urn

b

O BHod, D

forAvailable online 4 April 2006

MDA is a three-carbon dialdehyde that can exist in variousforms in aqueous solutions depending on the pH of the solu-

pholipids, and aldehydes [18]) to produce a pink chromophorethat can be measured by UV or fluorescence detection. Thesesubstances are termed TBARS (thiobarbituric acid reactingsubstances). Because MDA seems to be the main aldehyde that

Clinical Biochemistry 39 (20Introduction

Malondialdehyde (MDA) is generated as a relatively stableend product from the oxidative degradation of polyunsaturatedfatty acids (PUFA) [1]. This free radical-driven lipid peroxida-tion has been causatively implicated in the aging process [26],as well as in atherosclerosis [712], Alzheimer's disease[13,14], and cancer [15]. Plasma MDA has thus been used asa biomarker of lipid peroxidation and has served as an indicatorof free radical damage in the conditions mentioned above.

tion [16]. At acidic pH (pH b4.5) MDA is found mainly as -hydroxyacrolein in equilibrium with a dicarbonyl form. Inneutral and alkaline conditions, the predominant form of MDAis as an enolate anion. Initial acid and basic hydrolysis of plasmasamples contribute to the measurement of free or total levels ofTBARS, respectively.

Themeasurement of plasmaMDAhas been carried out by thethiobarbituric acid test [17]. When heated under acidicconditions, thiobarbituric acid (TBA) reacts with a number ofchemical species (nucleic acids, amino acids, proteins, phos-Abstract

Objectives: Malondialdehyde (MDA) as a part of thiobarbituric acid reacting substances (TBARS) is frequently used as an indicator of lipidperoxidation. Most methods for the measurement of TBARS require long derivatization time and addition of antioxidants in the samples.Furthermore, comparison of these methods with commercially available HPLC kits is lacking.

Design and methods: We investigated column performance of five different columns, tested eight different acids for the hydrolysis of thesamples, and estimated stability of derivatized plasma samples with different anticoagulants. The samples were derivatized with TBA. The peakfor the TBA2MDA adduct was separated and detected by HPLC.

Results: Performance of the Phenomenex Gemini column was best. PCA at the concentration of 0.1125 N was used in this method. Coefficientof variation (CV %) within the run and between the run was 4.1% and 6.7%, and analytical recovery was 9094%. The retention time of theTBA2MDA peak was 1.8 min.

Reference intervals for TBARS in serum from 250 individuals were 0.53 and 2.1 mol/L using our HPLC method and 0.07 and 0.24 mol/Lusing the Chromsystems assay. Linear regression with log converted values revealed weak relationship between the two methods (r2 = 0.064).

Conclusions: Our HPLC method for the analysis of TBARS in serum and plasma is fast and accurate and therefore can be used in clinicalstudies. 2006 The Canadian Society of Clinical Chemists. All rights reserved.A novel HPLC method for the meassubstances (TBARS). A compariso

Esben Seljeskog a,, Tor Herviga Division of Medical Biochemistry, P

b Haukeland Universityc Agder University College, Kristiansan

Received 3 February 2006; revised in revisedAbbreviations: MDA, malondialdehyde; TBARS, thiobarbituric acidreacting substances; TEP, 1,1,3,3-tetraethoxypropane. Corresponding author. Fax: +47 51519907.

E-mail address: sees@ext.sir.no (E. Seljeskog).

0009-9120/$ - see front matter 2006 The Canadian Society of Clinical Chemistsdoi:10.1016/j.clinbiochem.2006.03.012ement of thiobarbituric acid reactivewith a commercially available kit

, Mohammad Azam Mansoor c

ox 8100, N-4068 Stavanger, Norwayspital, Bergen, Norwayepartment of Natural Sciences, Norway

m 21 March 2006; accepted 22 March 2006

06) 947954reacts with TBA [19] and because TBARS measurements areusually correlated with standards made from pure MDA (acidhydrolysis of 1,1,3,3-tetrahydroxypropane; TEP), researchers

. All rights reserved.

iocoften use the term MDA when measuring TBARS. Forsimplicity, the term MDAwill also be used in this article.

