also showed that mature leaves
predicted based on desirability functions. The overall desirability contained less chlorogenic acids, mangiferin, rutin, TPC and
that represents the desirability of the multi-responses can be cal- antioxidant activities compared with their young counterparts
culated from the individual desirability values using the following (Chen et al., 2019; Chen et al., 2018).
equation (Derringer & Suich, 1980). The ranking of the impact of each factor on the twelve
responses are shown in Table 1S (supplement data). L:S ra-
d = (d1 · d2 · . . . · dm) 1/m
(2) tio was the most significant factor that affected the yield of

Vol. 0, Iss. 0, 2020 r Journal of Food Science 3

Food Chemistry Optimizing extraction of coffee leaf . . .

Figure 2–Taguchi experimental results of the main effects of different factors on mean values of different responses. (A) Trigonelline; (B) Caffeine;
(C) 3-CQA; (D) 4-CQA; (E) 5-CQA; (F) 3,4-diCQA; (G) 3,5-diCQA; (H) 4,5-diCQA; (I) Mangiferin; (J) Rutin; (K) TPC; (L) DPPH scavenging capacity. T:
temperature (°C), L:S: liquid-to-solid ratio, UP: ultrasound power (W), t: time (min), Y: young coffee leaves, M: mature coffee leaves. TPC: total phenolic
content, expressed as mg gallic acid per g coffee leaf (dry weight). DPPH is expressed as µmol Trolox per g coffee leaf (dry weight).

phytochemicals and the antioxidant activity and ethanol concen- are affected by several independent variables and thus reach the
tration was the second most important factor followed by the optimization purpose.
extraction temperature. Extraction time, ultrasound power, and The results of RSM are shown in Table 2. The ranges of
age of leaves did not significantly affect most of the responses. mangiferin, 5-CQA, TPC, and DPPH scavenging capacity were
A studied conducted by Belwal, Dhyani, Bhatt, Rawal, and 0.53 to 4.29 mg/g, 4.53 to 11.98 mg/g, 22.57 to 57.43 mg
Pande (2016) also showed that extraction temperature, L:S ratio, GA/g, and 122.23 to 359.36 µmol Trolox/g, respectively. The
and solvent concentration significantly affected the TPC, total dependent variables and the independent variables were fitted
flavonoids, total anthocyanins, total tannins, and antioxidant into the second-order polynomial equations that were shown in
activities of the extracts derived from Berberis asiatica fruits, supplement data. The ANOVA Tables 2S -12S (supplement data)
whereas, extraction time and pH did not have significant effects. listed the P values of model, linear, quadratic, and interaction
Therefore, L:S ratio, ethanol concentration, and extraction terms for each response. After deleting the non-significant
temperature were selected for further optimization using BBD. quadratic and interaction terms, the variables were fitted into the
following equations:
Trigonelline = 2.560 + 0.4561 × 1 − 0.0236 × 2 + 0.0118 ×
3.2 Results for RSM
3 − 0.007780 × 12 − 0.001038 × 1 ∗ X2 −
RSM is a combined mathematical and statistical techniques
that are used to model and analyze the response variables that 0.000848 × 1 ∗ X3 + 0.000735 × 2 ∗ X3

4 Journal of Food Science r Vol. 0, Iss. 0, 2020

 Table 1–Taguchi design and the results for the factor screening experiments.

Taguchi design Phytochemical content (mg/g)

DPPH
T EtOH L:S Time UP Age Trigon- 3- 4- 5- 3,4- 3,5- 4,5- Mangi- TPC (mg (µmol
No. (°C) (%) (w:v) (min) (W) (M) elline Caffeine CQA CQA CQA diCQA diCQA diCQA Rutin ferin GA/g) Trolox/g)

1 30 0 10 10 0 Y 3.7 ± 0.0 6.0 ± 0.1 0.1 ± 0.1 0.2 ± 0.0 4.0 ± 0.2 0.0 ± 0.0 0.4 ± 0.1 N.D. 1.0 ± 0.0 1.1 ± 0.0 19.1 ± 1.5 78.0 ± 6.0
2 30 0 10 40 210 M 4.0 ± 0.1 5.6 ± 0.1 0.1 ± 0.0 0.2 ± 0.0 5.0 ± 0.4 0.1 ± 0.0 0.5 ± 0.0 N.D. 1.4 ± 0.0 1.4 ± 0.0 25.3 ± 0.9 105.3 ± 6.6
3 30 60% 40 10 0 M 9.3 ± 0.7 8.7 ± 0.6 0.3 ± 0.0 0.4 ± 0.0 10.7 ± 0.7 0.3 ± 0.0 1.4 ± 0.1 0.5 ± 0.0 2.2 ± 0.3 2.2 ± 0.3 48.0 ± 2.6 272.4 ± 5.1
4 30 60% 40 40 210 Y 10.7 ± 0.1 12.2 ± 1.0 0.4 ± 0.0 0.5 ± 0.0 12.3 ± 0.8 0.4 ± 0.0 2.1 ± 0.2 0.5 ± 0.0 2.4 ± 0.4 2.4 ± 0.2 52.6 ± 3.2 295.0 ± 12.7
5 80 0 40 10 210 Y 8.5 ± 0.4 12.0 ± 0.1 0.3 ± 0.0 0.4 ± 0.0 9.8 ± 0.1 0.3 ± 0.0 1.7 ± 0.0 0.4 ± 0.0 2.2 ± 0.2 2.1 ± 0.0 47.2 ± 0.2 255.6 ± 11.2
Optimizing extraction of coffee leaf . . .

