Food Chemistry 143 (2014) 4047

Contents lists available at ScienceDirect

Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem

Discrimination of commercial cheeses from fatty acid proles and

phytosterol contents obtained by GC and PCA
Nam Sook Kim, Ji Hyun Lee, Kyoung Moon Han, Ji Won Kim, Sooyeul Cho, Jinho Kim
Advanced Analysis Team, Ministry of Food and Drug Safety, Osong Health Technology Administration Complex, Chungcheongbuk-do 363-700, South Korea

a r t i c l e

i n f o

Article history:
Received 12 April 2013
Received in revised form 27 June 2013
Accepted 18 July 2013
Available online 27 July 2013
Keywords:
Fatty acid composition
Phytosterols
Cheeses
Derivatization
GC
GCMS
PCA

a b s t r a c t
In this study, a method for discriminating natural mozzarella cheese from cheese substitutes, using fatty
acid proles, phytosterol contents, and statistical comparison, was developed. A total of 27 cheeses were
evaluated: eight natural mozzarella cheeses (NMCs), four imitation mozzarella cheeses (IMCs), 12 processed cheeses (PCs) and three mixed cheeses (MCs) composed of NMCs and IMCs. The fatty acid composition of the NMC class was distinct from those of the IMC and MC classes, but statistically similar
(p < 0.05) to that of the PC class. The phytosterol content of the NMC class, determined via gas chromatographymass spectrometry, was distinct from the IMCs, but similar (p < 0.05) to a portion of the PCs.
Principal component analysis (eigenvalue P 1) indicated that the NMCs can be differentiated from the
IMCs, but discrimination between the NMCs and the PCs could not be achieved.
2013 Elsevier Ltd. All rights reserved.

1. Introduction
In recent years, the demand for natural cheese, processed
cheese, and cheese substitutes has been ever increasing (Jana &
Mandal, 2011; Jana, Patel, Suneeta, & Prajapati, 2010). As dened,
natural cheeses are made from milk or milk products, whereas processed cheese must contain more than 50% milk solids, and cheese
substitutes, also called imitation cheeses, are generally manufactured from non-dairy fats or proteins to make cheese-like products
(Bachmann, 2001; Jana & Mandal, 2011; Ntakatsane, Liu, & Zhou,
2013). In this study, mixed cheese is simply a mixture of natural
cheese mixed with imitation mozzarella cheese. This type of
cheese product, commonly used as a pizza topping in Korea, is relatively cheap and has high storage stability (Jana et al., 2010; Korea
Food, 2012; Ntakatsane et al., 2013). In addition to the three broad
classes listed above, cheeses can be further segmented by fat content and other characteristics.
Mixed cheese and imitation mozzarella cheese consumption is
on the rise because they are easily prepared and cost considerably
less than natural mozzarella cheese, as milk is substituted with relatively inexpensive vegetable products. Mozzarella cheese substitutes are used widely, especially in pizza and related products,
and the lower quality has the potential to impact human health

Corresponding author. Tel.: +82 43 719 5303; fax: +82 43 719 5300.
E-mail addresses: kimjinho1@yahoo.com, heide2@empas.com (J. Kim).
0308-8146/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.foodchem.2013.07.083

(Pellegrino, Resmini, De Noni, & Masotti, 1996). Currently, it is difcult to discriminate between natural and imitation mozzarella
cheeses (Middleton, 1989), as analytical differentiation methods
are lacking (Pellegrino et al., 1996).
Cheese is composed of many chemicals, some of which may be
used as markers. Furosine, formed in the initial stages of the Maillard reaction, is a marker for lactose (Resmini, Pellegrino, & Massotti, 1993), and lysinoalanine, which occurs in commercial
caseinate and milk, can be used to differentiate between natural
and imitation mozzarella cheeses (Pellegrino et al., 1996). Additionally, fatty acid composition can be compared to help differentiate cheese products (Ntakatsane et al., 2013; Prandini, Sigolo, &
Piva, 2011). To date, there have been few studies investigating phytosterols, except for studies on the nutritional benets of phytosterol-enriched products (Garca-Llatas & Rodrguez-Estrada, 2011).
Chemometrics is a combination of logic-based methods, such as
mathematics and statistics, to effectively handle and interpret
large amounts of chemical data (Haswell, 1992), and principal
component analysis (PCA) is a commonly used multivariate chemometric method. PCA can be used to identify patterns and to explain similarities and differences in data. Also, PCA can be used to
show relationships that exist between objects and arbitrary principal components (Kadegowda, Piperova, & Erdman, 2008; Matos
et al., 2007). Recently, Ntakatsane et al. (2013) used PCA to efciently screen milk products for adulteration.

N.S. Kim et al. / Food Chemistry 143 (2014) 4047

In this study, a method to discriminate natural mozzarella

cheese from cheese substitutes is reported. Specically, natural,
imitation, processed, and mixed cheeses were evaluated based on
fatty acid proles and phytosterol contents via gas chromatography-ame ionisation detector (GC-FID) and GCmass spectrometry
(GCMS) experiments, and PCA data analysis.