Sinnhuber et al. characterized the chromogen produced in theTBA test as a condensation product of two molecules of TBAwith one molecule of MDA: This TBA2MDA complex can bemeasured by HPLC separation [6,2027]. Using UV/visibledetection, the maximum absorption of the complex has beenfound at 532 nm (acetic acid solution).

TBARS reference values for several individuals, with aselection between age groups and gender, have been reportedusing EDTA-treated plasma [6,32].

The aim of this study was to develop a new HPLC-basedmethod for analysis of TBARS and to determine reference valuesfrom serum samples. The reference sample group was an inter-national, Nordic population, divided into age groups and gender.

We also compared the analytical performance of our assaywith the Chromsystems malondialdehyde assay for the mea-surement of TBARS in serum samples.

Methods

Chemicals and reagents

All chemicals were of HPLC grade. Perchloric acid (PCA-HClO4), potassium dihydrogen phosphate (KH2PO4), potassiumhydroxide (KOH), 2-thiobarbituric acid (2-TBA; C4H4N2O2S),and 1,1,3,3-tetraethoxypropane (TEP; C11H24O4) were pur-chased from Sigma (St. Louis, MO, USA)/Fluka, and trichlor-oacetic acid (TCA; CCl3COOH), methanol (CH3OH), andacetonitrile (LiChrosolv for chromatography: CH3CN)was fromMerck (Darmstadt, Germany).

Sterile water was used for all the solutions

A 40-mM 2-TBA solution was prepared by dissolving576 mg of 2-TBA in 80 ml of water, and then heating to 55C for45 min in a water bath. The solution was cooled to room tem-perature and filled with water to 100 ml.

Buffer: A 50-mM KH2PO4 solution was prepared andadjusted to pH 6.8 with KOH. The buffer was then filteredthrough a Millipore membrane filter (Millipore, Billerica, MA)with a pore size of 0.45 m.

HPLCmobile phase consisted of 72:17:11 (by vol) KH2PO4methanolacetonitrile. The mobile phase was degassed beforeuse both by helium and a Shimadzu GT-154 vacuum degasser.

Reagent kit for HPLC analysis of malondialdehyde inplasma/serum was obtained from Chromsystems Instrumentsand Chemicals (Cat. no. 67,000, Chromsystems Instrumentsand Chemicals GmbH, Munich, Germany).

HPLC instrumentation

High performance liquid chromatographic analysis wasperformed using a Shimadzu HPLC system (Shimadzu

948 E. Seljeskog et al. / Clinical BCorporation, Kyoto, Japan): Pump (LC-10AT), autosampler(SIL-10AD), fluorescence detector (RF-10AXL), degasser(GT-154), and system controller (SCL-10A) with a PC controlprogram (Shimadzu Class VP, version 6.12 SP2). We tested thefollowing columns in our assay:

(a) 5.0, 10.0, and 15.0 0.3 cm Hypersil ODS C18 (ThermoElectron Corporation, USA). Columns were equippedwith a guard column (Thermo Uniguard direct-connectiondrop-in guard cartridge holder for 10 mm drop-in car-tridges, 3 mm bore, Hypersil ODS C18.

(b) A 5.0 0.32 cm Varian Polaris C8-A column with 5 particles (Varian Inc., CA, USA).

(c) A wide range pH-stable (pH 114), 5.0 0.3 cmPhenomenex Gemini C18 column (Phenomenex Inc., CA,USA) with 5 particles and equipped with a guard column(Phenomenex Securityguard drop-in guard cartridge holderfor 4 mm stacking rings, 3 mm bore, ODS C18).

In the present method, the Phenomenex Gemini column waschosenwith a buffer flow rate of 0.8mL/min, sample runwas 4min,injection volume 10 L and spectrofluorimetric detector wave-lengths were set at 525 nm (excitation) and 560 nm (emission).

The same HPLC system was used for the Chromsystems kitwith a 10-cm column (diameter and packing material not listed).Elution flow rate was 1.0 mL/min, sample run was 5 min, injec-tion volume was 20 L, and spectrofluorimetric detector wave-lengths were set to 515 nm (excitation) and 553 nm (emission),according to the manufacturer's instructions.