6 80 0 40 40 0 M 8.0 ± 0.5 9.0 ± 0.7 0.2 ± 0.0 0.4 ± 0.0 8.4 ± 0.1 0.2 ± 0.0 1.1 ± 0.1 0.4 ± 0.0 2.0 ± 0.2 2.2 ± 0.2 43.2 ± 0.9 242.8 ± 19.2
7 80 60% 10 10 210 M 6.8 ± 0.1 6.9 ± 0.4 0.2 ± 0.0 0.3 ± 0.1 7.7 ± 0.2 0.2 ± 0.0 0.9 ± 0.3 0.3 ± 0.1 1.7 ± 0.2 2.0 ± 0.2 40.9 ± 3.9 252.2 ± 10.6
8 80 60% 10 40 0 Y 8.2 ± 0.4 9.3 ± 0.1 0.3 ± 0.0 0.4 ± 0.0 9.4 ± 0.7 0.2 ± 0.0 1.7 ± 0.1 0.4 ± 0.0 2.1 ± 0.0 1.7 ± 0.0 42.1 ± 0.6 267.0 ± 19.7

Data are expressed as mean ± SD of triplicate experiments. T: temperature, L:S: liquid-to-solid ratio, UP: ultrasound power, Y: young coffee leaves, M: mature coffee leaves. TPC: total phenolic content, expressed as mg gallic acid per g coffee
leaf. DPPH is expressed as µmol Trolox per g coffee leaf.

Table 2–Box–Behnken design and results for optimization experiments.

Box-Behnken design Phytochemical content (mg/g)

L: S EtOH T Trigon- 3- 4- 5- 3,4- 3,5- 4,5- Mangi- TPC (mg DPPH (µmol
No. (w: v) (%) (°C) elline Caffeine CQA CQA CQA diCQA diCQA diCQA Rutin ferin GA/g) Trolox/g)

1 10 (-1) 0 (-1) 55 (0) 6.6 ± 0.1 5.5 ± 0.1 0.3 ± 0.0 0.4 ± 0.0 5.5 ± 0.3 0.4 ± 0.0 0.4 ± 0.0 0.3 ± 0.0 1.4 ± 0.1 1.6 ± 0.1 22.6 ± 2.0 122.2 ± 5.7
2 40 (+1) 0 (-1) 55 (0) 6.6 ± 0.5 7.1 ± 0.2 0.2 ± 0.0 0.2 ± 0.0 4.5 ± 0.5 0.2 ± 0.0 0.3 ± 0.0 N.D. 0.3 ± 0.0 0.5 ± 0.1 34.5 ± 1.4 144.4 ± 5.7
3 10 (-1) 60 (+1) 55 (0) 7.2 ± 0.0 7.1 ± 0.2 0.3 ± 0.0 0.5 ± 0.0 8.7 ± 0.1 0.4 ± 0.0 1.3 ± 0.0 0.5 ± 0.0 2.2 ± 0.0 3.1 ± 0.3 38.5 ± 1.0 286.2 ± 11.2
4 40 (+1) 60 (+1) 55 (0) 5.3 ± 0.3 7.0 ± 0.5 0.1 ± 0.0 0.4 ± 0.0 8.4 ± 0.5 0.3 ± 0.0 1.1 ± 0.0 0.4 ± 0.0 0.7 ± 0.0 2.7 ± 0.1 42.9 ± 0.3 340.3 ± 19.4
5 10 (-1) 30 (0) 30 (-1) 5.5 ± 0.1 6.0 ± 0.1 0.2 ± 0.0 0.3 ± 0.0 8.0 ± 0.1 0.2 ± 0.0 0.8 ± 0.0 0.2 ± 0.0 0.6 ± 0.0 2.2 ± 0.1 35.4 ± 1.2 229.4 ± 14.2
6 40 (+1) 30 (0) 30 (-1) 6.5 ± 0.1 7.6 ± 0.2 0.2 ± 0.0 0.4 ± 0.0 10.4 ± 0.2 0.3 ± 0.0 1.2 ± 0.0 0.4 ± 0.0 0.8 ± 0.0 3.0 ± 0.0 47.8 ± 2.0 263.5 ± 1.4
7 10 (-1) 30 (0) 80 (+1) 7.6 ± 0.2 6.7 ± 0.6 0.3 ± 0.0 0.6 ± 0.0 8.7 ± 0.8 0.4 ± 0.0 1.0 ± 0.0 0.4 ± 0.0 1.9 ± 0.0 2.7 ± 0.0 31.9 ± 3.1 207.4 ± 9.1
8 40 (+1) 30 (0) 80 (+1) 7.3 ± 0.1 8.2 ± 0.1 0.2 ± 0.0 0.4 ± 0.0 11.1 ± 0.3 0.4 ± 0.0 1.4 ± 0.1 0.4 ± 0.0 0.9 ± 0.0 3.4 ± 0.0 53.0 ± 1.0 336.0 ± 1.3
9 25 (0) 0 (-1) 30 (-1) 9.2 ± 0.9 7.4 ± 0.2 0.4 ± 0.0 0.4 ± 0.0 6.6 ± 0.2 0.4 ± 0.0 0.6 ± 0.0 0.3 ± 0.0 1.2 ± 0.0 2.0 ± 0.1 33.6 ± 3.5 132.3 ± 2.8
10 25 (0) 60 (+1) 30 (-1) 7.4 ± 0.1 8.9 ± 0.1 0.3 ± 0.0 0.6 ± 0.0 10.7 ± 0.5 0.4 ± 0.0 1.4 ± 0.1 0.5 ± 0.0 2.4 ± 0.0 3.6 ± 0.0 50.2 ± 1.8 318.2 ± 7.1
11 25 (0) 0 (-1) 80 (+1) 8.0 ± 0.2 7.9 ± 0.1 0.4 ± 0.0 0.6 ± 0.0 8.4 ± 0.1 0.5 ± 0.0 0.9 ± 0.0 0.3 ± 0.0 2.2 ± 0.1 2.7 ± 0.1 28.7 ± 0.6 161.8 ± 14.0
12 25 (0) 60 (+1) 80 (+1) 8.4 ± 0.0 9.5 ± 0.0 0.3 ± 0.0 0.8 ± 0.0 12.0 ± 0.2 0.5 ± 0.0 1.7 ± 0.0 0.5 ± 0.0 3.0 ± 0.0 4.3 ± 0.2 57.4 ± 3.7 359.4 ± 8.0
13 25 (0) 30 (0) 55 (0) 8.4 ± 0.5 7.9 ± 0.3 0.3 ± 0.0 0.6 ± 0.0 10. 5 ± 0.2 0.5 ± 0.0 1.3 ± 0.0 0.5 ± 0.0 2.3 ± 0.1 3.3 ± 0.1 44.5 ± 0.8 253.7 ± 0.1
14 25 (0) 30 (0) 55 (0) 9.0 ± 0.8 7.9 ± 0.2 0.3 ± 0.0 0.6 ± 0.0 10.4 ± 0.1 0.4 ± 0.0 1.2 ± 0.1 0.4 ± 0.0 2.3 ± 0.0 3.5 ± 0.0 47.0 ± 2.1 270.7 ± 11.7
15 25 (0) 30 (0) 55 (0) 7.8 ± 0.3 8.0 ± 0.2 0.3 ± 0.0 0.6 ± 0.1 10.2 ± 0.7 0.4 ± 0.0 1.2 ± 0.0 0.4 ± 0.0 2.3 ± 0.0 3.4 ± 0.2 48.8 ± 2.3 274.2 ± 17.2