2. Materials and methods

2.1. Materials and chemicals
2.1.1. Samples
Cheese and cheese products were purchased from local supermarkets (Chungcheongbuk-do, Korea) and on-line (http://itempage3.auction.co.kr). In total, eight natural mozzarella cheeses
(NMCs), four imitation mozzarella cheeses (IMCs), 12 processed
cheeses (PCs), and three mixed cheeses (MCs), composed of NMCs
and IMCs, were evaluated. The cheeses, described in Table 1, were
stored at 50 C and defrosted at room temperature prior to
analysis.
2.1.2. Standards and reagents
PUFA-2 (Supelco, Bellefonte, USA), a fatty acid sourced from animals, and undecanoic acid (C11:0, Sigma, Saint Louis, USA), an
internal standard, were purchased. Brassicasterol (99%), campesterol (98%), stigmasterol (98%), b-sitosterol (78.6%) and 5a-cholestane (99.9%) were acquired from Sigma (Saint Louis, USA). All stock
1 mg/mL phytosterol solutions were prepared separately in ethanol and stored at 20 C. Working standards were prepared daily
by diluting stock solutions with ethanol. Butylated hydroxytoluene
(BHT; Sigma, Saint Louis, USA), sodium chloride (NaCl), potassium
hydroxide (KOH), chloroform, methanol, isooctane, ethanol, and nhexane were obtained from Merck (Darmstad, Germany), and
water was obtained from a Milli-Q Advantage A10 SYSTEM
(EMD Millipore Corporation, Billerica, USA). A 14% boron triuoridemethanol solution and N,O-bis-(trimethylsilyl)triuoroacetamide (BSTFA) with 1% trimethylchlorosilane (TMCS) were
purchased from SIGMA (Saint Louis, USA).
2.1.3. Instruments
To prepare samples, a PT 2100 homogenizer (Kinematica AG,
Luzern, Switzerland), a Vibracell-VCX 750 ultrasonicator (Sonics
& Materials Inc., Newton, USA), a 5810R Centrifuge (Eppendorf,
Hamburg, Germany), and a NE-1001 rotary evaporator (EYELA
co., ltd, Tokyo, Japan), were used. Analyses were carried out with
an Agilent 7890A GC System (Agilent Technologies Inc., Santa
Clara, USA) equipped with a 7693 autosampler and FID and an Agilent 7890A GC System (Agilent Technologies Inc., Santa Clara, USA)
interfaced with a 5975 mass-selective detector and a 7683 auto-

sampler, controlled by Chem Station (Agilent Technologies Inc.,

Santa Clara, USA).
2.2. Determination of fatty acid composition
2.2.1. Lipid extraction
A modied version of Folchs method (Christie, 1989) was used
for lipid extraction. To start, 2.5 g of cheese were added to a 25 mL
chloroformmethanol mixture (2:1, v/v). Butylated hydroxytoluene (0.001%), an antioxidant, was then added to the mixture. The
mixture was homogenised (30 mm polytron, 2500 rpm, 30 min),
ultrasonicated (Amplier 35%, 20 min), and 10 mL of saturated
NaCl solution were added. The suspension was then centrifuged
for 20 min at 4 C and 4000 rpm. The chloroform phase was
recovered and transferred into a round ask (25 mL), and each
fat extract was dried via rotary evaporator at 45 C under vacuum.
2.2.2. Preparation of fatty acid methyl esters
The extracted fats were esteried using a modied esterication method (Christie, 1989; Christie, Sbdio, & Juanda, 2001).
First, 1 mL of undecanoic acid (5.05 mg/mL to chloroform) was
added to each fat extract. Then 1.5 mL of 0.5 N methanolic NaOH,
a dissolving agent, were added. The ask was then shaken vigorously for 30 s. The mixture was transferred into a Teon-lined
screw-top test tube, heated at 100 C for 5 min, and cooled at
20 C for 3 min. Then, 2 mL of 14% methanolic BF3 were added,
and the reaction was allowed to proceed for 20 min at 100 C. After
the reaction, the reactant was cooled at 4 C; 2.5 mL of isooctane
and 5 mL of saturated NaCl were added to the reaction and the
mixture was stirred. Two phases appeared and were separated,
one of which contained fatty acid methyl esters (FAME). The hexane phase was ltered with a PTFE syringe lter (0.2 lm, Whatman, Kent, UK), dehydrated with anhydrous sodium sulphate
(Na2SO4), and the ltrate was diluted to half with isooctane for
GC analysis. The same methylation procedure was carried out on
200 lL of the PUFA-2 standard solution (10 mg/mL to chloroform)
to obtain FAMEs.
2.2.3. Gas chromatography
GC analysis was conducted in accordance with a previously described method (Prandini et al., 2011). Qualitative analysis of
FAMEs was carried out on a GC equipped with a FID and a SP2560 capillary column (100 m  0.25 mm i.d.; 0.25 mm lm thickness; Supelco, Bellefonte, USA). An initial oven temperature of
120 C was maintained for 1 min. Then, the temperature was increased to 200 C at a rate of 25 C/min and the oven temperature
was held at 200 C for 5 min. Then, the temperature was increased
at a rate of 2 C/min to reach 240 C, and retained for 15 min. FAME
(1 lL) was injected into the instrument using a split ratio of 100:1.
The injector port and detector temperatures were set at 240 C and
250 C, respectively. The column ow rate was 1.0 mL/min, and

Table 1
The list of commercial cheese samples.
Classication

41

Quantity

Material description from product labels

Natural mozzarella cheeses

NMC

Imitation mozzarella cheeses

IMC

Processed cheeses

PC

12

Mixed cheeses

MC

Mozzarella cheese 99%

Other ingredients 1%
Vegetable oil (palm oil, palm olein oil) 99%
Other ingredients 1%
Mozzarella cheese 7080%,
Vegetable oil (palm oil, palm olein oil) 1929%
Other ingredients 1%
Mozzarella cheese 5070%,
Imitated mozzarella cheeses 2949%
Other ingredients 1%

Other ingredients: rened sugar, starch, our, emulsier, rened salt, citric acid and so on.

