See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/223530238

Measurement of malondialdehyde in fish: A comparison study between HPLC

methods and the traditional spectrophotometric test

Article  in  Food Chemistry · February 2009

DOI: 10.1016/j.foodchem.2008.06.052

CITATIONS READS

76 347

3 authors, including:

Rogério Oliveira Mendes Carlos Cardoso

Instituto Português do Mar e da Atmosfera University of Porto
100 PUBLICATIONS   1,567 CITATIONS    69 PUBLICATIONS   832 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

FreshFish View project

QCA III View project

All content following this page was uploaded by Rogério Oliveira Mendes on 23 November 2017.

The user has requested enhancement of the downloaded file.

 Author's personal copy

Food Chemistry 112 (2009) 1038–1045

Contents lists available at ScienceDirect

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

Analytical Methods

Measurement of malondialdehyde in fish: A comparison study between HPLC

methods and the traditional spectrophotometric test
Rogério Mendes *, Carlos Cardoso, Carla Pestana
Department of Upgrading of Fish Products, National Institute of Biological Resources, INRB/IPIMAR, Av. Brasília, 1449-006 Lisbon, Portugal

a r t i c l e i n f o a b s t r a c t

Article history: Alternative methods to the traditional spectrophotometric determination of the malondialdehyde-thio-
Received 5 March 2008 barbituric acid (MDA-TBA) complex (method A) and to the overestimation of MDA levels in the TBA reac-
Received in revised form 28 May 2008 tion have been develop for the evaluation of lipid oxidation in fish. In this study, two HPLC separation
Accepted 22 June 2008
methods of the MDA-TBA (method B) and MDA-dinitrophenylhydrazine (MDA-DNPH) adduct (method
C) were investigated and compared to the traditional spectrophotometric TBA test (method A) in samples
of chilled fish (hake, sea bream and sardine). Detection limits were 0.16, 0.10 and 0.20 lM MDA and quan-
Keywords:
tification limits were 0.23, 0.17 and 0.26 lM MDA, for methods A, B and C, respectively. Recovery of method
Malondialdehyde
Fish
B ranged between 100% and 108% and of method C between 90% and 112%. Method A presented low recov-
Lipid oxidation ery levels (under 71%). Overall method performance followed the order HPLC method MDA-DNPH > HPLC
2-Thiobarbituric acid method MDA-TBA > traditional spectrophotometric TBA test. Though showing a better accuracy and spec-
2,4-Dinitrophenylhydrazine ificity, method C had, however, some disadvantages, a relatively high limit of detection (0.20 lM MDA) and
a lower reproducibility at lower MDA contents in standards and samples. Nevertheless, these are not crit-
ical drawbacks for an application in routine fish analysis, given the high MDA concentrations in oxidised
fish. The application of the modified HPLC methods in fish samples with different levels of MDA, showed
that these methods are useful for the samples with low amounts of oxidation products, such as chilled hake
as well as in samples with high levels of oxidation, like 15 days chilled stored sardine.
Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction whenever an assessment of lipid oxidation is required (Panpipat &

Yongsawatdigul, 2008). Reaction occurs by attack of the monoenol-
Lipid oxidation in food is associated with the development of ic form of MDA on the active methylene groups of TBA. Visible and
rancidity and oxidative deterioration. Fish, on account of its high ultraviolet spectrophotometry of the pigment confirms its primary
content of polyunsaturated fatty acids (PUFA), is highly susceptible maximum at 532–535 nm and a secondary one at 245–305 nm
to lipid oxidation during manipulation, processing, and cooking. As (Sinnhuber, Yu, & Yu, 1958). The intensity of colour is a measure
a consequence of oxidative spoilage, lipid hydroperoxides are of MDA concentration (Tarladgis, Pearson, & Dugan, 1964) and
formed, which, in turn, are unstable and decompose to aldehydes, has been correlated with rancidity (Zipser, Kwon, & Watts, 1964).
ketones, alcohols, acids or hydrocarbons (St. Angelo, 1996). These Reaction kinetics depends on the concentration of TBA solution,
so-called secondary oxidation products can change food quality, temperature and pH (Fernández et al., 1997).
namely, colour, texture, flavour and odour (Fernández, Pérez- Several variations of MDA-TBA method exist, namely, a method
Álvarez, & Fernández-López, 1997). One of the most important for fish lipids was described by Ke and Woyewoda (1979) and for
products of oxidation is malondialdehyde (MDA), which is thought fish by Vyncke (1970). Furthermore, various procedures are gener-
to be a carcinogenic initiator and mutagen. MDA has often been ally performed in food: direct heating of the sample with TBA
used as marker of oxidative damage in biological samples (Kinter, (Sinnhuber & Yu, 1958), sample distillation (Ke, Cervantes, & Ro-
1995) and foods (St. Angelo, 1996). The most widely used method bles-Martinez, 1984), lipid extraction with organic solvents (Pikul,
for determination of MDA is the spectrophotometric determination Leszczynski, & Kummerow, 1989) or acid extraction of MDA
of the pink fluorescent MDA-thiobarbituric acid (MDA-TBA) com- (Squires, 1990), followed by acid reaction with TBA. In spite of
plex produced after reaction with 2-thiobarbituric acid (TBA) at these improving modifications, the traditional spectrophotometric
low pH and high temperature (Bergamo, Fedele, Balestrieri, Abres- TBA test has been criticised for its lack of sensitivity (Squires, 1990)
cia, & Ferrara, 1998). This simple technique is used in fish analysis, and its high inaccuracy, since TBA reacts not only with MDA but
also with many other compounds (for instance, carbohydrates,
* Corresponding author. Tel.: +351 213027036; fax: +351 213015948.
amino acids, pyridines, pigments, etc.) (Guillen-Sans & Guzman-
E-mail address: [email protected] (R. Mendes). Chozas, 1998), interfering in the TBA assay and resulting in

0308-8146/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.foodchem.2008.06.052
 Author's personal copy