Data are expressed as mean ± SD of triplicate experiments. T: temperature, L:S: liquid-to-solid ratio. TPC: total phenolic content, expressed as mg gallic acid per g coffee leaf. DPPH is expressed as µmol Trolox per g coffee leaf.

Vol. 0, Iss. 0, 2020 r Journal of Food Science 5

Food Chemistry
 Optimizing extraction of coffee leaf . . .

0.836 to 0.946, and the adjusted R2 s are in between 0.781 and

Caffeine = 3.696 + 0.3492 × 1 + 0.04281 × 2 − 0.0673 ×
0.962 indicating that all the models adequately agree with the
3 − 0.005647 × 12 + 0.000716 × 32 − 0.000939 experimental results. ANOVA analysis (Table 2S to 12S) showed
that all regression models are significant (P < 0.001) for all
× 1 ∗ X2 these twelve responses. According to the P values of lack of fit
3 − CQA = 0.0184 + 0.02399 × 1 − 0.002378 × 2 + (Table 3), the models adequately fitted the experimental data for
trigonelline, caffeine, 3-CQA, 4-CQA, 3,4-diCQA, TPC, and
0.002244 × 3 − 0.000469 × 12 + 0.000023 DPPH radical scavenging capacity, whereas, they did not fit the
experimental data of 5-CQA, 3,5-diCQA, 4,5-diCQA, rutin,
× 22 − 0.000059 × 1 ∗ X3
and mangiferin due to the small P-value (P < 0.05).
4 − CQA = −0.2094 + 0.05144 × 1 + 0.003696 × 2 3.2.1 Effects of L:S ratio. According to the results of the
Taguchi design, L:S ratio was the most important factor that
+ 0.00118 × 3 − 0.000955 × 12 − 0.000047 affected the yield of phytochemicals and the antioxidant activity
× 22 + 0.000042 × 32 − 0.000140 × 1 ∗ X3 of coffee leaf extract. In general, the higher of L:S ratio the
greater amounts of phenolic compounds dissolved in the solvent.
+ 0.000041 × 2 ∗ X3 However, the increased solvent volume resulted in a waste of
organic solvent, therefore, an appropriate L:S ratio is crucial
5 − CQA = 4.36 + 0.4267 × 1 + 0.1903 × 2 − 0.1504 for an efficient extraction process. The yield of phytochemicals
× 3 − 0.007894 × 12 − 0.002130 × 22 increased with increased L:S ratio followed by a decrease with
a further increase of L:S ratio. The optimal L:S ratio for the
+ 0.001570 × 32 twelve responses ranged from 19.7:1 to 40:1 and the results varied
3, 4 − diCQA = −0.1890 + 0.03549 × 1 + 0.000958 × 2 depending on the responses. This is partly due to the different
Food Chemistry

polarities of those phytochemicals. Other researches also found

+ 0.004962 × 3 − 0.000631 × 12 different optimal L:S ratio for phenolic compounds. For example,
Yang et al. (2009) found that the extraction rate of phenolic
− 0.000111 × 1 ∗ X3
compounds increased significantly when L:S ratio increased from
3, 5 − diCQA = 0.191 + 0.05786 × 1 + 0.02808 × 2 10:1 to 20:1, however, further increase of L:S ratio, could not
significantly increase the extraction rate. Rajha et al. (2014)
− 0.01771 × 3 − 0.001072 × 12 − 0.000229 reported that an L:S ratio of 3:1 was the best to recover total
× 22 + 0.000199 × 32 phenolics from grape pomace. However, a much higher L:S ratio,
50:1, was used by Belwal et al. (2016) to optimize the extraction of
4, 5 − diCQA = −0.1962 + 0.02463 × 1 + 0.00655 × 2 phenolic compounds and antioxidant activity from Berberis asiatica
fruits.
+ 0.00488 × 3 − 0.000453 × 12 − 0.000078 3.2.2 Effects of ethanol concentration. Due to the varied
× 22 + 0.000094 × 1 ∗ X2 − 0.000108 polarity of phytochemicals in plants, various solvents are used for
phytochemical extraction. Ethanol, dichloromethane, and hexane
× 1 ∗ X3 were used to extract pigments including pheophytin α, zeaxan-
thin, crocetin, β-carotene and chlorophyll b from Coffea arabica L.
Mangiferin = 0.550 + 0.1957 × 1 + 0.06737 × 2 − 0.0507
leaves and ethanol was found to be the best solvent for the yield and
× 3 − 0.003853 × 12 − 0.000663 × 22 the antioxidant activities of coffee leaf extract(Marcheafave et al.,
2019). Partition of methanol extract of coffee leaves with ethyl
+ 0.000566 × 32 acetate produced a fraction that contained the highest amount of
Rutin = −2.636 + 0.2757 × 1 + 0.01302 × 2 + 0.03589 TPC, total flavonoids, 5-CQA, and antioxidant activities (Segheto
et al., 2018). Phenolic compounds are distributed in the cell
× 3 − 0.005187 × 12 − 0.000817 × 1 ∗ X3 vacuoles or in the cell wall where they are bonded with proteins or
polysaccharides with hydrogen bond or hydrophobic interaction
TPC = 18.39 + 1.526 × 1 + 0.464 × 2 − 0.2463 × 3 (Yang et al., 2009). Alcohol and water mixtures have been found
− 0.02856 × 12 − 0.00662 × 22 + 0.00577 to be more effective to extract phenolic compounds compared
with the mono-component solvent (Yilmaz & Toledo, 2006).
× 1 ∗ X3 + 0.00405 × 2 ∗ X3 Water and low concentration of ethanol can easily penetrate into
the cell to dissolve the phenolic compounds, whereas, high con-
DPPH = 112.1 + 1.61 × 1 + 5.100 × 2 − 0.9760 × 3
centration of ethanol was often found to decrease the extraction
− 0.00617 × 12 − 0.03337 × 22 + 0.0630.1 × 1 ∗ X3 rate because it can cause protein denaturation, thus prevents the
dissolving of phenolic compounds. Yang et al. (2009) found that
where X1: L:S ratio, X2: EtOH concentration (%), and X3: the extraction rate for polyphenols from the bark of Phyllanthus
temperature. emblica L. increased from 30 to 70% ethanol followed by a decrease
Surface plots, contour plots, and optimal extraction conditions with a further increase of ethanol concentration. The optimal
for each response are shown in Figure 3, Figure 1S (supplementary ethanol concentration was 75% under microwave-assisted extrac-
data) and Table 3, respectively. The coefficient of determination tion condition. In the present study, for most of the responses,
(R2 ) greater than 0.8 for a chemical experiment indicates a the increased ethanol resulted in greater phytochemical content,
satisfy fitness of the regression model (Lundstedt et al., 1998). TPC and antioxidant activity, whereas, in the case of 3-CQA, a
In the present study, R2 for all the response models ranged from decreasing trend was found. The optimal ethanol concentration

6 Journal of Food Science r Vol. 0, Iss. 0, 2020

Optimizing extraction of coffee leaf . . .

Food Chemistry
Figure 3–Response surface plots for the effects of L:S ratio, ethanol concentration, and temperature on different responses. (A) Trigonelline; (B)
Caffeine; (C) 3-CQA; (D) 4-CQA; (E) 5-CQA; (F) 3,4-diCQA; (G) 3,5-diCQA; (H) 4,5-diCQA; (I) Mangiferin; (J) Rutin; (K) TPC; (L) DPPH scavenging capacity.
X1, L:S ratio; X2, ethanol concentration; X3, temperature.

Table 3–Optimal extraction conditions, P, R2 , R2 (adj.) and predicted values for RSM.

Responses L:S ratio (w/v) EtOH conc. (%) Temperature (°C) P-value R2 R2 (adj.) Predicted values (mg/g)
Trigonelline 21.2 55.8 80 0.743 0.836 0.781 8.98
Caffeine 25.8 60.0 80 0.302 0.925 0.899 9.28
3-CQA 20.9 4.0 80 0.116 0.928 0.904 0.39
4-CQA 21.0 60.0 80 0.561 0.971 0.962 0.82
5-CQA 27.2 43.6 80 0.033 0.906 0.875 12.33
3,4-diCQA 21.5 60.0 80 0.067 0.842 0.789 0.55
3,5-diCQA 26.4 60.0 80 0.185 0.907 0.875 1.70
4,5-diCQA 23.6 57.0 80 0.000 0.846 0.794 0.58
Rutin 19.7 60.0 80 0.000 0.951 0.934 2.99
Mangiferin 26.1 51.5 80 0.007 0.899 0.865 4.34
TPC 33.0 57.6 80 0.195 0.910 0.880 56.4 mg gallic acid/g
DPPH 40.0 60.0 80 0.966 0.964 0.952 402.0 µmol trolox/g
P, significant (P < 0.05) of lack of fit; R2 is the coefficient of determination of each model; R2 (adj.) is the adjusted R2 .

(Table 3) was close to 60% for most of the responses, except for present study was 80 °C, indicating that increased temperature ac-
3-CQA. celerated the yield of phytochemicals, which positively correlated
3.2.3 Effects of temperature. Our results found that in with the increased DPPH free radical scavenging capacity.
the range of temperature that was tested in the present study, the Rajha et al. (2014) also found that the increase of temperature
increase of temperature resulted in the greater phytochemical con- caused the increases of TPC, flavonoid, tannin, anthocyanin, and
tent, TPC and antioxidant activity, whereas, it showed an opposite antioxidant activity of grape pomace extract and the optimum
tend in case of 3-CQA. The optimal temperature in the range of extraction temperature was 93 °C in consideration of the above

Vol. 0, Iss. 0, 2020 r Journal of Food Science 7

 Optimizing extraction of coffee leaf . . .

DPPH (µmol
multiple responses. Other researches reported that the extraction

2.7a∗
3.2b
0.3b

4.5b
0.8c
2.8a
Trolox/g)
yield of phenolic compounds and antioxidant activities was

±
±
±
±
±
±
enhanced with increasing temperature (Santos et al., 2016). High

Data are expressed as mean ± SD of triplicate experiments. TPC is expressed as mg gallic acid per g coffee leaf. DPPH is expressed as µmol Trolox per g coffee leaf. The same group data in the same column with different letters denote
356.9
42.2
4.9
325.8
49.3
5.1
temperature accelerates the mass transfer, increases the solubility
of solute as well as reduces the surface tension and viscosity
(Rajha et al., 2014). However, heat-labile phenolic compounds

(mg GA/g)

1.3b

0.8b
0.1c

0.3c
4.8a

1.5a
can be degraded under high-temperature long-time extraction,

TPC

±
±
±
±
±
±
thus resulting in the low TPC. Yang et al. (2009) found that the

62.1
11.1
1.9
62.3
12.8
2.4
extraction rate gradually increased when temperature increased

significantly different at P < 0.05. N.D.: not detectable. Asterisk denotes significantly different when compared phytochemical contents extracted with or without ultrasound treatment at the same extraction time
from 20 °C to 50 °C followed by a decrease when further
increased the temperature. In the present study, the optimum
Mangiferin

0.1b

0.1b
0.0b
0.0c
0.1a

0.9a
temperature for all the responses was 80 °C, which was similar to
±
±
±
±
±
±
the results obtained by Belwal et al. (2016).
4.1
1.0
0.1
3.5
1.1
0.2
3.2.4 Optimal conditions when considered multi-
responses. In some experiments, researchers are interested in
0.1b

0.0b
0.0b
0.0c
0.1a

0.3a

several responses rather than a single response. In the case of

Rutin

extraction bioactive components from plant, both yield and

±
±
±
±
±
±

bioactivities are the target responses because a single response

2.5
0.6
0.1
2.1
0.7
0.1

does not adequately represent the quality of the plant extract.

3,4-diCQA 3,5-diCQA 4,5-diCQA

Moreover, due to the different polarity of phytochemicals, their

0.1 ± 0.0b

0.1 ± 0.0b
0.1 ± 0.0b
0.4 ± 0.0a

0.6 ± 0.1a

solubility varied depending on different solvents and extraction

N.D.

conditions. Therefore, the optimal extraction conditions for

Food Chemistry

different phytochemicals and the bioactivities are different. To

reach the goal of simultaneous optimization of multiple responses,
0.5 ± 0.1b

0.5 ± 0.0b
0.1 ± 0.0b
1.5 ± 0.1a

1.3 ± 0.2a

an optimal response surface model for each response is built first,

N.D.

then the condition that keeps all responses in the desired range is
assessed later. Derringer’s desirability function is one of the tech-
Phytochemical content(mg/g)

niques that is used for simultaneous optimizing multiple responses

0.1 ± 0.0b

0.1 ± 0.0b
0.4 ± 0.0a

0.4 ± 0.0a

(Montgomery, 2013). The closer of the d value to 1, the more

N.D.

N.D.

desirable response. In the present study, the appropriate response

surface models for twelve responses including trigonelline, caf-
feine, chlorogenic acids, mangiferin, rutin, TPC and antioxidant
1.3a∗
0.4b

0.2b
0.0b
0.1c
1.2a

activities were built. 5-CQA is the most abundant phenolic that

5-CQA

exists in coffee leaves. Mangiferin is a xanthone that is unique in

±
±
±
±
±
±

Coffea Arabica leaves but not detectable in coffee beans. It is well

12.4
3.2
0.4
9.9
3.5
0.5

known that mangiferin possesses various bioactivities that are able

to prevent or inhibit inflammation, diabetics, hyperlipidemia,
0.1 ± 0.0b

0.1 ± 0.0b
0.5 ± 0.2a

0.7 ± 0.1a
4-CQA

oxidation, microbial, and neuro-dysfunction (Jyotshna, Khare &

N.D.

N.D.

Shanker, 2016). The choice of the extraction conditions must

consider both the chemical compositions of the extracts and the
antioxidant activity; therefore, the aim of this research was to
0.1 ± 0.0b

0.1 ± 0.0b
0.4 ± 0.0a

0.5 ± 0.2a

establish the extraction conditions that mangiferin, 5-CQA, TPC,

3-CQA

N.D.

N.D.

and DPPH scavenging capacity were simultaneously optimized

to keep in a desirable range. Our results showed that the optimal
conditions were L:S ratio, 30.3:1; ethanol concentration, 54.5%;
temperature, 80 °C when considered those four responses
0.5a∗
0.3b

0.1b
0.0c

0.0c
1.0a
Caffeine
Table 4–Results of confirmation experiment.

simultaneously. The predicted results for the above responses were

±
±
±
±
±
±

12.01 mg/g, 4.27 mg/g, 56.1 mg GA/g, and 365.8 µmol Trolox/g,
8.5
2.1
0.2
7.1
2.3
0.3

respectively. The overall desirability for these four responses is

0.957, which is close to 1, indicating that the optimal extraction
Trigonelline
0.4 a
0.3b

0.1b
0.0c

0.0c
0.8a

conditions were overall desirable for these four responses.

±
±
±
±
±
±
7.0
2.2
0.19
6.3
2.4
0.3

3.3 Conformation experiment

The confirmation experiments were done under the above opti-
mal conditions with and without ultrasound treatment. The leaves
1
2
3
1
2
3

were also extracted three times sequentially to compare the differ-

Without ultrasound

ent extraction times on the yield of phytochemicals and antioxi-

Extraction times
With ultrasound

dant activity. Under the optimal conditions, the yield of 5-CQA,

mangiferin, and TPC were 12.4 mg/g and 4.1 mg/g, 62.1 GA/g,
respectively, and DPPH scavenging capacity reached 356.9 µmol
Trolox/g (Table 4). The experimental results were similar to the
predicted values. The first time extraction obtained 74.5 to 83.3%

8 Journal of Food Science r Vol. 0, Iss. 0, 2020

Optimizing extraction of coffee leaf . . .

and 68.4 to 87.5% phytochemicals under ultrasonic condition and ACKNOWLEDGMENTS

without ultrasound treatment, respectively. About 12.4 mg/g 5- This study was supported by Jiangsu Specially-Appointed
CQA was extracted from coffee leaves under ultrasound condition Professor Program (1711360016) and Senior Talent Startup
at the first time extraction, which is significantly higher than the Fund of Jiangsu University (4111360002) to X. Chen, Graduate
yield of 9.9 mg/g that was obtained under the condition without Student Innovative Program of Jiangsu University (5561360022)
ultrasound treatment. Similar results were also found in the case to J. Ding and X. Chen. We thank Dr. David D Kitts from
of DPPH scavenging capacity, which was 356.9 and 325.8 µmol University of British Columbia, Arnaud Petitvallet and Max
Trolox/g, respectively. The phytochemicals in the first two times Rivest from Wize Monkey Inc. (Vancouver, BC, Canada) for the
extraction accounted for 96.9% of the total three times extraction, kind donation of coffee leaves, as well as the Finca La Aurora
indicating most of the phytochemicals were extracted at the first coffee farm in Nicaragua for providing samples of coffee leaves.
two times extraction. A greater amount of 5-CQA (16.0 mg/g vs
13.9 mg/g) and caffeine (10.8 mg/g vs 9.7 mg/g) were extracted AUTHOR CONTRIBUTIONS
under ultrasonic condition compared with no ultrasonic treatment X. Chen conceived and designed the study, collected and
(P < 0.05), which was associated with the higher DPPH scav- interpreted the test data, drafted and revised the manuscript. J.
enging capacity (404.0 µmol Trolox/g vs 379.4 µmol Trolox/g). Ding, D. Ji, and S. He collected and analyzed the test data. H.
Green extraction techniques such as accelerated solvent, Ma gave critical suggestions.
ultrasound-assisted, microwave-assisted, high hydrostatic pressure,
subcritical fluid, supercritical fluid, and enzyme-assisted extraction CONFLICT Of INTEREST
have been developed for the extraction of bioactive compounds The authors have no conflicts of interest to declare.
(Engmann, Ma, Tchabo, Ma, & Zhang, 2014; Luo et al., 2018a;
Luo, Cui, Zhang, & Duan, 2018b; Tchabo, Ma, Engmann, & REFERENCES
Zhang, 2015). Among these, ultrasound-assisted extraction (UAE) Ashengroph, M., Nahvi, I., & Amini, J. (2013). Application of Taguchi design and response
surface methodology for improving conversion of isoeugenol into vanillin by resting cells of

Food Chemistry
is considered as a more economical, and competent extraction Psychrobacter sp. CSW4. Iranian Journal of Pharmaceutical Research, 12(3), 411–421.
method because of its simplicity, low-cost, rapidity, and low Ayim, I., Ma, H., & Alenyorege, E. A. (2018). Optimizing and predicting degree of hydrolysis
of ultrasound assisted sodium hydroxide extraction of protein from tea (Camellia sinensis L.)
solvent volumes (Vilkhu, Mawson, Simons, & Bates, 2008). UAE residue using response surface methodology. Journal of food science and technology, 55(12), 5166–
has been widely utilized for the extraction of antioxidant phenolic 5174. https://doi.org/10.1007/s13197-018-3407-4.
Belwal, T., Dhyani, P., Bhatt, I. D., Rawal, R. S., & Pande, V. (2016). Optimization ex-
compounds from various teas (Both, Chemat, & Strube, 2014; traction conditions for improving phenolic content and antioxidant activity in Berberis
Li, Raza, Wang, Xu, & Chen, 2017); phenolic compounds from asiatica fruits using response surface methodology (RSM). Food Chemistry, 207, 115–124.
https://doi.org/10.1016/j.foodchem.2016.03.081.
red sorghum barn (Luo et al., 2018a), and phytochemicals from Both, S., Chemat, F., & Strube, J. (2014). Extraction of polyphenols from black tea - Con-
mulberry (Tchabo et al., 2015). ventional and ultrasound assisted extraction. Ultrasonics Sonochemistry, 21(3), 1030–1034.
https://doi.org/10.1016/j.ultsonch.2013.11.005.
Acoustic cavitation and the mechanical effect produced by Campa, C., Mondolot, L., Rakotondravao, A., Bidel, L. P. R., Gargadennec, A., Couturon, E.,
ultrasound wave can accelerate solvent penetrating into the . . . Davis, A. P. (2012). A survey of mangiferin and hydroxycinnamic acid ester accumulation
plant tissue and increase the contact surface area between in coffee (Coffea) leaves: Biological implications and uses. Annals of Botany, 110(3), 595–613.
https://doi.org/10.1093/aob/mcs119.
sample and solvent, thus enhance extraction efficiency (Ghafoor, Chen, X., Mu, K., & Kitts, D. D. (2019). Characterization of phytochemical mix-
Choi, Jeon, & Jo, 2009; Zou, Jia, Li, Wang, & Wu, 2013). tures with inflammatory modulation potential from coffee leaves processed by green
and black tea processing methods. Food Chemistry, 2019(271), 248–258. https://doi.
Although the ultrasonic treatment did not significantly enhance org/10.1016/j.foodchem.2018.07.097.
the phytochemical yield and antioxidant activity of coffee leaf Chen, X. M. (2019). A review on coffee leaves: Phytochemicals, bioactivities
and applications. Critical Reviews in Food Science and Nutrition, 59(6), 1008–1025.
extract, it did increase 5-CQA and DPPH scavenging capacity https://doi.org/10.1080/10408398.2018.1546667.
in the confirmation experiment. Therefore, in further study, Chen, X. M., Kitt, D. D., Ji, D. Y., & Ding, J. (2019). Free radical scavenging activ-
ities of phytochemical mixtures and aqueous methanolic extracts recovered from pro-
more ultrasonic conditions such as power, frequency and the cessed coffee leaves. International Journal of Food Science and Technology, 54(10), 2872–2879.
combination of different frequency can be investigated to further https://doi.org/10.1111/ijfs.14099.
Chen, X. M., Ma, Z., & Kitts, D. D. (2018). Effects of processing method and age of leaves
increase the yield under ultrasound treatment. on phytochemical profiles and bioactivity of coffee leaves. Food Chemistry, 249, 143–153.
https://doi.org/10.1016/j.foodchem.2017.12.073.
Chen, X. M., Tait, A. R., & Kitts, D. D. (2017). Flavonoid composition of orange peel and
its association with antioxidant and anti-inflammatory activities. Food Chemistry, 218, 15–21.
https://doi.org/10.1016/j.foodchem.2016.09.016.
4. CONCLUSION Dash, R. N., Mohammed, H., & Humaira, T. (2016). An integrated Taguchi and response surface
methodological approach for the optimization of an HPLC method to determine glimepiride
An ultrasound-assisted extraction method has been developed in a supersaturatable self-nanoemulsifying formulation. Saudi Pharmaceutical Journal, 24(1),
92–103. https://doi.org/10.1016/j.jsps.2015.03.004.
for extracting phytochemicals from coffee leaves using the Derringer, G., & Suich, R. (1980). Simultaneous optimization of several response variables. Jour-
combination of Taguchi design and Box–Behnken Design. nal of Quality Technology, 12, 214–219. https://doi.org/10.1080/00224065.1980.11980968.
Taguchi design was effectively employed to screen the significant Engmann, F. N., Ma, Y. K., Tchabo, W., Ma, H., & Zhang, H. N. (2014). Optimization of
ultrasonic and high hydrostatic pressure conditions on quality parameters of mulberry (Morus
factors including L:S ratio, ethanol concentration and extraction Moraceae) juice using response surface methodology. Journal of Food Quality, 37(5), 297–308.
temperature from the non-significant factors such as ultrasound https://doi.org/10.1111/jfq.12100.
FAOSTAT (2017). Harvest year 2017, Retrieved from http://www.fao.org/faostat/
power, extraction time and age of leaves. The RSM was success- en/#data/QC.
fully used to optimize the extraction conditions and the optimum Ghafoor, K., Choi, Y. H., Jeon, J. Y., & Jo, I. H. (2009). Optimization of ultrasound-
assisted extraction of phenolic compounds, antioxidants, and anthocyanins from grape
included L:S ratio = 30.3:1, ethanol concentration = 54.5%, and (Vitis vinifera) seeds. Journal of Agricultural and Food Chemistry, 57(11), 4988–4994.
temperature = 80 °C coupling with ultrasound power of 210W https://doi.org/10.1021/jf9001439.
Jyotshna, Khare, P., & Shanker, K. (2016). Mangiferin: A review of sources and interventions
and extraction time for 10 min. Under the optimal conditions, for biological activities. Biofactors, 42(5), 504–514. https://doi.org/10.1002/biof.1308.
TPC, mangiferin, rutin, and DPPH scavenging capacity were Kenconojati, H., Ulkhaq, M. F., Budi, D. S., & Azhar, M. H. (2019). In vitro study of an ethanolic
62.1 mg GA/g, 4.1 mg/g, 11.4 mg/g, and 356.9 µmol Trolox/g, extract of coffea leaves to inhibit freshwater pathogenic bacteria. In The 1st International
Conference on Fisheries and Marine Science (Vol. 236, pp. 01208). Bristol, UK: IOP Publishing.
respectively. The results indicated that combination of Taguchi Li, F., Raza, A., Wang, Y. W., Xu, X. Q., & Chen, G. H. (2017). Optimization of surfactant-
mediated, ultrasonic-assisted extraction of antioxidant polyphenols from rattan tea (Ampelopsis
design and RSM is an effective method for the extraction of grossedentata) using response surface methodology. Pharmacognosy Magazine, 13(51), 446–453.
bioactive components from coffee leaves. https://doi.org/10.4103/pm.pm_159_16.

Vol. 0, Iss. 0, 2020 r Journal of Food Science 9

 Optimizing extraction of coffee leaf . . .

Lundstedt, T., Seifert, E., Abramo, L., Thelin, B., Nystrom, A., Pettersen, J., & Bergman, R. Zhang, H. N., & Ma, Y. K. (2017). Optimisation of High Hydrostatic Pressure Assisted Extraction
(1998). Experimental design and optimization. Chemometrics and Intelligent Laboratory Systems, of Anthocyanins from Rabbiteye Blueberry Pomace. Czech Journal of Food Sciences, 35(2), 180–
42, 3–40. https://doi.org/10.1016/S0169-7439(98)00065-3. 187. https://doi.org/10.17221/189/2016-CJFS.
Luo, X., Cui, J., Zhang, H., Duan, Y., Zhang, D., Cai, M., & Chen, G. (2018a). Ultrasound Zou, T. B., Jia, Q., Li, H. W., Wang, C. X., & Wu, H. F. (2013). Response Surface Methodology
assisted extraction of polyphenolic compounds from red sorghum (Sorghum bicolor L.) bran and for Ultrasound-Assisted Extraction of Astaxanthin from Haematococcus pluvialis. Marine
their biological activities and polyphenolic compositions Industrial. Crops and Products, 112, Drugs, 11(5), 1644–1655. https://doi.org/10.3390/md11051644.
296–304. https://doi.org/10.1016/j.indcrop.2017.12.019.
Luo, X. P., Cui, J. M., Zhang, H. H., & Duan, Y. Q. (2018b). Subcritical water extraction
of polyphenolic compounds from sorghum (Sorghum bicolor L.) bran and their biological
activities. Food Chemistry, 26214–20. https://doi.org/10.1016/j.foodchem.2018.04.073.
Marcheafave, G. G., Tormena, C. D., Pauli, E. D., Rakocevic, M., Bruns, R. E., &
Scarminio, I. S. (2019). Experimental mixture design solvent effects on pigment extraction
and antioxidant activity from Coffea arabica L. leaves. Microchemical Journal, 146, 713–721.
https://doi.org/10.1016/j.microc.2019.01.073 Supporting Information
Montgomery, D. C. (2013). Response surface methods and designs. In D. C. Montgomery (Ed.),
Design and analysis of experiments (8th Edition ed., pp. 478–553). Hoboken, NJ: John Wiley & Additional supporting information may be found online in the
Sons Inc. Supporting Information section at the end of the article.
Patay, E. B., Bencsik, T., & Papp, N. (2016). Phytochemical overview and medicinal importance
of Coffea species from the past until now. Asian Pacific Journal of Tropical Medicine, 9(12),
1127–1135. https://doi.org/10.1016/j.apjtm.2016.11.008.
Table S1. Ranks of each factor on the responses in Taguchi design.
Rajha, H. N., Darra, N. E., Hobaika, Z., Boussetta, N., Bvorobiev, E., Maroun, R. G., &
Louka, N. (2014). Extraction of total phenolic compounds, flavonoids, anthocyanins and Table S2. Analysis of Variance for trigonelline in RSM.
tannins from grape byproducts by response surface methodology. Influence of solid-liquid
ratio, particle size, time, temperature and solvent mixtures on the optimization process. Food Table S3. Analysis of variance for caffeine in RSM.
and Function Scicences, 5, 397–409. https://doi.org/10.4236/fns.2014.54048.
Ross, I. A. (2005). Cannabis arabica. Medicinal Plants of the World (pp. 155–184). Hoboken, NJ: Table S4. Analysis of variance for 3-CQA in RSM.
Humana Press.
Santos, J. S., Deolindo, C. T. P., Esmerino, L. A., Genovese, M. I., Fujita, A., Marques, M. B.,
. . . Granato, D. (2016). Effects of time and extraction temperature on phenolic composition
Table S5. Analysis of variance for 4-CQA in RSM.
and functional properties of red rooibos (Aspalathus linearis). Food Research International, 89,
476–487. https://doi.org/10.1016/j.foodres.2016.08.041. Table S6. Analysis of variance for 5-CQA in RSM.
Segheto, L., Santos, B. C. S., Werneck, A. F. L., Vilela, F. M. P., de Sousa, O. V., & Rodarte, M.
Food Chemistry

P. (2018). Antioxidant extracts of coffee leaves and its active ingredient 5-caffeoylquinic acid Table S7. Analysis of variance for 3,4-diCQA in RSM.
reduce chemically-induced inflammation in mice. Industrial Crops and Products, 126, 48–57.
https://doi.org/10.1016/j.indcrop.2018.09.027. Table S8. Analysis of variance for 3,5-diCQA in RSM.
Tchabo, W., Ma, Y., Engmann, F. N., & Zhang, H. (2015). Ultrasound-assisted enzymatic
extraction (UAEE) of phytochemical compounds from mulberry (Morus nigra) must and
optimization study using response surface methodology. Industrial Crops and Products, 63, Table S9. Analysis of variance for 4,5-diCQA in RSM.
214–225. https://doi.org/10.1016/j.indcrop.2014.09.053.
Thakur, R., Das, D., & Das, A. (2016). A Combined Taguchi and Response Surface Approach Table S10. Analysis of variance for mangiferin in RSM.
to Improve Electret Charge in Corona-charged Fibrous Electrets. Fibers and Polymers, 17(11),
1790–1800. https://doi.org/10.1007/s12221-016-6312-3. Table S11. Analysis of variance for rutin in RSM.
Vilkhu, K., Mawson, R., Simons, L., & Bates, D. (2008). Applications and opportunities for
ultrasound assisted extraction in the food industry - A review. Innovative Food Science and Table S12. Analysis of variance for TPC in RSM.
Emerging Technologies, 9(2), 161–169. https://doi.org/10.1016/j.ifset.2007.04.014.
Yang, L., Jiang, J. G., Li, W. F., Chen, J., Wang, D. Y., & Zhu, L. (2009). Opti-
mum extraction Process of polyphenols from the bark of Phyllanthus emblica L. based Table S13. Analysis of variance for DPPH scavenging capacity in
on the response surface methodology. Journal of Separation Science, 32(9), 1437–1444. RSM.
https://doi.org/10.1002/jssc.200800744.
Yilmaz, Y., & Toledo, R. T. (2006). Oxygen radical absorbance capacities of grape/wine industry Figure S1. Contour plots for the effects of L:S ratio, ethanol
byproducts and effect of solvent type on extraction of grape seed polyphenols. Journal of Food
Composition and Analysis, 19(1), 41–48. https://doi.org/10.1016/j.jfca.2004.10.009. concentration, and temperature on different responses.

10 Journal of Food Science r Vol. 0, Iss. 0, 2020