Biological materials

Serum samples from 250 healthy subjects (123 female and127 male) were acquired from the NOBIDA biobank (NordicReference Interval Project Bio-bank and Databasea databasemanaged by NORIP; Nordic Reference Interval Project [28]).Subjects had given informed consent to NORIP, and permissionfrom the Norwegian ethical committee of the NORIP wasgranted. Selection was performed from the biobank's 3036reference individuals by the NOBIDA Committee based on thecriteria of equal number of men and women, ages 1884 yearsfrom individuals living in Norway (28%), Denmark/Iceland(24%), Sweden (20%), and Finland (28%). Serum samples werestored at 80C until they were analyzed.

EDTA plasma used for development was obtained from thelocal blood bank (Stavanger University Hospital) and labora-tory colleagues. Plasma was either processed immediately orstored at 80C until analysis. Derivatized samples were storedeither at 80C, 20C, +4C, and +22C.

Standard and sample preparation

Plasma/serum or standard/blank (50 L) was mixed withPCA (0.1125 N, 150 L) and TBA (40mM, 150 L) in a 1.5-mLscrew cap Eppendorf tube, vigorously mixed for 10 s, and placedin a heating cabinet at 97C for 60 min. After cooling in a freezerat 20C for 20 min, methanol (300 L) and 20% TCA (100 L)

hemistry 39 (2006) 947954were added to the suspension and mixed for 10 s. The sampleswere centrifuged at 13,000g for 6 min, and 100 L of thesupernatant was transferred to autosampler vials. As acid

BiocE. Seljeskog et al. / Clinicalhydrolysis of TEP yields stoichiometric amounts of MDA, astandard curve was made from TEP dissolved in methanol anddiluted in water at concentrations of 10.0, 5.0, 2.5, 1.25, 0.62,0.31, 0.16 M, and blank. TEP standards were heated at 50C for60 min and were stored in a refrigerator for maximum 1 week.

Samples (2040 NOBIDA serum samples per day) werethawed, derivatized, and measured with our own method for1 day and with the Chromsystems assay the following day.Serum samples were refrozen at 20C in between 1st and2nd day.

EDTA plasma aliquots were frozen in 1.5-mL screw capEppendorf tubes at 80C and were thawed just before analyses.

Analytical performance

Within-run precision was calculated from 10 samples doneon the same day, whereas between-run precision was cal-culated from 10 samples done three times over a period of17 days. Peak maximum by acid type when acidifying plasma/serum/standard and TBA were tested with 8 different acids(perchloric acid, acetic acid, hydrobromic acid, phosphoricacid, phosphinic acid, nitric acid, sulfuric acid, and hydro-chloric acid) and at three different concentrations (0.1125,0.075, and 0.0375 N).

Fig. 1. Typical HPLC chromatograms of the MDATBA2 adduct in (A) a blank samplL P-MDA. The recurring peak with a retention time of 2.6 min is unidentified.949hemistry 39 (2006) 947954The assay from Chromsystems Instruments and ChemicalsGmbH

The 100 L of plasma/standards was mixed with 500 L ofprecipitation reagent in a 1.5-ml light-protected vial andmixed for 10 s. After 5 min of centrifuging at 13,000 rpm,500 L of the supernatant was transferred to a glass vial. 100 Lof derivatization reagent was added, the mixture mixedbriefly, and the vial was incubated at 95C for 60 min in aheating cabinet. After cooling in a freezer at 20C for 20 min,500 L of Neutralisation buffer was added to the vial. After abrief mixing, the mixture was pipetted into an HPLC vial forelution. Injection volume was 20 L.