R. Mendes et al. / Food Chemistry 112 (2009) 1038–1045 1039

considerable overestimation, as well, variability in the results (Merluccius capensis) was purchased frozen from a local fish pro-
(Mateos, Lecumberri, Ramos, Goya, & Bravo, 2005). The traditional cessor and thawed overnight in a refrigerated chamber. For the
spectrophotometric TBA test has also shown problems when used study of rancidity development all fish species were kept in a
in frozen Gadidae fish species, like saithe, haddock, cod and others. refrigerated chamber (2 ± 1 °C) and analysed periodically over a
During such storage, these species can produce an important period of 19 days.
amount of formaldehyde which in turn was shown to interfere in
the traditional spectrophotometric TBA test (Aubourg, 1999; Car- 2.3. Sample preparation
eche & Tejada, 1988). Furthermore, the high temperatures (95–
100 °C), extended incubation times (up to 150 min) (Sakai, Habiro, For MDA extraction, a portion (15 g) of minced fish was
& Kawahara, 1999; Volpi & Tarugi, 1998) and strong acidic condi- weighed in a 50 ml centrifuge tube and diluted with 30 ml of
tions (pH 1.5–3.5) commonly required for the reaction of MDA 7.5% TCA solution (7.5% (p/v) TCA, 0.1% (p/v) EDTA, 0.1% (p/v) pro-
with TBA may cause an artifactual peroxidation of sample constit- pyl gallate). Mixture was then homogenised with a Polytron
uents even in the presence of added antioxidants. For this reason PT3000 blender for 1 min at 5000 rpm and filtered through filter
and in order to eliminate interferences in the formation of the paper (Whatman #1). Filtrate was centrifuged for 10 min at
MDA-TBA red pigment, more sensitive and advanced methods of 6000 rpm. Supernatant (extract) was used in both TBA and DNPH
analysis of biological matrices by capillary gas chromatography methods.
with electron-capture and mass spectrometry and liquid chroma-
tography–mass spectrometry were developed (Cighetti, Debiasi, 2.4. Traditional spectrophotometric TBA determination of MDA-TBA
Paroni, & Allevi, 1999; Jardine, Antolovich, Prenzler, & Robards, (TBA Test)
2002; Stalikas & Konidari, 2001). Also more specific high-perfor-
mance liquid chromatographic (HPLC) approaches using re- This MDA determination (method A) was performed according
versed-phase chromatography (Draper & Hadley, 1990; Squires, to Vyncke’s methodology (Vyncke, 1970). TEP (1,1,3,3-tetraethoxy-
1990) with UV/VIS absorption (Cesa, 2004) or with fluorimetric propane) was used as the MDA standard, without hydrolysis prior
detection (De las Heras, Schoch, Gibis, & Fischer, 2003) have been to the TBA reaction. A standard curve was made from TEP diluted
used. Derivatisation of MDA with 2,4-dinitrophenylhidrazine in 7.5% TCA solution, at concentrations 2.0, 4.0, 6.0, 8.0 and
(DNPH) and conversion into pyrazole and hydrazone derivatives, 10.0 lM. 5 ml sample supernatant, standard or blank was trans-
has been also found to allow a specific estimation of MDA, partic- ferred into a screw-capped tube, 5 ml of 20 mM TBA solution
ularly, if combined with HPLC separation (Mateos et al., 2005). was added, mixture was vigorously agitated in a vortex and placed
Absorption at 310 nm is used to calculate MDA concentration. This in a boiling water bath for 60 min. After cooling, MDA-TBA com-
method has been used to determine MDA levels in biological sam- plex was measured at 530 nm using an UNICAM Helyos spectro-
ples, such as rat and human plasma (Pilz, Meineke, & Gleiter, 2000) photometer. Results were expressed as micromoles MDA present
or rat urine (Ekström, Garberg, Egestad, & Högerg, 1988). However, in 1 kg of muscle.
it has not been applied yet to fish samples.
Though the mentioned disadvantages, conventional spectro- 2.5. HPLC determination of MDA-TBA
photometric MDA-TBA methods are preferred because of their sim-
plicity. Therefore, development of a simple, sensitive and specific HPLC separation of MDA-TBA adduct (method B) was performed
MDA-TBA method has remained a challenge. The objective of this according to the method described by Seljeskog, Hervig, and Mans-
study was to develop an improved sensitive and specific HPLC oor (2006), with modifications in the sample deproteinisation pro-
method for MDA determination in fish with different fat contents cedure (see sample preparation, above). TEP was used as the MDA
and different degrees of rancidity, using TBA or DNPH as derivatis- standard, without hydrolysis prior to the TBA reaction. A standard
ing reagents and compare these two methods with the traditional curve was made from TEP diluted in 7.5% TCA solution, at concen-
spectrophotometric technique (TBA test). trations of 0.6, 1.3, 2.5, 5.0, 10.0 lM. 0.5 ml sample supernatant,
standard or blank was transferred into a screw-capped tube,
1.5 ml of 40 mM TBA solution was added, mixture was vigorously
2. Materials and methods agitated in a vortex and placed in a hot water bath at 97 °C for
60 min. After cooling in a freezer at 20 °C for 20 min, 3 ml meth-
2.1. Chemicals and reagents anol was added and mixed in a vortex. The resulting solution was
filtered through a 0.2 lm PTFE membrane (AcrodiscÒ CR 25 mm
All the chemicals and reagents used were analytical grade of the Syringe Filter, PALL Life Sciences) into autosampler vials. HPLC
highest purity. Potassium dihydrogenphosphate (KH2PO4), potas- analysis was performed using an Agilent 1100 Series system,
sium hydroxide (KOH), hydrochloric acid fuming 37% (HCl), glacial equipped with pump, degasser, autosampler, spectrofluorimetric
acetic acid (CH3COOH), trichloroacetic acid (TCA, CCl3COOH), per- detector and system controller with a PC control program. Separa-
chloric acid (PCA, HCLO4), ethylenediaminetetracetic acid (EDTA, tion of MDA-TBA was done using a Phenomenex Gemini C18 col-
C10H16N2O8) and propyl gallate (C10H12O5) were purchased from umn (5 lm, 150  4.6 mm), operated isocratically with a HPLC
Merck (Darmstadt, Germany). 2-thiobarbituric acid (2-TBA, mobile phase pumped at 1.0 ml/min and consisting of 50 mM
C4H4N2O2S) and 1,1,3,3-tetraethoxypropane (TEP, C11H24O4) were KH2PO4 buffer solution, methanol and acetonitrile in the propor-
from Sigma (St. Louis, MO, USA). 2,4-dinitrophenylhydrazine tion 72:17:11 (v/v). In this method, injection volume was 10 ll,
(DNPH, C6H6N4O4) was from Fluka (Deisenhofen, Germany). HPLC sample run took 8 min and retention time for MDA-TBA was near
grade organic solvents were used: methanol (CH3OH) and acetoni- 5.5 min. Spectrofluorimetric detector wavelengths were set at
trile (CH3CN) were from Merck (Darmstadt, Germany). Aqueous 525 nm (excitation) and 560 nm (emission). Results were ex-
solutions were prepared with Milli-Q purified water. pressed as micromoles MDA present in 1 kg of muscle.

2.2. Materials 2.6. HPLC determination of MDA-DNPH

Sardine (Sardina pilchardus) and farmed sea bream (Sparus aura- MDA-DNPH adduct detection (method C) was based on the
ta) were bought fresh from a local supermarket. South African hake method described by Mateos et al. (2005), modified in order to ob-
 Author's personal copy

1040 R. Mendes et al. / Food Chemistry 112 (2009) 1038–1045

tain a more practical and rapid method. TEP was used as the MDA val. Coefficients of variation were obtained from the ratio of the
standard, without hydrolysis prior to the TBA reaction. A standard standard deviation to the standards mean.
curve was made from TEP diluted in 7.5% TCA solution, at concen- Samples: Within-run precision, for a same extract of hake or sea
trations of 0.6, 1.3, 2.5, 5.0, 10.0 lM. Sample supernatant, standard bream, was calculated for 13 aliquots of the respective extract.
or blank was filtered through a 0.2 lm PTFE membrane (AcrodiscÒ Within-run precision, for different extracts of hake, sea bream
CR 25 mm Syringe Filter, PALL Life Sciences) and 200 ll was trans- and sardine, was calculated for 13 aliquots of 13 different extracts
ferred into an autosampler vial. Derivatisation of MDA as well in each given species. Each group of 13 aliquots was analysed on
MDA-DNPH complex separation and analysis were performed by the same day. Between-run precision was determined for sea
an Agilent 1100 Series HPLC system, equipped with pump, degas- bream using eight different extracts on eight different days over
ser, autosampler, UV/VIS absorbance detector and system control- a period of 11 days. Coefficients of variation were obtained from
ler with a PC control program. Each vial was placed in the the ratio of the standard deviation to the samples mean.
autosampler tray and using automated derivatization 20 ll of
5 mM DNPH (in 2 M HCl) solution were added. Injector drew and 2.10. Specificity
ejected two times 100 ll of the reactive mixture in order to achieve
a good homogenisation. After waiting 10 min, 50 ll of the mixture For the evaluation of the specificity of the studied methods,
was injected. For each injection cycle, special care was taken in MDA contents in the muscle of hake, sea bream and sardine stored
washing the needle with acetonitrile. Separation of MDA-DNPH in a refrigerated chamber (2 ± 1 °C) were followed for 19 days by
was done using a Phenomenex Gemini C18 column (5 lm, methods A, B and C. The same extracts were used for all methods.
250  4.6 mm), operated isocratically at 30 °C with a HPLC mobile These three fish species were chosen in order to investigate the
phase pumped at 1.0 ml/min and consisting of water, acetonitrile methods potential in assessing lipid oxidation of fish with different
and glacial acetic acid in the proportion 55:45:0.2 (v/v). In this fat contents.
method, sample run took 18 min time and retention time for
MDA-DNPH was near 12.2 min. UV/VIS detector wavelength was 2.11. Statistical analysis
set at 310 nm and results were expressed as micromoles MDA
present in 1 kg of muscle. For each standard/sample and aliquot, all MDA determinations
were done in duplicate. For comparison of lipid oxidation in differ-
2.7. Detection and quantification limits and linearity ent fish species over storage time, a general linear model —factorial
analysis of variance (ANOVA)— was used to determine significant
Limits of detection (LD) and quantification (LQ) were calculated, differences (p 6 0.05) among methods. Multiple comparisons were
respectively, as the mean blank or trace solution response plus 3 done by the Tukey HSD test. All statistical treatment was done with
and 5 standard deviations of blank or trace solution response. For the software STATISTICAÓ from StatSoft, Inc. (Tulsa, OK, USA), ver-
method A and B, 10 blank replicates were measured. In method sion 6.1, 2003.
C, 10 trace solution replicates with 0.08 lM MDA were measured,
since blank had no signal. Trace solution concentration was deter- 3. Results and discussion
mined by consecutive diluting the lowest standard until no signal
was attained. The lowest concentration enabling a signal was con- In order to achieve a rapid and simple MDA determination
sidered the trace level. Concerning linearity, correlation coeffi- method with application in fish products, an HPLC method for
cients of the regression lines from all three methods were the analysis of MDA-TBA in serum and plasma (Seljeskog et al.,
determined. 2006) and an HPLC method for the analysis of MDA-DNPH in serum
and liver (Mateos et al., 2005) were modified, namely, concerning
2.8. Recovery/Accuracy deproteinisation. The main purpose of these modifications was to
adapt techniques to routine fish analysis and to optimize experi-
Recovery of MDA was evaluated by addition of known amounts mental conditions. In this context, two alternative deproteinisation
of TEP (lower range: 0.15-0.30 lM; higher range: 2.50–5.00 lM) to strategies were tested for TBA method (see below, Recovery/accu-
the sample extracts. The amount of MDA was determined by using racy section). A fundamental objective was the development of
specific standard curves. Determinations were done in quadrupli- simple methods, in accordance to the principle that sample prepa-
cate and each recovery value was the mean of all determinations. ration procedures should be kept as simple as possible, in order to
For all methods, usual MDA extraction procedures were followed. avoid error additions and to maximize analyte recoveries (extra
However, for MDA-TBA detection after HPLC separation, two ex- preparation steps should be added only if needed).
tracts/sample preparation methods were compared: Typical chromatograms obtained from the MDA-TBA and MDA-
DNPH adducts are shown in Fig. 1. The MDA-TBA and MDA-DNPH
(i) a direct 7.5% TCA extract treated according to method B; peaks were identified by the elution profile of authentic standard.
(ii) a water extract (0.5 ml) treatment with 0.1125 N PCA The average retention time for the MDA-TBA adduct was around
(1.5 ml) and 40 mM TBA (1.5 ml), followed by reaction at 5.6 min at a flow rate of 1.0 ml/min and it produced the main peak
97 °C for 60 min and deproteinisation with 20% TCA in chromatogram. On the other hand, the MDA-DNPH complex, de-
(1.0 ml) and methanol (3.0 ml) according to the method of tected near 12.3 min at a flow rate of 1.0 ml/min, was a small peak
Seljeskog et al. (2006). behind other peaks produced by TCA solution components and
DNPH reagent. Hence, these peaks were also present in the blank
(see peak near 7.5 min in Fig. 1). No interfering peak was detected
2.9. Analytical reproducibility in fish samples and chromatographic separation was very good for
both MDA-TBA and MDA-DNPH analysis.
Standards: Within-run precision was determined for 13 aliquots
of standard solutions with 5 concentrations (n = 65), injected on 3.1. Detection and quantification limits and linearity
the same day. Between-run precision was calculated for 13 ali-
quots of standard solutions with 5 concentrations (n = 65), each Detection limits were 0.16, 0.10 and 0.20 lM MDA and quanti-
aliquot being injected in one different day, within an 18 day inter- fication limits were 0.23, 0.17 and 0.26 lM MDA, for the traditional
 Author's personal copy

R. Mendes et al. / Food Chemistry 112 (2009) 1038–1045 1041

tween 100% and 108% (Table 1). This represents a significant

A improvement over the 66–71% recovery of MDA, when determined
LU 5.578
MDA Standard by the traditional spectrophotometric TBA test. Lower TBA concen-
1.6 tration in method A (20 mM vs 40 mM) is a possible factor hinder-
(5 μM)
ing recoveries near to 100%. Samples extracted with water and
1.4 deproteinised after TBA derivatization (method ii) showed also
after analysis with method B a reduced recovery (65–75%), pre-
1.2 sumably, because, as Pilz et al. (2000) and Bird, Hung, Hadley,
and Draper (1983) refer, strong acidic conditions are essential for
1 the release of protein bound MDA. This acid extraction was carried
out in sample preparation method (i), but not in (ii), where TCA
0.8 5.615 deproteinisation occurs after MDA reaction with TBA. Method C
Hake sample presented acceptable recovery levels in the range 90–112%. Other
0.6 studies have also reported good recovery levels for DNPH HPLC
method in plasma or liver samples (Mateos et al., 2005).
0.4 1.718 3.263 5.604 Blank
3.3. Analytical reproducibility
0 1 2 3 4 5 6 7 mi
n
Analytical reproducibility was studied with analysis of standard
solutions and sample extracts. Within-run and between-run (Table
B
mAU 2) precision of method B ranged, for the different standard concen-
trations tested, between 0.2–3.1% and 0.3–4.6%, with mean global
values of 1.6% and 2.0%, respectively for within-run and be-
8
tween-run precision. Concerning method C, within-run and be-
12.278
tween-run (Table 2) precision ranged between 4.7–10.2% and
MDA Standard
6 1.0–10.9%, with mean global values of 8.0% and 4.8%, respectively.
(5 μM)
The traditional spectrophotometric TBA test presented coefficients
12.293
of variation (data not shown) ranging from 1.0% to 4.3%, being the
4
Hake sample
highest percentage attained with a between-run analysis of a
2.0 lM solution.
2 Comparison of methods B and C shows that a higher precision
Blank was reached with method B, which can be related to the small
chromatographic area of the MDA-DNPH peak (Fig. 1) and, thus,
0
to a greater proximity of the concentration of the tested solutions
to the limit of quantification (see above) of method C. However,
8 9 10 11 12 13 14 m
high coefficients of variation were also determined for higher con-
Fig. 1. A: HPLC chromatograms of MDA-TBA complex (retention time 5.6 min) in centrations of MDA-DNPH in the within-run experiment which can
blank, MDA standard (5 lM) and hake sample (method B). B: HPLC chromatograms be credited to the instability of the signal observed after prolonged
of MDA-DNPH complex (retention time 12.2 min) in blank, MDA standard (5 lM) equipment operation (after 15 h). For between-run precision,
and hake sample (method C).
method C performance was more similar to method B and, more-
over, coefficients of variation declined from 10.9% for lowest stan-
dard concentration to 1.0% for highest, thus, reinforcing the limit of
spectrophotometric TBA test (method A) and the HPLC methods of quantification vicinity hypothesis. Standard precision values for
MDA-TBA (method B) and MDA-DNPH (method C) adducts analy- MDA-TBA determination were comparable to those mentioned in
sis, respectively. These values were well below the lowest standard literature (Khoschsorur et al., 2000). On the other hand lower co-
concentration, 0.63 lM MDA and in view of the high MDA values efficients of variation, not exceeding 5%, have been reported by
typically found in fish considered suitable for the evaluation of li- other authors working on the DNPH method and on the range
pid oxidation. For HPLC separation of MDA-TBA limits of detection 0.2–20.0 lM (Mateos et al., 2005).
have been reported to range between 0.015 lM (Khoschsorur et al., Within-run and between-run precision of method B results
2000; Tsaknis, Lalas, & Evmorfopoulos, 1999) and 1 lM or more determined with hake and sea bream extracts are presented in Ta-
(Kakuda, Stanley, & Van de Voort, 1981). Regarding chromato- ble 3. Coefficients of variation ranged from 2.1% for 13 within-run
graphic analysis of MDA-DNPH, limits of detection and quantifica- aliquots of the same hake extract to 9.1% for 8 different sea bream
tion were similar to those found in other studies (Mateos et al., extracts in different runs. Regarding method C, coefficients of var-
2005). iation ranged from 5.4% for 13 within-run aliquots of the same sea
In what concerns linearity, method A exhibited a linear re- bream extract to 31.6% for 13 different sea bream extracts within
sponse between 2.00 lM and 10.00 lM and a collection of 15 cal- the same run (Table 3). In comparison to the above precision val-
ibration curves presented a modest correlation coefficient, ues, the traditional spectrophotometric TBA test showed a with-
R2  0.90 (data not shown). On the other hand, both HPLC methods in-run coefficient of variation of the same hake extract of 14.0%
showed a linear response of MDA in a range of concentrations from (MDA content averaged 7.33 lmol/kg; data not shown). As with
0.63 lM to 10.00 lM and calibration curves presented high corre- standard solutions, within-run trial reproducibility of method C
lation coefficients R2 > 0.99 (p = 0.0001; n = 15). determined with different fish extracts was much worse than in
between-run situation, thus, suggesting a run duration effect sim-
3.2. Recovery/Accuracy ilar to the one observed in the standards precision study. However,
this effect was magnified by the vicinity of the method quantifica-
Recovery of the samples analysed according to method B and tion limit, since the sea bream samples analysed had very low MDA
extracted with 7.5% TCA (sample preparation method i) ranged be- contents (below 2 lmol/kg). The importance of this factor is
 Author's personal copy

1042 R. Mendes et al. / Food Chemistry 112 (2009) 1038–1045

Table 1
Recovery data of MDA in fish samples extracts by different analytical methods*

Analytical method MDA in sample (lM MDA ± SD) MDA added (lM MDA) MDA detected (lM MDA ± SD) Recovery(%)
Method A 0.075 ± 0.011 0.15 0.175 ± 0.04 66
0.075 ± 0.011 0.30 0.289 ± 0.13 71
Method B (i) 0.098 ± 0.019 0.15 0.249 ± 0.023 100
0.098 ± 0.019 0.30 0.430 ± 0.004 108
Method B (ii) 0.440 ± 0.004 0.15 0.538 ± 0.047 65
0.440 ± 0.004 0.30 0.667 ± 0.030 75
Method C 4.395 ± 0.478 2.50 6.651 ± 0.146 90
4.395 ± 0.478 5.00 10.013 ± 0.480 112
*Presented values correspond to mean ± standard deviation (n = 4).
(i) Sample extraction with 7.5% TCA.
(ii) Sample extraction with water and deproteinisation with 20% TCA + PCA 0.1125 N (Seljeskog et al., 2006).

Table 2
Reproducibility data (precision within run and from run to run) for the HPLC determinations of the MDA-TBA and MDA-DNPH complex in standard solutions

Expected 0.625 lM Expected 1.250 lM Expected 2.500 lM Expected 5.000 lM Expected 10.000 lM
Measured (lM) Measured (lM) Measured (lM) Measured (lM) Measured (lM)
Within run
Method B: MDA-TBA
Mean 0.585 1.172 2.468 5.083 10.034
SD ± 0.018 0.030 0.012 0.010 0.160
CV(%) 3.1 2.5 0.5 0.2 1.6
Method C: MDA-DNPH
Mean 0.637 1.276 2.516 4.906 10.039
SD ± 0.055 0.131 0.159 0.229 1.006
CV(%) 8.6 10.2 6.3 4.7 10.0
From run to run
Method B: MDA-TBA
Mean 0.645 1.201 2.447 4.992 10.014
SD ± 0.030 0.025 0.040 0.070 0.032
CV(%) 4.6 2.1 1.6 1.4 0.3
Method C: MDA-DNPH
Mean 0.576 1.253 2.548 5.015 9.983
SD ± 0.063 0.045 0.097 0.245 0.100
CV(%) 10.9 3.6 3.8 4.9 1.0

highlighted when a comparison is made between sea bream and (0.6 g fat/100 g fresh weight), sea bream (9.7 g/100 g) and sardine
highly oxidised sardine samples (MDA above 300 lmol/kg, requir- (14.3 g/100 g). In what concerns specificity, methods A, B and C
ing a 1:10 dilution). Within-run coefficient of variation for different presented parallel MDA variations during storage time (Fig. 2).
extracts was in this case quite lower for sardine (9.6% vs 31.6%). The only exception was MDA content determined by method A
This MDA concentration effect was almost absent for method B in sardine from day 15. However, MDA concentrations were sub-
(3.0 vs 3.5%), although contents range was not so wide. Method B stantially different between techniques. In hake and sea bream,
reproducibility was in a precision range comparable to that found method C produced significantly lower (p < 0.05) MDA contents
by other authors using also HPLC separation of MDA-TBA in differ- than methods A and B. In fact, after day 8 in hake and day 4 in
ent biological material, albeit with some technical differences sea bream, a gap was formed between method C values and the
(Khoschsorur et al., 2000; Knight, Smith, Kinder, & Pieper, 1988). others and the difference progressively widened. This suggests that
Variation coefficients of up to 5.6% (between-run), for very low other compounds released by further fish spoilage reacted with
MDA levels (0.74 lmol/kg) were also reported by authors working TBA, while DNPH remained specific and only reacted with MDA.
on HPLC separation of MDA-TBA in fish samples (Bergamo et al., In sea bream, after day 6, whereas MDA measured by methods A
1998). Method C precision was not so good as reported in the liter- and B suffered only minor fluctuations, method C variations were
ature for an equivalent DNPH derivatization method (Fenaille, greater, reaching 0.9 lmol MDA/kg on 11th day.
Mottier, Turesky, Ali, & Guy, 2001), where, for milk powder (4– For sardine and most of the time, method C mean values were
9 lmol MDA/kg), variation coefficients of up to 3% were calculated. lower. The considerable production of MDA in sardine possibly re-
Comparison of methods B and C with the traditional spectrophoto- duces the effect of other thiobarbituric acid reactive substances,
metric technique shows that for MDA concentrations in the same thus, narrowing the difference between TBA and DNPH techniques.
range both HPLC methods B and C have greater reproducibility With respect to former techniques (methods A and B), there were
(2.1% and 5.4% vs 14.0%: data not shown). Analysis of the data significant differences (p < 0.05) only in sea bream and higher val-
shows that both methods have adequate reproducibility for appli- ues were determined by MDA-TBA HPLC method. This may be
cation in routine fish analysis. unexpected since HPLC should separate MDA-TBA from other com-
ponents that interfere in the traditional spectrophotometric TBA
3.4. Methods comparison and specificity test by also absorbing at 530 nm. Though, almost no such compo-
nent was found in method B chromatograms (see Fig. 1). A plausi-
Methods comparison and evaluation of potential in assessing ble explanation for the difference may be related to the low
different levels of lipid oxidation was carried out during chilled recovery values found for method A (see above recovery/accuracy).
storage of three fish species with different fat content, hake This can possibly also explain the relatively low MDA content mea-
 Author's personal copy

R. Mendes et al. / Food Chemistry 112 (2009) 1038–1045 1043

Table 3
Reproducibility data (precision within run and from run to run) for the HPLC determinations of the MDA-TBA and MDA-DNPH complex in hake, sea bream and sardine samples

Hake Same extract (lmol Hake Different extract (lmol Sea bream Different extract Sardine Different extract (lmol
MDA/ kg) MDA/ kg) (lmol MDA/ kg) MDA/ kg)
Within run
Method B: MDA-TBA
Mean 19.89 19.95 7.29
SD ± 0.41 0.54 0.25
CV(%) 2.1 3.0 3.5
Method C: MDA-DNPH
Mean 7.08 1.81 335.55
SD ± 0.38 0.57 32.38
CV(%) 5.4 31.6 9.6
From run to run
Method B: MDA-TBA
Mean 7.47
SD ± 0.68
CV(%) 9.1
Method C: MDA-DNPH
Mean 2.30
SD ± 0.26
CV(%) 11.4

sured by method A in sardine on 15th day. The consistently lower tween MDA and nucleophilic compounds included in fish muscle
values estimated by method C and the fact that these are not linked (namely, protein-type compounds) led to the formation of tertiary
to any recovery problems, show that both methods based on the oxidation products. Similar changes in MDA content were reported
quantification of the MDA-TBA adduct (methods A and B) may by Ozden, Inugur, and Erkan (2007) and Mendes and Gonçalves (in
overestimate MDA and are therefore less accurate. Overestimation press) in sea bass and sea bream, by Rodriguez-Casado et al. (2007)
may result from TBA’s low specificity, but, also, from an artefactual in sardine and by Ruiz-Capillas, Saavedra, and Moral (2003) in
peroxidation of sample constituents (Mateos et al., 2005) even in hake. However, in hake and sea bream, all methods showed a later
the presence of added antioxidants (EDTA and propyl gallate), as MDA increase (near day 15). Similar behaviour was also reported
a consequence of TBA reaction requiring treatment at high temper- by Ruiz-Capillas and Moral (2001) in chilled hake, where a slight
atures for extended incubation times. increasing trend of MDA content over three storage weeks was ob-
Similar conclusions have been reported by other authors com- served. Moreover, in sardine, after an initial peak on 4th day, MDA
paring TBA and DNPH techniques in other kinds of sample, be it maximum was reached on 15th day. Changes during chilled stor-
milk powder (Fenaille et al., 2001), MDA estimates reduced up to age suggest that MDA content — as peroxide value — can be the re-
17-fold with DNPH, or plasma and tissues (Türközkan, Erdamar, sult of a balance between rates of formation and destruction
& Seven, 2006), reductions up to 100-fold. However, MDA content reactions and, as such, the following hypothesis can be formulated:
values were not so drastically reduced by DNPH technique in fish initial MDA content increases occur as lipid oxidation proceeds at
samples. Probably, the cause for this lies much more in the greater slow rate, intermediate declines (as seen between 8th and 11th
levels of lipid oxidation in fish, conducive to stronger MDA produc- day in hake muscle or between 4th and 6th day in sardine) may
tion, than in lower artefactual MDA formation or any MDA overes- be caused by the same modest rates of lipid oxidation coupled with
timation by method C in fish samples. increased MDA degradation, greater increases after several storage
In comparison with published data, MDA levels in hake and sar- days may be due to a markedly high rate of MDA formation and,
dine were relatively high (particularly, exceeding 29 lmol MDA/kg finally, a deceleration of this process brings about a final decline
hake on 8th day and 700 lmol MDA/kg sardine on 15th day). Nev- in MDA concentration (clearly seen in sardine after day 15). Higher
ertheless, about 100 lmol MDA/kg was reported for hake stored in rates of MDA formation may be due to greater release of free iron
ice (Ruiz-Capillas & Moral, 2001). Concerning sardine, in spite of and other prooxidants from the muscle, which becomes increas-
being a much fatter fish and, thereby, more prone to oxidation, lit- ingly degraded when storage time increases (Chaijan et al., 2006).
erature values determined by classical spectrophotometric tech- In order to evaluate the reliability of the HPLC methods the cor-
nique (TBA test) in iced whole sardine reached 250 lmol MDA/kg relation between them was followed (Fig. 3). Whereas, correlation
flesh after 14 days (Nunes, Batista, & Campos, 1992) or, only, coefficients between method A and the HPLC methods were low
111 lmol MDA/kg after nine days (Aubourg, Sotelo, & Gallardo, (0.785 for B and 0.746 for C), correlation coefficient between meth-
1997). However, Chaijan, Benjakul, Visessanguan, and Faustman ods B and C was considerably high (0.974) and similar to the ones
(2006) reported over 8000 lmol MDA/kg in sardine muscle after reported between the MDA-TBA complex determined by HPLC and
15 days. This wide range may arise from different biological condi- by the classical spectrophotometric TBA test in oxidised (0.9794)
tion of fish, post-mortem handling before storage in ice and, also, and non-oxidised (0.9819) traditional fish products (Tsaknis
from a different method used by the latter authors, since ground et al., 1999). Therefore, independent determinations by methods
sample was homogenised with a solution containing TBA and B and C agree to each other, being MDA variation estimated by C
TCA. This simultaneous derivatisation and deproteinisation proce- largely a translation to a lower range of B values (correlation
dure promotes TBA reaction with other sample components be- slope < 1). This, in turn, reinforces the artefactual MDA formation
sides MDA and, thus, may lead to overestimated MDA contents. hypothesis. On the other hand, method A deviations from methods
For all methods, MDA content in fish samples varied during B and C seem to highlight a greater uncertainty in classical spectro-
chilled storage in a broadly typical way for this oxidation product, photometric TBA technique without HPLC separation. However,
namely, an initial phase of MDA content increase followed by a la- standard deviations of the method A results were not particularly
ter phase of decreasing MDA levels in, which the interaction be- higher in comparison to the others (Fig. 2). It should be noticed
 Author's personal copy

1044 R. Mendes et al. / Food Chemistry 112 (2009) 1038–1045

HAKE Correlation between methods A and B

40
800
y = 1.3398x - 5.3101
700 R2 = 0.7854
30 600

Met. B (μmol MDA/kg)

μmol MDA/kg)

500
20 400

300

10 200

100

0 0
0 5 10 15 20 05 0 100 150 200 250 300 350
Time (days) Met. A (μmol MDA/kg)
A B C

Correlation between methods A and C

SEA BREAM 800
30 y = 1.2555x - 13.505
700 R2 = 0.746
600
Met. C (μmol MDA/kg) 500
20
μmol MDA/kg)

400

300
10 200

100

0
0 0 50 100 150 200 250 300 350
0 5 10 15 20 Met. A (μmol MDA/kg)
Time (days)
A B C
Correlation between methods B and C
800
SARDINE y = 0.9491x - 9.605
900 700 R2 = 0.9744
Met. C (μmol MDA/kg)

600

675 500
μmol MDA/kg)

400

300
450
200

100
225
0
0 100 200 300 400 500 600 700 800
Met. B (μmol MDA/kg)
0
0 5 10 15 20
Time (days) Fig. 3. Correlations between MDA levels (lmol MDA/kg) determined by methods A,
A B C B and C in hake, sea bream and sardine samples during chilled storage at 2 ± 1 ± °C.

Fig. 2. Changes in MDA levels (lmol MDA/kg) of hake, sea bream and sardine
during chilled storage at 2 ± 1 °C, measured by methods A, B and C. that in fish samples both developed HPLC methods are fast, simple,
sensitive and stable, as shown by short retention and elution times
(not exceeding 20 min), linearity, acceptable limits of detection
that method A had a great deviation with the sardine sample of day
and quantification, inter- and intra-assay CV precision testing
15, which also explains correlation slopes above the unity (Fig. 3).
and recovery assay. On the other hand, the traditional spectropho-
tometric TBA test presented low recovery levels (under 71%).
4. Conclusions Regarding accuracy and specificity for MDA, methods A and B
(MDA-TBA) were inferior to method C (MDA-DNPH) as a result of
Methods for MDA determination using HPLC separation after possible artefactual peroxidation of sample constituents, namely,
TBA or DNPH derivatisation were developed for routine fish analy- polyunsaturated fatty acids and, also, of TBA reactivity with other
sis and optimized. Overall method performance based on accuracy components besides MDA. However, method C had some disad-
and specificity followed the order, HPLC method MDA- vantages, a relatively high limit of detection (0.20 lM MDA) and
DNPH > HPLC method MDA-TBA > Traditional spectrophotometric a lower reproducibility at lower MDA contents in standards and
TBA test. For sample extraction it was found that TCA deproteiniza- samples. Nevertheless, these are not critical drawbacks for an
tion must be done before the addition of derivatising reagent in or- application in routine fish analysis, given the high MDA concentra-
der to ensure a better MDA extraction. The present study showed tions in oxidised fish. For a finer understanding of lipid oxidation in
 Author's personal copy

R. Mendes et al. / Food Chemistry 112 (2009) 1038–1045 1045

fish, MDA determination by method C should be preferred to TBA malondialdehyde, and application of the method to different biological
materials. Chromatographia, 52(3-4), 181–184.
techniques that only give a rough estimate of lipid oxidation in
Kinter, M. (1995). Analytical technologies for lipid oxidation products analysis.
fish, encompassing thiobarbituric acid reactive substances and Journal of Chromatography, 671, 223–236.
artefactual MDA. The application of the modified HPLC methods Knight, J. A., Smith, S. E., Kinder, V. E., & Pieper, R. K. (1988). Urinary lipoperoxides
in fish samples with different levels of MDA showed that these quantified by lipid chromatography, and determination of reference values for
adults. Clinical Chemistry, 34(6), 1107–1110.
methods are useful for samples with low amounts of oxidation Mateos, R., Lecumberri, E., Ramos, S., Goya, L., & Bravo, L. (2005). Determination of
products, as well, for samples with high levels of oxidation. malondialdehyde (MDA) by high-performance liquid chromatography in serum
and liver as a biomarker for oxidative stress. Application to a rat model for
hypercholesterolemia and evaluation of the effect of diets rich in phenolic
Acknowledgment antioxidants from fruits. Journal of Chromatography B, 827, 76–82.
Mendes, R., & Gonçalves, A. (2008). Effect of Soluble CO2 Stabilization on the Quality
This study was financial supported by EU-QCA III-MARE/FEDER: of Fillets from Farmed Gilthead Sea Bream (Sparus aurata) and European Sea
Bass (Dicentrarchus labrax). Journal of Aquatic Food Product Technology, 17(4),
Project ‘‘Quality and Innovation of Fishery Products”. 19–32.
Nunes, M. L., Batista, I., & Campos, R. M. (1992). Physical, chemical and sensory
analysis of sardine (Sardina pilchardus) stored in ice. Journal of the Science of Food
References and Agriculture, 59, 37–43.
Ozden, O., Inugur, M., & Erkan, N. (2007). Effect of gamma radiation and
Aubourg, S. P. (1999). Lipid damage detection during the frozen storage of an refrigeration on the chemical and sensory properties and microbiological
underutilized fish species–denaturation of fish proteins during frozen storage: status of aqua cultured sea bass (Dicentrarchus labrax). Radiation Physics and
Role of formaldehyde. Food Research International, 32(6), 497–502. Chemistry, 76, 1169–1178.
Aubourg, S. P., Sotelo, C. G., & Gallardo, J. M. (1997). Quality assessment of sardines Panpipat, W., & Yongsawatdigul, J. (2008). Stability of potassium iodide and omega-
during storage by measurement of fluorescent compounds. Journal of Food 3 fatty acids in fortified freshwater fish sausage. Lebensmittel Wissenschaft und
Science., 62(2), 295–304. Technologie, 41, 483–492.
Bergamo, P., Fedele, E., Balestrieri, M., Abrescia, P., & Ferrara, L. (1998). Pikul, J., Leszczynski, D. E., & Kummerow, F. A. (1989). Evaluation of three modified
Measurement of malondialdehyde levels in food by high-performance liquid TBA methods for measuring lipid oxidation in chicken meat. Journal of
chromatography with fluorometric detection. Journal of Agricultural and Food Agricultural and Food Chemistry, 37, 1309–1313.
Chemistry, 46, 2171–2176. Pilz, J., Meineke, I., & Gleiter, C. H. (2000). Measurement of free and bound
Bird, R. P., Hung, S. S. O., Hadley, M., & Draper, H. H. (1983). Determination of malondialdehyde in plasma by high-performance liquid chromatography as the
malonaldehyde in biological materials by high-pressure liquid chromatography. 2,4-dinitrophenylhydrazine derivative. Journal of Chromatography B, 742(2000),
Analytical Biochemistry, 128, 240–244. 315–325.
Careche, M., & Tejada, M. (1988). Interference by formaldehyde in the 2- Rodriguez-Casado, A., Carmona, P., Moreno, P., Sánchez-González, I., Macagnanoc,
thiobarbituric acid test for rancidity. Journal of the Science of Food and A., Di Natalec, C., & Careche, M. (2007). Structural changes in sardine (Sardina
Agriculture, 43, 49–58. pilchardus) muscle during iced storage: Investigation by DRIFT spectroscopy.
Cesa, S. (2004). Malonaldehyde contents in infant milk formulas. Journal of Food Chemistry, 103(3), 1024–1030.
Agricultural and Food Chemistry, 52, 2119–2122. Ruiz-Capillas, C., & Moral, A. (2001). Correlation between biochemical and sensory
Chaijan, M., Benjakul, S., Visessanguan, W., & Faustman, C. (2006). Changes of lipids quality indices in hake stored in ice. Food Research International, 34, 441–447.
in sardine (Sardinella gibbosa) muscle during iced storage. Food Chemistry, 99, Ruiz-Capillas, C., Saavedra, A., & Moral, A. (2003). Hake slices stored in retail
83–91. packages under modified atmospheres with CO2- and O2-enriched gas mixes.
Cighetti, G., Debiasi, S., Paroni, R., & Allevi, P. (1999). Free and total European Food Research and Technolology, 218, 7–12.
malondialdehyde assessment in biological matrices by gas chromatography– Sakai, T., Habiro, A., & Kawahara, S. (1999). High-performance liquid
mass spectrometry: what is needed for an accurate detection. Analytical chromatographic analysis of 1,3-diethyl-thiobarbituric acid-malonaldehyde
Biochemistry, 266(8), 222–229. adduct in fish meat. Journal of Chromatography B, 726, 313–316.
De las Heras, A., Schoch, A., Gibis, M., & Fischer, A. (2003). Comparison of methods Seljeskog, E., Hervig, T., & Mansoor, M. A. (2006). A novel HPLC method for the
for determining malondialdehyde in dry sausage by HPLC and the classic TBA measurement of thiobarbituric acid reactive substances (TBARS). A comparison
test. European Food Research and Technology, 217, 180–184. with a commercially available kit. Clinical Biochemistry, 39, 947–954.
Draper, H. H., & Hadley, M. (1990). Malondialdehyde determination as index of lipid Sinnhuber, R. 0., & Yu, T. C. (1958). 2-Thiobarbituric acid method for the
peroxidation. Methods in Enzymology, 186, 421–431. measurement of rancidity in fishery products II. The quantitative
Ekström, T., Garberg, P., Egestad, B., & Högerg, J. (1988). Recovery of determination of malonaldehyde. Food Technology, 12, 9–12.
malondialdehyde in urine as a 2,4-dinitrophenylhidrazine derivative analysed Sinnhuber, R. O., Yu, I. C., & Yu, T. C. (1958). Characterization of the red pigment
with high-performance liquid chromatography. Chemico-Biological Interactions, formed in the 2-thiobarbituric acid determination of oxidative rancidity. Food
66(3-4), 177–187. Research, 23, 624–634.
Fenaille, F., Mottier, P., Turesky, R. J., Ali, S., & Guy, P. A. (2001). Comparison of Squires, E. J. (1990). High-performance liquid chromatographic analysis of the
analytical techniques to quantify malondialdehyde in milk powders. Journal of malondialdehyde content of chicken liver. Poultry Science, 69, 1371–1376.
Chromatography A, 921, 237–245. St. Angelo, A. J. (1996). Lipid oxidation in foods. Critical Reviews in Food Science and
Fernández, J., Pérez-Álvarez, J. A., & Fernández-López, J. A. (1997). Thiobarbituric Nutrition, 36, 175–224.
acid test for monitoring lipid oxidation in meat. Food Chemistry, 59(3), 345–353. Stalikas, C. D., & Konidari, C. N. (2001). Analysis of malondialdehyde in biological
Guillen-Sans, R., & Guzman-Chozas, M. (1998). The thiobarbituric acid (TBA) matrices by capillary gas chromatography with electron-capture detection and
reaction in foods: a review. Critical Reviews in Food Science and Nutrition, 38(4), mass spectrometry. Analytical Biochemistry, 290(8), 108–115.
315–330. Tarladgis, B. G., Pearson, A. M., & Dugan, L. Jr., (1964). Chemistry of the 2-
Jardine, D., Antolovich, M., Prenzler, P. D., & Robards, K. (2002). Liquid thiobarbituric acid test for determination of oxidative rancidity in foods. II.
chromatography–mass spectrometry (LC–MS) investigation of the Formation of the TBA malonaldehyde complex without acid heat treatment.
thiobarbituric acid reactive substances (TBARS) reaction. Journal of Agricultural Journal of the Science of Food and Agriculture, 15, 602–607.
and Food Chemistry, 50, 1720–1724. Tsaknis, J., Lalas, S., & Evmorfopoulos, E. (1999). Determination of malondialdehyde
Kakuda, Y., Stanley, D. W., & Van de Voort, F. R. (1981). Determination of TBA in traditional fish products by HPLC. The Analyst, 124, 843–845.
number by high-performance liquid chromatography. Journal of the American Türközkan, N., Erdamar, H., & Seven, I. (2006). Measurement of total
Oil Chemistry Society, 58, 773–776. malondialdehyde in plasma and tissues by high-performance liquid
Ke, P. J., Cervantes, E., & Robles-Martinez, C. (1984). Determination of thiobarbituric chromatography and thiobarbituric acid assay. Fırat Tıp Dergisi, 11(2), 88–92.
acid reactive substances (TBARS) in fish tissue by an improved distillation Volpi, N., & Tarugi, P. (1998). Improvement in the high-performance liquid
spectrophotometric method. Journal of the Science of Food and Agriculture, 35, chromatography malondialdehyde level determination in normal human
1248–1254. plasma. Journal of Chromatography B, 713, 433–437.
Ke, P. J., & Woyewoda, A. D. (1979). Microdetermination of thiobarbituric acid Vyncke, W. (1970). Direct determination of the thiobarbituric acid value in
values in marine lipids by a direct spectrophotornetric method with a trichloroacetic extracts of fish as a measure of oxidative rancidity. Fette Seifen
monophasic reaction system. Analytical Chimica Acta., 106, 279–284. anstrichm., 12, 1084–1087.
Khoschsorur, G. A., Winklhofer-Raab, B. M., Rabl, H., Auer, T., Peng, Z., & Schaur, R. J. Zipser, M. W., Kwon, T. W., & Watts, B. M. (1964). Changes in cured and uncured
(2000). Evaluation of a sensitive HPLC method for the determination of frozen cooked pork. Journal of Agricultural and Food Chemistry, 12, 101–109.

View publication stats