42

N.S. Kim et al. / Food Chemistry 143 (2014) 4047

nitrogen was used as the carrier and make up gas. Identication of

fatty acids was achieved by comparing the relative retention times
with the internal and PUFA-2 standards. PUFA-2 standards included 14 fatty acids and allowed only for qualitative analysis.
Fatty acid compositions (mol%) were determined based on the relative chromatographic areas to compare only intergroup differences in the fatty acid compositions. All measurements were
performed in triplicate.
2.3. Determination of phytosterols
2.3.1. Sample preparation
Cheese (2.5 g) was mixed with 25 mL of chloroformmethanol
(2:1, v/v), and 50 lL of the internal standard solution. After extraction, as described above, the liquid phase was collected from the
mixture. The liquid phase was dried via rotary evaporator at
45 C in vacuo. Then 40 mL of ethanol and 10 mL of 0.1 N ethanolic
KOH were added, and saponication, at 95 C for 1 h, was carried
out according to the methods described in the Health Functional
Food Code (Korea Food & Drug Administration, 2012; Santos
et al., 2007). Saturated NaCl (5 mL) and 150 mL of n-hexane were
added and the fraction not saponied was extracted with a separation funnel. The separated hexane phase was ltered (Whatman,
circles 185 mm U), dehydrated with anhydrous sodium sulphate,
and dried by rotary evaporator at 45 C under vacuum. After dissolving the extract in 3 mL of ethanol, the mixture was ltered
with a PTFE syringe (0.2 lm) and dehydrated with anhydrous sodium sulphate. The ltrate was then dried under a gentle stream
of nitrogen at 60 C. The dry residue was derivatized with 500 lL
of BSTFA:TMCS (99:1) at 60 C for 30 min. Finally, 1 lL of the derivatized solution was analysed by GC (Schummer, Delhomme,
Appenzeller, Wennig, & Millet, 2009; Wu, Hu, Yue, Yang, & Zhang,
2009).
2.3.2. Gas chromatographymass spectrometry
The MS detector transfer line was kept at 280 C and tuning was
conducted on a daily basis with a peruorotributylamine (PFTBA)
standard composed of three masses (m/z 69, 219, 502). Samples
(1 lL) were injected automatically in split mode (10:1) at 300 C.
An Agilent J&W DB-5MS (crosslinked 95% dimethyl/5% phenyl
polysiloxane) capillary column (30 m  0.25 mm i.d., 0.25 lm lm
thickness; Agilent Technologies Inc., Santa Clara, USA) was used for
separation, and helium (99.9999%) was used as the carrier gas at a
ow rate of 1 mL/min. The column temperature was initially set to
200 C and then held for 1 min. Then, the column was heated to
280 C at a rate of 10 C/min, and held at 280 C for 11 min. The column was then heated to 300 C at a rate of 4 C/min and retained
for 5 min. The total run time was 30 min. The samples were ionized
via electron impact ionisation (EI) with the following conditions:
70 eV electron energy; 230 C electron source temperature; and
150 C quad temperature. Retention times and characteristic fragments were determined by total ion monitoring (SCAN) in the
range m/z 50500. Abundant ions and/or parent ions without
apparent cross-contribution and interferences were chosen as target ions for SIM mode quantication (Fiamegos, Nanos, Vervoort, &
Stalikas, 2004). When quantication was complete, the peak areas
were normalised to the internal standard, 5a-cholestane, peak.
2.3.3. Validation method
Linearity, accuracy, precision, limits of detection (LODs), limits of quantitation (LOQs), and recoveries were determined to
help validate the method. The LODs and LOQs were determined
as the lowest concentration that could be calculated for the
major ions at a signal-to-noise ratio (peak-to-peak S/N) greater
than 3 and 10, respectively, using the MSD chemstation (Agilent
Technologies Inc., Santa Clara, USA). Linearity was calculated

over a variety of ranges: brassicasterol (2.550 lg/mL); campesterol (2.550 lg/mL); stigmasterol (2.430 lg/mL); b-sitosterol
(2.5120 lg/mL).
Accuracy and precision were determined at six different concentrations for the standard solutions: brassicasterol (2.5, 5, 10,
20, 30, 40 lg/mL); campesterol (2.5, 5, 10, 20, 30, 40 lg/mL); stigmasterol (2.4, 5, 10, 15, 20, 25 lg/mL); b-sitosterol (2.5, 10, 20, 50,
100, 120 lg/mL). Interday (n = 3) and intraday (n = 3 + 3 + 3) accuracy and precision were calculated as percentages and percentages
of the relative standard deviation (%RSD) at the different
concentrations.
Natural mozzarella and imitation mozzarella cheese samples
were spiked at three different concentrations: brassicasterol (5,
20, 40 lg/mL); campesterol (5, 20, 40 lg/mL); stigmasterol (5, 15,
25 lg/mL); b-sitosterol (5, 50, 100 lg/mL). Sample extractions
were carried out on two spiked cheese samples and two blank
cheese samples, according to the extraction method described
above. The spiked and blank samples were spiked with the internal
standard solution before sample preparation. Recoveries (%) were
obtained in triplicate at each concentration, and determined by
comparison of the mean peak area ratios of analytes with the internal standard spike peak area (Lu et al., 2010; Santos et al., 2007).
2.4. Statistical analysis
The data are expressed as a mean standard deviation (SD). The
mean fatty acid and phytosterol content values were compared
using the KruskalWallis test; as a method of median was applied
which was one of the nonparametric statistics as distribution-free
tests in common with the one-way ANOVA. This test does not rely
on parameter estimates or assumptions about variable distributions. Scheffes method was used for a posteriori tests. PCA, with
Varimax rotation, was conducted in two factors, on the fatty acid
proles and phytosterol levels for the cheese classes. SPSS software
(IBM SPSS ver. 18, Armonk, USA) was used for statistical analysis
(Galina, Osnaya, Cuchillo, & Haenlein, 2007; Mapekula, Chimonyo,
Mapiye, & Dzama, 2011).
3. Results and discussion
3.1. Fatty acid proles by GC
Representative examples of FAMEs from cheese extracts are
presented in Table 2. Fourteen unique fatty acids were identied
based on retention times obtained using a PUFA-2 animal-sourced
qualitative standard and undecanoic acid (C11:0) as an internal
Table 2
Fatty acid composition of PUFA-2, animal source as a qualitative standard.
Fatty acids
Myristic acid
Palmitic acid
Palmitoleic acid
Stearic acid
Oleic acid
Vaccenic acid
Linoleic acid
c-Linolenic acid
Linolenic acid
Eicosenoic acid
Arachidonic acid
Eicosapentaenoic acid
Docosatetraenoic acid
Docosahexaenoic acid

C14:0
C16:0
C16:1n7
C18:0
C18:1n9
C18:1n7, trans
C18:2n6
C18:3n6
C18:3n3
C20:1n9
C20:4n6
C20:5n3
C22:4n6
C22:6n3

Retention
time (min)

Composition
(mol%)

15.163
16.890
17.800
19.060
20.081
20.198
21.604
22.854
23.500
22.945
27.311
29.710
31.386
35.619

0.79
25.16
1.67
17.50
22.29
1.93
9.62
2.31
2.38
0.36
8.31
1.77
2.85
3.08

PUFA-2 standard was from animal source, which had a detail description for only
qualitative identication. Undecanoic acid (C11:0, 13.278 min) was used as a
internal standard.

Table 3
Fatty acid proles (mol%) of commercial cheese samples.
Classication
Natural mozzarella cheeses

NMC1
NMC2
NMC3
NMC4
NMC5
NMC6
NMC7
NMC8
IMC1
IMC2
IMC3
IMC4

Processed cheeses

PC1
PC2
PC3
PC4
PC5
PC6
PC7
PC8
PC9
PC10
PC11
PC12

Mixed cheeses

MC1
MC2

C16:0

C16:1
n7

C18:0

C18:1
n9

C18:1
n7

C18:2
n6

C18:3
n6

C18:3
n3

C20:1
n9

C20:4
n6

C20:5
n3

C22:4
n6

C22:6
n3

mean
SD
mean
SD
mean
SD
mean
SD
mean
SD
mean
SD
mean
SD
mean
SD

13.25de
0.07
15.31fg
0.21
13.00d
0.01
13.01d
0.01
13.06d
0.04
15.34fg
0.01
14.95f
0.00
15.31fg
0.01

37.68a
0.08
41.14de
0.13
37.56a
0.00
37.49a
0.01
37.46a
0.02
41.12de
0.02
40.97de
0.01
41.30e
0.00

1.71f
0.01
1.96g
0.03
1.70ef
0.00
1.69ef
0.00
1.69ef
0.00
1.97g
0.03
1.91g
0.00
1.98g
0.00

15.31i
0.07
13.94f
0.15
15.37i
0.00
15.43i
0.01
15.42i
0.01
13.94f
0.02
14.17fg
0.00
13.81f
0.00

26.72d
0.07
24.16abc
0.13
26.75d
0.00
26.82d
0.01
26.82d
0.04
24.14abc
0.02
24.39bc
0.00
24.11abc
0.00

0.87hi
0.01
0.59abcde
0.01
0.92i
0.00
0.93i
0.00
0.92i
0.00
0.64defg
0.00
0.64defg
0.00
0.66fg
0.00

3.39d
0.03
1.67ab
0.06
3.55d
0.00
3.58d
0.00
3.57d
0.00
1.61a
0.05
1.73ab
0.00
1.63a
0.00

0.20ijk
0.00
0.18h
0.00
0.20k
0.00
0.20jk
0.00
0.20jk
0.00
0.18h
0.00
0.18hi
0.00
0.18h
0.00

0.50bcd
0.06
0.76def
0.01
0.57cde
0.00
0.46abcd
0.00
0.46abcd
0.00
0.76def
0.00
0.76def
0.00
0.75def
0.00

0.06b
0.00
0.04a
0.00
0.07b
0.00
0.07b
0.00
0.06b
0.00
0.04a
0.00
0.04a
0.00
0.04a
0.00

0.23g
0.00
0.12e
0.00
0.23g
0.00
0.23g
0.00
0.23g
0.00
0.12e
0.00
0.12e
0.00
0.12e
0.00

0.04c
0.00
0.10d
0.00
0.05c
0.00
0.04c
0.00
0.05c
0.00
0.11d
0.00
0.10d
0.00
0.10d
0.00

0.05f
0.00
0.02c
0.00
0.04ef
0.00
0.04ef
0.00
0.05f
0.00
0.02c
0.00
0.02c
0.00
0.02c
0.00

<0.00ab
0.00
0.01cd
0.00
<0.00ab
0.00
<0.00bc
0.00
<0.00bc
0.00
0.01d
0.00
0.01d
0.00
0.01d
0.00

mean
SD
mean
SD
mean
SD
mean
SD

1.11a
0.02
1.19a
0.00
1.12a
0.00
1.40a
0.02

44.64i
0.07
40.57d
0.06
44.63i
0.10
43.70h
0.15

0.17a
0.00
0.19a
0.00
0.17a
0.00
0.20a
0.00

4.50a
0.01
4.46a
0.02
4.51a
0.02
4.70a
0.02

38.97g
0.06
41.76h
0.04
38.86g
0.08
38.90g
0.13

0.61bcdef
0.01
0.64defg
0.00
0.60bcdef
0.01
0.64cdefg
0.00

9.69g
0.01
10.83h
0.00
9.78g
0.01
9.98g
0.02

0.04a
0.00
0.05ab
0.00
0.04a
0.00
0.06bc
0.00

0.15a
0.00
0.17ab
0.00
0.15ab
0.00
0.29abc
0.00

0.14ef
0.00
0.15ef
0.00
0.13e
0.00
0.15f
0.00

nd

nd

nd

nd

nd

nd

nd

nd

nd

nd

nd

nd

nd

nd

nd

nd

mean
SD
mean
SD
mean
SD
mean
SD
mean
SD
mean
SD
mean
SD
mean
SD
mean
SD
mean
SD
mean
SD
mean
SD

12.69d
0.09
16.57gh
0.90
14.39ef
0.72
12.61d
0.00
12.57d
0.01
12.58d
0.01
17.02h
0.01
17.03h
0.02
17.01h
0.01
14.64f
0.00
14.64f
0.01
14.55ef
0.01

38.66b
0.04
39.75c
0.22
42.14f
0.45
38.60b
0.00
38.58b
0.04
38.60b
0.01
39.89c
0.00
39.88c
0.00
39.88c
0.00
42.34fg
0.01
42.34fg
0.01
42.35fg
0.01

1.69def
0.01
1.69def
0.05
1.59d
0.04
1.68def
0.00
1.68def
0.00
1.68def
0.00
1.71f
0.00
1.71f
0.00
1.71f
0.00
1.61de
0.00
1.61de
0.00
1.60de
0.00

13.99f
0.04
14.63gh
0.38
12.40e
0.14
14.02f
0.00
14.04f
0.01
14.05f
0.00
14.92hi
0.00
14.91hi
0.01
14.92hi
0.01
12.52e
0.00
12.53e
0.00
12.58e
0.01

27.41d
0.08
23.47ab
0.75
25.02c
0.64
27.44d
0.01
27.42d
0.03
27.44d
0.01
23.04a
0.01
23.04a
0.01
23.05a
0.01
24.74c
0.01
24.74c
0.01
24.80c
0.02

0.81h
0.00
0.59abcde
0.04
0.53a
0.04
0.88i
0.00
0.89i
0.00
0.89i
0.00
0.63cdefg
0.00
0.63cdefg
0.00
0.63cdefg
0.00
0.58abcd
0.00
0.58abcd
0.00
0.56ab
0.02

3.73d
0.05
1.63a
0.52
2.63c
0.42
3.75d
0.01
3.81d
0.09
3.74d
0.01
1.28a
0.00
1.28a
0.00
1.29a
0.00
2.40bc
0.02
2.39bc
0.02
2.39bc
0.01

0.19hij
0.01
0.14f
0.01
0.16g
0.00
0.19hijk
0.00
0.19hijk
0.00
0.19hijk
0.00
0.14f
0.00
0.14f
0.00
0.14f
0.00
0.16g
0.00
0.16g
0.00
0.16g
0.00

0.47abcd
0.02
1.24g
0.23
0.90efg
0.23
0.46abcd
0.00
0.46abcd
0.00
0.46abcd
0.00
1.09fg
0.00
1.09fg
0.00
1.09fg
0.00
0.76def
0.00
0.76def
0.00
0.76def
0.00

0.06b
0.00
0.04a
0.01
0.06b
0.01
0.07b
0.00
0.07b
0.00
0.07b
0.00
0.04a
0.00
0.04a
0.00
0.04a
0.00
0.06b
0.00
0.06b
0.00
0.06b
0.00

0.21f
0.00
0.10d
0.01
0.07c
0.01
0.21f
0.00
0.21f
0.00
0.21f
0.00
0.09d
0.00
0.09d
0.00
0.09d
0.00
0.07c
0.00
0.07c
0.00
0.07c
0.00

0.04c
0.00
0.12e
0.01
0.11d
0.00
0.05c
0.00
0.05c
0.00
0.04c
0.00
0.13e
0.00
0.13e
0.00
0.13e
0.00
0.11d
0.00
0.11d
0.00
0.11d
0.00

0.04def
0.00
0.01b
0.00
0.01b
0.00
0.04d
0.00
0.04de
0.00
0.04de
0.00
0.01b
0.00
0.01ab
0.00
0.01b
0.00
0.01ab
0.00
0.01ab
0.00
0.01ab
0.00

<0.00ab
0.00
0.01e
0.00
0.01d
0.00
<0.00ab
0.00
<0.00ab
0.00
<0.00ab
0.00
0.01e
0.00
0.01e
0.00
0.02e
0.00
0.01d
0.00
0.01d
0.00
0.01d
0.00

mean
SD
mean
SD

6.54c
0.05
6.78c
0.44

42.06f
0.07
42.91g
0.15

0.87c
0.00
0.90c
0.05

9.15d
0.04
7.89c
0.25

33.36e
0.08
33.78e
0.40

0.69g
0.00
0.57abc
0.00

6.70e
0.01
6.52e
0.23

0.11e
0.00
0.08d
0.00

0.30abc
0.00
0.39abc
0.02

0.10cd
0.00
0.10c
0.00

0.09d
0.00
0.03a
0.00

0.02b
0.00
0.05c
0.00

0.02c
0.00
<0.00a
0.00

<0.00a
0.00
<0.00ab
0.00

N.S. Kim et al. / Food Chemistry 143 (2014) 4047

Imitation mozzarella cheeses

C14:0

(continued on next page)

43

0.01

0.01b
0.00
0.02
0.01a
0.00
0.04
0.04b
0.00
0.07
0.11d
0.00
0.04
0.06c
0.00
0.05
7.81f
0.01
3.07
35.29f
0.04
5.85

0.65efg
0.00
0.13

3.2. Phytosterol analysis by GCMS

SEM

Mean values of three replicates standard deviation; nd, not detected.

15 Means in the same column with different superscripts are different at p < 0.05, n = 3.

6.79b
0.01
3.91
0.53b
0.00
0.62
44.44i
0.04
2.25
4.00b
0.01
5.48
mean
SD
MC3

Classication

Table 3 (continued)

standard. Table 3 reports the fatty acid compositions (mol%) of the

fat extracts from 27 commercial cheeses.
Arachidonic, eicosapentaenoic, docosatetraenoic, and docosahexaenoic acids could not be detected in the IMCs. Also, docosahexaenoic acid could not be detected in one of the mixed cheeses
(MC3).
The oleic acid, vaccenic acid, and linoleic acid proles for each
cheese class were found to be consistent with those from other
studies (Prandini, Sigolo, & Piva, 2011; Zhang, Mustafa, & Zhao,
2006).
NMCs had oleic acid, 24.1126.82 mol%; vaccenic acid, 0.59
0.93 mol%; and linoleic acid, 1.613.58 mol%, respectively. IMCs
contained oleic acid, 38.8641.76 mol%; vaccenic acid, 0.60
0.64 mol% and linoleic acid, 9.6910.83 mol%, respectively. PCs
contained oleic acid, 23.0427.44 mol%; vaccenic acid, 0.53
0.89 mol%, and linoleic acid, 1.283.73 mol%. The mixed cheeses
(MCs) consisted of oleic acid, 33.3635.29 mol%; vaccenic acid,
0.570.69 mol%, and linoleic acid, 6.527.81 mol%.
The proportion of oleic and linoleic acids in NMCs were statistically similar (p < 0.05) to those of PCs, and these acids were significantly (p < 0.05) more common in NMCs and PCs than in the MCs.
The proportion of oleic and linoleic acids in MCs was lower than
that of IMCs but higher than that of others. There were no signicant differences (p < 0.05) among the concentrations of vaccenic
acid in the four cheese classes. From the fatty acid proles, more
fat (p < 0.05) was present in the MCs and IMCs than in the NMCs
and PCs.

0.25abc
0.00
0.31

C22:4
n6
C20:5
n3
C20:4
n6
C20:1
n9
C18:3
n6
C16:0
C14:0

C16:1
n7

C18:0

C18:1
n9

C18:1
n7

C18:2
n6

C18:3
n3

nd

N.S. Kim et al. / Food Chemistry 143 (2014) 4047

C22:6
n3

44

3.2.1. Phytosterol content

Retention times and m/z ratios of four phytosterols from the
cheese extracts and the internal standard 5a-cholestane, used for
quantitative purposes, are given in Table 4. The major fragments
were identied by GCMS in SCAN mode (Fig 1). For 5a-cholestane, peaks at m/z 372 (the parent ion, M), 357 (M 15), and
217 (M 155), were used for quantitation. Four peaks for each trimethylsilyl derivative were used for quantitation; the derivatives
were
brassicasterol
(m/z
470 = M + 72,
380 = M 18,
255 = M 143), campesterol (m/z 472 = M + 72, 382 = M 18,
367 = M 33), stigmasterol (m/z 484 = M + 72, 394 = M 18,
255 = M 157) and b-sitosterol (m/z 486 = M + 72, 396 = M 18,
381 = M 33). The fragmentation patterns observed for the trimethylsilyl ethers were consistent with those from previous studies (Schummer et al., 2009; Wu, Hu, Yue, Yang, & Zhang, 2009).
Plant sterol contents (lg/g), determined by SIM-mode GCMS,
for the 27 cheese samples are given in Table 5. Quantitation was
based on the major ions for brassicasterol, campesterol, stigmasterol, b-sitosterol, and 5a-cholestane at m/z 255, 382, 255, 396,
and 217, respectively. For all of the samples tested, campesterol
concentrations ranged from 1.17 to 13.49 lg/g, stigmasterol concentrations ranged from 0.47 to 7.02 lg/g, and b-sitosterol concentrations ranged from 0.59 to 47.70 lg/g. The brassicasterol content
could not be quantied in all of the samples, and stigmasterol
could not be quantied in NMC1 and PC2.
The NMCs had 1.171.43 lg/g campesterol, 0.470.49 lg/g stigmasterol, 0.591.05 lg/g b-sitosterol and 1.772.95 lg/g total content. The IMCs had 10.9013.49 lg/g campesterol, 5.957.02 lg/g
stigmasterol, 40.7647.70 lg/g b-sitosterol and 57.6168.21 lg/g
total content. The PCs contained 1.603.18 lg/g campesterol,
0.511.19 lg/g stigmasterol, 0.855.88 lg/g b-sitosterol, and
2.4510.04 lg/g total content. The MCs contained 5.888.45 lg/g
campesterol, 3.225.77 lg/g stigmasterol, 18.0628.10 lg/g bsitosterol, and 27.1742.32 lg/g total content.
Phytosterols, the most common being b-sitosterol, are found in
fats and oils (Contarini, Povolo, Bontto, & Berardi, 2002). Phytos-

N.S. Kim et al. / Food Chemistry 143 (2014) 4047

Table 4
Main characteristic ions with relative abundances and retention times of the
derivatized phyterosterols in the GCMS full-scan mass spectra.
Target compounds

Diagnostic ions

5a-Cholestane
Brassicasterol
Campesterol
Stigmasterol
b-Sitosterol

217
255
382
255
396

(100%),
(100%),
(100%),
(100%),
(100%),

357
380
367
394
381

Retention
time (min)
(22.6%),
(38.6%),
(70.0%),
(33.6%),
(73.6%),

372
470
472
484
486

(15.0%)
(29.1%)
(47.5%)
(26.3%)
(59.3%)

11.105
15.868
17.087
17.731
19.089

terols in NMCs are derived from milk fats, whereas phytosterols in

cheese substitutes are derived from vegetable oils (Bachmann,
2001). Overall phytosterol contents in NMCs and PCs were lower
(p < 0.05) than in the IMCs and MCs. The total phytosterol contents
in PCs were statistically similar (p < 0.05) to those of NMCs. This result corresponds with product description said NMCs was used for
one of the materials for PCs at 7080%. The highest (p < 0.05) phytosterol levels were found in the IMCs, sterols which were likely
derived from vegetable fats. The MCs had 4050% the same phytosterol content as IMCs, and conrmed that the MCs were blends of
NMCs and IMCs. The campesterol, b-sitosterol, and total phytosterol contents for the MCs were distinct from the IMCs. These results are consistent with the results obtained from the fatty acid
proles reported above.
3.2.2. Phytosterol determination method validation
Linearity, accuracy, precision, LODs, LOQs and recovery data for
the GCMS analysis of plant sterols are given in Tables 6 and 7. The
rapid GCMS method (30 min) was capable of separating four
plant sterols and the internal standard (Fig. 1). The LODs of brassicasterol, campesterol, stigmasterol, and b-sitosterol were all lower
than 1 lg/mL. LOQs for brassicasterol, campesterol, stigmasterol,
and b-sitosterol were 2.64, 2.67, 2.31, and 2.43 lg/mL, respectively.
Linearity, obtained from the relative mass ratios of each phytosterol to the internal standard, were 2.550 lg/mL for brassicasterol
and campesterol, 2.430 lg/mL for stigmasterol, and 2.5120 lg/
mL for b-sitosterol. The correlation coefcients obtained from the
linearity plots were greater than 0.995 for all of the samples. Intraday precision for the method (expressed as the RSD) ranged from
0.6% to 8.7%, while interday precision ranged from 0.6% to 9.4%.

45

Intraday accuracy ranged from 93.1% to 108.2%, and interday

accuracy ranged from 92.6% to 108.4%. Recoveries for the
phytosterols, added prior to sample preparation were greater than
80% at the spike levels evaluated.
3.3. Principal component analysis
PCA was performed with SPSS software, and the KruskalWallis
test was used to compare multiple independent groups. The variance eigenvalue was greater than 1. The loadings (or factor scores)
corresponding to the principal components were calculated from
the correlation matrix (Massart, Vandeginste, Deming, Michotte,
& Kaufman, 1988). The PCA results were used to identify important
experimental factors, and factor score plots were used to indicate
similar, dissimilar, typical, or outlier data (Qiu et al., 2007; Shin,
Craft, Pegg, Phillips, & Eitenmiller, 2010).
PCA plots for the data obtained are shown in Fig. 2 (A). The principal component contributed 58.910% and was composed primarily
of nine fatty acids: myristic acid, palmitoleic acid, stearic acid, oleic
acid, linoleic acid, linolenic acid, eicosenoic acid, eicosapentaenoic
acid, and docosahexaenoic acid (Table 8). The secondary component accounted for 40.901%, and was composed of palmitic acid,
vaccenic acid, c-linolenic acid, arachidonic acid, and docosatetraenoic acid. In the PCA plot with two components, three distinct
groups were identiable. Group A was correlated with the IMC
class, group B with the MC class, and group C was associated with
the NMC and PC classes. Using this analysis, it was possible to discriminate between NMCs and IMCs, and between IMCs and MCs.
On the other hand, it was not possible to differentiate between
the PC class and NMCs.
The score plot for the phytosterols generated from comparison
of the two principal components (eigenvalue P1) is depicted in
Fig. 2 (B). The primary component contributed 51.974%, and consisted mainly of campesterol and b-sitosterol. The secondary component contributed 47.496%, and was composed primarily of
stigmasterol. From the PCA plot, three groups were identied.
Group D was correlated with NMCs and a part of the PC class
(PC2 and PC79), group E was associated with the rest of the PC
class (PC1, PC36, and PC1012), and Group F was composed of
the IMC and MC classes. From the PCA plot, it was possible to discriminate between the NMCs and the IMCs. On the other hand, it

Fig. 1. Total ion chromatogram of derivatized phytosterols in GC-MS. (1) 5a-cholestane (Internal standard), (2) Brassicasterol, (3) Campesterol, (4) Stigmasterol, (5) bsitosterol.

46

N.S. Kim et al. / Food Chemistry 143 (2014) 4047

Fig. 2. PCA score plot for 27 commercial cheeses. (A) Fatty acid proles (B) Phytosterol contents. NMC, natural mozzarella cheeses; IMC, imitated mozzarella cheeses; PC,
processed cheeses; MC, mixed cheeses of NMC and IMC.

was difcult to differentiate between some of the PCs from the

NMC class and between the MC and IMC classes.

4. Conclusion
PCA of the fatty acid prole and phytosterol content data enabled the differentiation of NMCs from IMCs. Unfortunately, the
PC and MC classes were not completely separated from the NMC
and IMC classes. The presented method to discriminate natural
mozzarella cheese from cheese substitutes is interesting approach,
but more research is needed to enable the complete differentiation
of the cheese classes investigated in this work.

Acknowledgements
This research was supported by a grant from the Criminal Investigations Ofce of the Korea Food & Drug Administration (KFDA) in
2012.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.foodchem.2013.
07.083.

N.S. Kim et al. / Food Chemistry 143 (2014) 4047

References
Bachmann, H.-P. (2001). Cheese analogues: A review. International Dairy Journal, 11,
505515.
Christie, W. W. (1989). Gas chromatography and lipidsa practical guide. Dundee:
The Oily Press.
Christie, W. W., Sbdio, J. L., & Juanda, P. (2001). A practical guide to the analysis
of conjugated linoleic acid. Inform, 12, 147152.
Contarini, G., Povolo, M., Bontto, E., & Berardi, S. (2002). Quantitative analysis of
sterols in dairy products: Experiences and remarks. International Dairy Journal,
12, 573578.
Fiamegos, Y. C., Nanos, C. G., Vervoort, J., & Stalikas, C. D. (2004). Analytical
procedure for the in-vial derivatizationextraction of phenolic acids and
avonoids in methanolic and aqueous plant extracts followed by gas
chromatography with mass-selective detection. Journal of Chromatography A,
1041, 1118.
Galina, M. A., Osnaya, F., Cuchillo, H. M., & Haenlein, G. F. W. (2007). Cheese quality
from milk of grazing or indoor fed Zebu cows and Alpine crossbred goats. Small
Ruminant Research, 71, 264272.
Garca-Llatas, G., & Rodrguez-Estrada, M. T. R. (2011). Current and new insights on
phytosterol oxides in plant sterol-enriched food. Chemistry and Physics of Lipids,
164, 607624.
Haswell, S. J. (1992). Introduction to chemometrics. In Practical guide to
chemometrics (pp. 13). New York: Marcel Dekker Inc.
Jana, A. H., & Mandal, P. K. (2011). Manufacturing and quality of mozzarella cheese:
A review. International Journal of Dairy Science, 6, 199226.
Jana, A. H., Patel, H. G., Suneeta, P., & Prajapati, J. P. (2010). Quality of casein based
mozzarella cheese analogue as affected by stabilizer blends. Journal of Food
Science Technology, 47, 240242.
Kadegowda, A. K. G., Piperova, L. S., & Erdman, R. A. (2008). Principal component and
multivariate analysis of milk long-chain fatty acid composition during diet
induced milk fat depression. Journal of Dairy Science, 91, 749759.
Korea Food and Drug Administration (2012). Health Functional Food Code.
Notication
No.68.
URL:
<http://www.mfds.go.kr/gwangju/
index.do?mid=26&seq=18826&cmd=v>.
Lu, J., Wang, S., Dong, Y., Wang, X., Yang, S., Zhang, J., et al. (2010). Simultaneous
analysis of fourteen tertiary amine stimulants in human urine for doping
control purposes by liquid chromatographytandem mass spectrometry and
gas chromatographymass spectrometry. Analytica Chimica Acta, 657, 4552.
Mapekula, M., Chimonyo, M., Mapiye, C., & Dzama, K. (2011). Fatty acid, amino acid
and mineral composition of milk from Nguni and local crossbred cows in South
Africa. Journal of Food Composition and Analysis, 24, 529536.

47

Massart, D. L., Vandeginste, B. G. M., Deming, S. N., Michotte, Y., & Kaufman, L.
(1988). Principal components and factor analysis. In Chemometrics: A textbook
(pp. 339370). New York: Elsevier Science Publishing Company Inc.
Matos, L. C., Cunha, S. C., Amaral, J. S., Pereira, J. A., Andrade, P. B., Seabra, R. M., et al.
(2007). Chemometric characterization of three varietal olive oils (Cvs.
Cobrancosa, Madural and Verdeal Transmontana) extracted from olives with
different maturation indices. Food Chemistry, 102, 406414.
Middleton, J.L. (1989). Process of using rennet casein for producing imitation
cheese. United States Patent 4 822 623 (pp. 18).
Ntakatsane, M. P., Liu, X. M., & Zhou, P. (2013). Short communication: Rapid
detection of milk fat adulteration with vegetable oil by uorescence
spectroscopy. Journal of Dairy Science, 96, 21302136.
Pellegrino, L., Resmini, P., De Noni, I., & Masotti, F. (1996). Sensitive determination
of lysinoalanine for distinguishing natural from imitation mozzarella cheese.
Journal of Dairy Science, 79, 725734.
Prandini, A., Sigolo, S., & Piva, G. (2011). A comparative study of fatty acid
composition and CLA concentration in commercial cheeses. Journal of Food
Composition and Analysis, 24, 5561.
Qiu, Y., Su, M., Liu, Y., Chen, M., Gu, J., Zhang, J., et al. (2007). Application of ethyl
chloroformate derivatization for gas chromatographymass spectrometry
based metabonomic proling. Analytica Chimica Acta, 583, 277283.
Resmini, P., Pellegrino, L., & Massotti, F. (1993). Evaluation of the Maillard reaction
extent for quality control of low-heat treated dairy products. IDF Special Issue,
9302:153. Brussels: International Dairy Federation.
Santos, R., Limas, E., Sousa, M., Castilho, M. C., Ramos, F., & Silveira, M. I. N. (2007).
Optimization of analytical procedures for GCMS determination of phytosterols
and p4, phytostanols in enriched milk and yoghurt. Food Chemistry, 102,
113117.
Schummer, C., Delhomme, O., Appenzeller, B. M. R., Wennig, R., & Millet, M. (2009).
Comparison of MTBSTFA and BSTFA in derivatization reactions of polar
compounds prior to GC/MS analysis. Talanta, 77, 14731482.
Shin, E.-C., Craft, B. D., Pegg, R. B., Phillips, R. D., & Eitenmiller, R. R. (2010).
Chemometric approach to fatty acid proles in Runner-type peanut cultivars by
principal component analysis (PCA). Food Chemistry, 119, 12621270.
Wu, J., Hu, R., Yue, J., Yang, Z., & Zhang, L. (2009). Determination of fecal sterols by
gas chromatographymass spectrometry with solid-phase extraction and
injection-port derivatization. Journal of Chromatography A, 1216, 10531058.
Zhang, R. H., Mustafa, A. F., & Zhao, X. (2006). Effects of feeding oilseeds rich in
linoleic and linolenic fatty acids to lactating ewes on cheese yield and on fatty
acid composition of milk and cheese. Animal Feed Science and Technology, 127,
220233.