Statistics

Statistical analyses of relations between TBARS and age andgender, and relations between our HPLC method and theChromsystems assay were performed with Statview version5.0.1 (SAS Institute) and SPSS 13.0 (SPSS, Inc.). Scatterplot analysis was performed on log-converted values. Referenceintervals were calculated as recommended by IFCC, withRefVal 4.0 [29]. Data were standardized to zero mean and unitvariance before exponential and modulus transformation to

e, (B) TEP standard, and (C) plasma from a healthy volunteer containing 0.7 M/

adjust for skewness and kurtosis. The final distribution was notsignificantly different from Gaussian (AndersonDarling'sA2 = 0.146, P=1.0) using our method. Transformed distribu-tion of Chromsystems assay data was not Gaussian, however,and this makes parametric estimates unreliable for these data.

Results

HPLC analysis and method development

Optimal conditions for separation of the TBA2MDA adductwere established by evaluating (a) trials of several commercialHPLC columns, (b) evaluations of a series of phosphate bufferacetonitrilemethanol mixtures, and (c) varying the pro ratavolumes and concentrations of derivatizing chemicals. Initialmethod development was done using EDTAplasma beforeestablishing the serum reference intervals.

The three types of columns produced equally satisfactorychromatograms, but the stability was greatly enhanced by the

Ten plasma aliquots were measured 3 times within a 17-dayperiod. Within-run and between-run CVs for these 10 plasmaaliquots were 4.1% (range 2.64.0%) and 6.7%, respectively.Analytical recovery was 9094% when 2.5 M TEP standardwas added to plasma (data not shown).

Limit of quantification was around 0.05 mol/l, when adding5 standard deviations of the blank measurement to the blankmeasurement [30].

Hydrolyzing acid

Plasma was acidified with 8 different acids. Perchloric acid at0.1125 N produced the highest peak of TBA adducts (Fig. 3) andwas therefore chosen as the acid for further measurements.

Anticoagulants and storage

We measured stability of derivatized plasma MDA with anti-coagulants EDTA, heparin, citrate, and serum. Results forderivatized samples kept for 28 days are shown in Fig. 4. Sampleswere stable when stored at 80C and 20C and showedincreasingly higher levels when stored at +4C and +22C,respectively. This pattern was the same with all anticoagulantsand serum.

The Chromsystems assay

950 E. Seljeskog et al. / Clinical BiocPhenomenex wide range pH-stable Gemini column. Increasedsystem pressure that was noticeable after 100 runs with theHypersile C18 columns did not occur after 800 runs with thePhenomenex Gemini column.

The HPLC profiles of a reagent blank, 0.62 mol/L standard,and a plasma sample are shown in Fig. 1. The retention time ofthe TBA2MDA adduct was 1.8 min at flow rate of 0.8 mL/min.Run time per sample were 4 min. The mean area of the smallpeak obtained with the blank control (H2O) was subtracted fromthe peak heights of calibration and samples.

Calibration

The standard curve generated with TEP standards from 0.0 to10.0 M was linear with a correlation coefficient r = 0.9991(Fig. 2).Fig. 2. Standard curve generated with TEP standards showing correlationbetween peak area and MDA concentration. Correlation coefficient for theregression line was r = 0.9991 in the TEP range from 0.0 to 10 M/L.Precision and recovery

Fig. 3. Peak area diagram of 8 different acids at concentrations 0.037, 0.075, and0.1125 N tested to find the optimal formation of the MDATBA2 adduct. Eachbar represents the mean MDATBA2 adduct peak area of three parallells.

hemistry 39 (2006) 947954A standard curve generated with TEP standards from 0.0 to10.0 M was linear with a correlation coefficient r = 0.991.

The retention time of the TBA2MDA adduct was 2.8 min atflow rate of 1.0 mL/min. The chromatograms showed a single

Optimal separation performance was accomplished by the newPhenomenex Gemini column, which is pH stable between pH 1and 14. There was no need for elaborate cleaning procedures ofthe column as described by Wong et al. [26].

Fig. 4. Storage of derivatized samples with EDTA as anticoagulant, peak area ofMDATBA2 adduct after 0, 1, 2, 3, 4, 7, 14, 21, and 28 days.

E. Seljeskog et al. / Clinical Biocclean peak (data not shown).

Serum reference intervals for TBARS

The reference intervals (parametric estimates) were definedby the 0.025 and 0.975 percentiles and estimated as recom-mended by the IFCC. Age intervals of 1839, 4059, and 6084 years and intervals for men and women separately andtogether are given in Table 1.

Values are listed both for our HPLC method and for theChromsystems assay (see Table 1).

Mean SD TBARS measurements for 250 men andwomen were 1.086 0.432 mol/L using our HPLC methodand 0.135 0.043 mol/L using the Chromsystems assay.Analysis of variance was used to assess relations between ourHPLC method and the Chromsystems assay. The analysisTable 1Reference intervals for TBARS (parametric estimate)

Source ofsamples

Our HPLC method Chromsystems assay

Ageintervals

0.025 and0.975percentiles(mol/L)

0.90confidenceintervals(mol/L)

0.025 and0.975percentiles(mol/L)

0.90confidenceintervals(mol/L)

1839 years 0.51 0.480.55 0.07 0.070.08(n = 90) 2.08 1.902.31 0.23 0.200.294059 years 0.39 0.290.48 0.07 0.070.08(n = 102) 2.05 1.812.35 0.25 0.220.286084 years 0.53 0.470.60 0.07 0.060.08(n = 58) 2.01 1.752.37 0.23 0.210.26Men 1884 0.61 0.580.64 0.07 0.060.08(n = 127) 2.02 1.872.20 0.25 0.220.28Women 1983 0.36 0.280.43 0.08 0.070.08(n = 123) 2.17 1.902.52 0.23 0.210.26Men andwomen

0.53 0.510.56 0.07 0.070.08

(n = 250) 2.10 1.952.27 0.24 0.220.26Mean total 1.0860.43 0.130.04revealed no significant difference in age or gender with eithermethod. A scatter plot of our method against Chromsystemsassay is shown in Fig. 5. Linear regression with log convertedvalues revealed only weak relations between the two methods(r2 = 0.064).

Discussion

The present study shows that our HPLC method for thedetection of TBARS in serum and plasma is fast, accurate, andstable as shown by short retention and elution times, the lowinter- and intra-assay CV precision testing, and recovery assay.

Fig. 5. Scatter plot of our own method against the Chromsystems assay. Linearregression with log-converted values revealed only weak relation between thetwo methods (r2 = 0.064).

951hemistry 39 (2006) 947954Increased lipoperoxide levels have been correlated toatherosclerosis, Alzheimer's disease, cancer, as well as theaging process [215]. Therefore, during the last decades, a lotof work has been done to refine measurement of lipidperoxidation products. Quantitative assessment of conjugateddienes, lipid hydroperoxides, alkanes, aldehydes, and isopros-tanes have been extensively studied as the effects of free radicaldamage have received increasing attention [17].

Since Yagi [31] applied the TBA assay to estimatelipoperoxide concentrations in human serum, it has been apopular method of TBARS detection. From the assay's initiallow specificity as a measurement of MDA levels byfluorometry, the introduction of HPLC after derivatization hasincreased the specificity. As the measured levels of MDA/TBARS depend on a variety of factors, the correlations betweenstudies have been difficult. Table 2 shows plasma and serumTBARS measurements by various studies.

EDTA plasma TBARS measurements seem to produce lowervalues than heparin, citrate, and serum samples [6,20,26,32].This is probably due to the chelation of iron by EDTA that limits

uth

n

23141010

104030304040

iocfurther in vitro peroxidation. Iron is necessary in thesuperoxide-driven Fenton reaction and in the iron-catalyzedHaberWeiss reaction [33] that produces hydroxyl radicalsleading to increased lipid peroxidation and increased levels ofTBARS. Studies using EDTA plasma will then detect lowervalues of TBARS than studies using heparin, citrate, or vialswithout anticoagulants when sampling blood.

In plasma, MDA may exist both in free and in hydrolysablebound forms. Pryor and Stanley [34] proposed a possiblemechanism for the formation of MDA from the precursors ofPUFAs, arachidonic acid (20:4), and docosahexaenoic acid(22:6) breakdown by thermal or acid-catalyzed reactions.

For the acid hydrolysis, however, different acids have beenused. Wong, Knight, Koschsorur, