Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
ABSTRACT: Many methods are available for determining food antioxidant capacity, which is an important topic
in food and nutrition research and marketing. However, the results and inferences from different methods may vary
substantially because each complex chemical reaction generates unique values. To get a complete and dynamic pic-
ture of the ranking of food antioxidant capacity, relative antioxidant capacity index (RACI), a hypothetical concept,
is created from the perspective of statistics by integrating the antioxidant capacity values generated from different
in vitro methods. RACI is the mean value of standard scores transformed from the initial data generated with differ-
ent methods for each food item. By comparing the antioxidant capacity of 20 commonly consumed vegetables in the
U.S. market that were measured with 7 chemical methods, we demonstrated that the RACI correlated strongly with
each method. The significant correlation of RACI with an independent data set further confirmed that RACI is a valid
tool to assess food antioxidant capacity. The key advantage of this integrated approach is that RACI is in a numerical
scale with no units and has consistent agreement with chemical methods. Although it is a relative index and may
not represent a specific antioxidant property of different food items, RACI provides a reasonably accurate rank of
antioxidant capacity among foods. Therefore, it can be used as an integrated approach to evaluate food antioxidant
capacity.
Keywords: antioxidant capacity, electron transfer, hydrogen atom transfer, vegetables
C 2007 Institute of Food Technologists Vol. 72, Nr. 9, 2007JOURNAL OF FOOD SCIENCE R159
doi: 10.1111/j.1750-3841.2007.00552.x
Further reproduction without permission is prohibited
R: Concise Reviews in Food Science Evaluating food antioxidant capacity . . .
results from standardized methods may vary substantially as dis- Demonstration of the Concept with Published Data
parate complex chemical reactions may induce different conclu- on Vegetable Antioxidant Capacity
sions when comparing antioxidant capacity. Indeed, a standard
relationship between methods may not exist (Prior and others Antioxidants and antioxidant capacity of vegetables
2005). Vegetables contain several types of antioxidants. Examples in-
clude anthocyanins, apigenin, apigenin glycoside, ascorbic acid, -
The Hypothetical Concept of Relative carotene, p-coumaric acid, caffeic acid, chlorogenic acid, cinnamic
Antioxidant Capacity Index acid, ferulic acid, p-hydroxybenzoic acid, kaempferol, kaempferol
glycoside, lutein, luteolin, luteolin glycoside, myricetin, quercetin,
A new concept, relative antioxidant capacity index (RACI), is pro-
posed herein from the perspective of statistics by integrating
the food antioxidant capacity data determined by several methods.
quercetin glycoside, and -tocopherol (Sun and Powers 2007). Both
the types and amounts of antioxidants differ among vegetables.
This concept not only allows for the comparison of food antioxi- Several methods have been used to measure the antioxidant capac-
dant capacity derived from different chemical methods, but also ity of vegetables. The data on antioxidant capacity of 20 commonly
provides a more comprehensive comparison. If measured with 1 consumed vegetables in the U.S. market were collected from pub-
chemical method, the relative ranking of the antioxidant capacity lished reports (Table 1) (Cao and others 1996; Vinson and others
between foods is only based on a single indicator. However, no sin- 1998; Ou and others 2002; Proteggente and others 2002; Pellegrini
gle chemical method can represent total food antioxidant capacity. and others 2003; Wu and others 2004; Eberhardt and others 2005;
To determine the antioxidant capacity of food items and the rela- Sun and others 2007). The antioxidant capacity of these vegeta-
tionship between foods for multiple assays, the most widely used bles was determined using 7 HAT and/or ET methods: ABTS, FRAP,
measure is central tendency (Zar 1996), where foods are ranked ORAC (Cu2+ ), ORAC (OH r), ORAC (ROO r), phenol antioxidant in-
based on the mean value for each food. However, it is impossible dex (PAOXI), and TRAP (Table 1).
to get a mean value for each food as the units and the scale of the
data from various chemical methods are different. If the values of
Brief review of methods
antioxidant capacity in each data set are transformed into standard The ABTS method measures the decrease of the absorbance at
scores, the data can be compared. In statistics, a standard score 734 nm of the ABTS radical, which is produced by the reaction
(also called z- or normal score) is a dimensionless quantity derived of ABTS, metmyoglobin, H 2 O 2 , and peroxidase, or between ABTS
by subtracting the mean from the raw data divided by the standard and persulfate. The ABTS method can determine both hydrophilic
deviation (Sharma 1996). The standard score is thus calculated as and lipophilic antioxidants in the same sample (Arnao and oth-
follows: ers 2001). The FRAP method measures the reduction of ferric 2, 4,
6-tripyridyl-s-triazine (TPTZ) to a colored product (Fe3+ -TPTZ to
x Fe2+ -TPTZ). FRAP can be affected by several factors such as inter-
ference from compounds that absorb at the wavelength of analy-
sis and reaction kinetics. In the ORAC (ROO r) method, the peroxyl
where x is the raw data, is the mean, and is the standard devia- radical 2,2 -azobis-(2-amidinopropane) dihydrochloride (AAPH)
tion (Sharma 1996). reacts with a fluorescent compound (fluorescein) to form a non-
Subtracting the mean centers the distribution and dividing the fluorescent product, but antioxidants can scavenge the peroxyl
difference by the standard deviation normalizes the distribution. radicals and retard the loss of the fluorescence; thus the area
Standard scores have a mean of 0 and a standard deviation equal under the timefluorescence curve can be used to quantify food an-
to 1. The standard score represents the distance between the raw tioxidant capacity (Cao and others 1993). ORAC (Cu2+ ) and ORAC
data and the mean in units of the standard deviation, which is (OH r) methods use Cu2+ and OH r as the radical, respectively.
negative when the raw data are smaller than the mean and pos- PAOXI is calculated by dividing the total phenol concentration of
itive when larger. The standard score reveals how many units a a vegetable (mol/kg) from the FolinCiocalteau assay by the half
case is smaller or larger than the mean and allows the comparison maximal inhibitory concentration (IC 50 ) in M from the inhibi-
of results with different units or with the same units but different tion assay of LDL oxidation. The FolinCiocalteau method deter-
normal distributions (Sharma 1996). The standard score transfor- mines the total phenolic content of the sample by monitoring the
mation is especially useful when seeking to compare the relative absorbance change during a color reaction. The LDL method mea-
standings of items from distributions with different means and/or sures the antioxidants ability to inhibit the auto-oxidation of LDL
standard deviations. The antioxidant capacity data generated by caused by Cu2+ or the AAPH radical (Handelman and others 1999).
different methods have different units and/or scales, and there is The TRAP method evaluates the power of antioxidants to interfere
no standard relationship between these methods. Thus, the direct with the reaction between peroxyl radicals generated by AAPH or 2,
comparison of antioxidant capacity of different foods using varying 2-azobis (2-amidinopropane) dihydrochloride (ABAP) and a target
methods is impossible (Prior and others 2005). To overcome these probe (Ghiselli and others 1995).
difficulties, RACI of each food item was created by averaging the
standard scores transformed from the raw data generated with dif- Development and validation of the
ferent chemical methods. After data transformation, the standard relative antioxidant capacity index
scores (without any units) from various methods with different dis- The antioxidant capacity of common vegetables determined by
tributions will have similar normal distributions with a mean of 0 7 chemical methods in published references was used to develop
and variance of 1. Therefore, the standard scores with the same RACI. For a vegetable with a range of data values, a mean value
scale from data generated with different methods will have simi- was calculated to represent the antioxidant capacity determined
lar contribution to the central tendency of the mean and can be by each method. The units and scales of the raw data were differ-
compared without interference from different units, scales, and dis- ent (Table 1). The ranking of antioxidant capacity of vegetables dif-
tributions. The following describes how to use RACI to compare fered between methods. Data from FRAP and ORAC (ROO r) meth-
antioxidant capacity of vegetables. ods had significant correlation with the data of most methods,
Table 2 --- Correlation coefficients between different methods for analyzing antioxidant capacity.
Methods ABTS FRAP ORAC (Cu2+ ) ORAC (OH r) ORAC (ROO r) PAOXI TRAP
ABTS 1.00
FRAP 0.93 1.00
ORAC (Cu2+ ) 0.37 0.49 1.00
ORAC (OH r) 0.53
0.54 0.20 1.00
ORAC (ROO r) 0.51
0.91
0.64
0.60 1.00
PAOXI 0.15 0.21 0.43 0.03 0.29 1.00
TRAP 0.65 0.74 0.54 0.39 0.85 0.10 1.00
The minimum, maximum, and 50% median standard scores were dependently used to validate RACI. In their research, a modified-
1.10, 1.90, and 0.20 for ABTS; 1.00, 2.40, and 0.40 for FRAP; FRAP method that determines the antioxidant capacity of both
0.70, 2.80, and 0.45 for ORAC (Cu2+ ); 1.20, 2.10, and 0.30 for water- and lipid-soluble antioxidants was used to determine the
ORAC (OH r); 1.10, 2.90, and 0.30 for ORAC (ROO r); 0.60, 2.90, antioxidant capacity of vegetables and the result was expressed as
and 0.40 for PAOXI; and 1.00, 2.50, and 0.30 for TRAP, respec- total antioxidant content. The RACI developed herein showed sig-
tively. The standardized scores by definition had 100% correlation nificant linear relationship with the total antioxidant content (r =
with the corresponding raw data from each chemical method. 0.75, P < 0.01; Figure 3). The scientific value of RACI is therefore
The RACI of each vegetable was calculated as the mean of stan- enhanced by its significant correlation with this independent data
dard scores transformed from the raw data generated with dif- set. Thus, the RACI is a valid approach to rank food antioxidant
ferent chemical methods. The difference in units and variances capacity.
in the raw data had no influence on the RACI. Stepwise regres-
sion between RACI and different chemical methods revealed that Method recommendations
(1) each of the 7 chemical assays was selected as a significant Ideally, the more chemical methods used in developing RACI,
variable with no single chemical method being removed, (2) each the closer RACI represents the ranking (high or low) of the food an-
chemical method contributed the same weight (1/7) in build- tioxidant capacity. In many situations, due to limitations of instru-
ing RACI, and (3) the regression was highly significant (r = 1, mentation, time, and budget, it is not practical to measure antioxi-
P < 0.001). Therefore, RACI of each vegetable is a scientific com- dant capacity of food items with more than 3 chemical methods. To
bination of data from different chemical methods with no unit choose a method to determine food antioxidant capacity, the types,
limitation and no variance among methods, and makes compar- and concentrations of antioxidants in the food, the type of sub-
ison of food antioxidant capacity possible and perhaps more ac- strates, and analysis conditions need to be considered. RACI can be
curate. Based on the RACI value of the 20 common vegetables, the used as a reference to choose a chemical assay that can better rep-
rank of antioxidant capacity (Figure 1) from the highest to the low- resent the food antioxidant status. Based on the concept of RACI,
est was garlic with a value of 1.71 to celery with a value of 0.88. the statistically more sensitive combined methods, which had high
Using RACI, garlic, spinach, and asparagus have high antioxi- correlation coefficients with RACI as well as a higher percentage of
dant capacity while celery, cucumber, carrot, and squash have low significant correlation with methods not used to develop RACI, are
antioxidant capacity. The trend of the RACI value matched with the recommended for determining antioxidant capacity (Table 4). RACI
standard score from the 7 methods (Figure 2). Furthermore, the cor- developed from the combination of ORAC (ROO r), FRAP, and ABTS,
relation analysis demonstrated that the RACI value correlated with PAOXI, or TRAP, had high correlation coefficients of 0.93, 0.95, or
each method; the correlation coefficients of RACI with ABTS, FRAP, 0.94, respectively, and correlated with other methods with the high-
ORAC (Cu2+ ), ORAC (OH r), ORAC (ROO r), PAOXI, and TRAP were est percentage (75%), compared with the other 3-method combina-
0.79, 0.84, 0.76, 0.56, 0.86, 0.44, and 0.87, respectively. Therefore, tions.
this novel approach is proven to be an appropriate way to quan- Due to a variety of reasons, 1 method may be justified. For deter-
tify food antioxidant capacity as it can represent each chemical mining antioxidant capacity using a single method, ORAC (ROO r),
method. the most statistically sensitive method in developing RACI, showed
Halvoren and others (2006) studied the antioxidant capacity of the highest percentage of significant correlation with other meth-
16 of the 20 vegetables used in the development of RACI (ex- ods (r = 0.83) and high correlation with RACI (86%). Further-
cludes cauliflower, beet, spinach, and garlic). These data were in- more, ORAC (ROO r) was recommended as a standard method by
other 6 methods (67%). For a 2-method combination, ORAC (ROO) chemical methods measuring different antioxidant properties may
and FRAP are recommended because RACI developed by these 2 reflect a more accurate RACI. Indeed, the value of RACI might be
methods had the highest correlation (r = 0.95) and the highest per- enhanced if the number and type of antioxidant methods was de-
centage of correlation with other methods (80%). fined and fixed. Furthermore, measuring antioxidant capacity of
food items with multiple chemical methods is not always practi-
Limitations of the relative antioxidant capacity index cal. Although one can collect data from independent experiments,
RACI correlated with each chemical assay and may be used as which is the case of the present analyses, assumptions are made
a reference for measuring food antioxidant capacity. As a simple that experimental conditions are similar and experimental errors
numerical approach with a relative value, RACI has its own weak- are ignored for the same chemical method.
nesses. First, RACI is not a fixed value and is thus subject to ma- When assessing health benefits from food antioxidants in bio-
nipulation by the methods it includes. These methods may differ logical systems, the combined effects of complex bioactive com-
substantially in chemical mechanism and thus contribute different pounds need to be considered (Eberhardt and others 2000). The
proportions to RACI. It may also change as different food items or RACI in the present study was developed from in vitro chemical
the same food with different genotypes are added or removed from assays. To achieve a more powerful RACI, more tests of food an-
the testing list, which affects the standard score of each food item. tioxidant capacity in real biological systems, particularly humans,
Second, although RACI was developed from chemical assays, it can- should be included to represent the bioavailability of antioxidants
not directly reflect a single specific chemical mechanism that is in- and their clinical effectiveness (Collins 2005). An ideal RACI would
volved in the antioxidant reactions of a selected food item. It is a be developed from data generated from both in vitro and in vivo
combination of transformed data generated from several mecha- assays with equal contribution from assays covering all antioxidant
nisms into a standard score. Food items with the same RACI may properties. If 1 special group of antioxidants is of interest, such as
have different antioxidants in quality and/or quantity. Third, even lipophilic or hydrophilic antioxidants, RACI can be developed by
though RACI was developed from 7 chemical assays, it may not methods that cover the properties of that special group.
cover all the chemical aspects of antioxidant reactions. A variety of
Conclusions
G
iven the fact that food has different classes of antioxidants
and each method to measure the antioxidant capacity has its
own limitations, RACI provides an alterative approach to integrate
several chemical assays. Although RACI may not directly represent
any individual chemical mechanism, which is the traditional way to
measure and define antioxidant capacity, it does correlate with re-
sults from different chemical assays. The key advantage of RACI is
that it is a numerical scale integrating multiple chemical methods,
allowing comparison of antioxidant capacity of a large number of
foods. RACI goes beyond chemical reactions and ranks antioxidant
capacity without the restriction of mechanism. Although RACI does
not allow avoidance of performing chemical assays, it is an alterna-
tive assessment to help standardize methods. RACI is a relative food
Figure 3 --- Regression (r = 0.75, P < 0.01) between rela- antioxidant capacity ranking tool that presents the antioxidant ca-
tive antioxidant capacity index (RACI) and total antiox-
idant content measured with a modified-ferric reducing pacity of each food in a standardized format for use by the food in-
antioxidant power assay in 16 vegetables. dustry, scientists, and consumers.
Table 4 --- Recommended chemical methods used to evaluate food antioxidant capacity.
Numbers of methods used to develop RACI
1 2 3 3 3
Recommended methods ORAC (ROO r) ORAC (ROO r) ORAC (ROO r) ORAC (ROO r) ORAC (ROO r)
FRAP FRAP FRAP FRAP
ABTS PAOXI TRAP
Correlation coefficient (r )a 0.86 0.95 0.93 0.95 0.94
Methods tested ABTS ABTS ORAC (Cu2+ ) ABTS ABTS
FRAP ORAC (Cu2+ ) ORAC (OH r) ORAC (Cu2+ ) ORAC (Cu2+ )
ORAC (Cu2+ ) ORAC (OH r) PAOXI ORAC (OH r) ORAC (OH r)
ORAC (OH r) PAOXI TRAP TRAP PAOXI
PAOXI TRAP
TRAP
Methods with significant correlation ABTS ABTS ORAC (Cu2+ ) ABTS ABTS
FRAP ORAC (Cu2+ ) ORAC (OH r) ORAC (Cu2+ ) ORAC (Cu2+ )
ORAC (Cu2+ ) ORAC (OH r) TRAP TRAP ORAC (OH r)
ORAC (OH r) TRAP
TRAP
Methods without significant correlation PAOXI PAOXI PAOXI ORAC (OH r) PAOXI
Number of methods with significant correlation 5 4 3 3 3
Number of methods without significant correlation 1 1 1 1 1
Significant correlation (%) 83.3 80.0 75.0 75.0 75.0
a
Correlation with relative antioxidant capacity index (RACI) developed from 7 chemical methods.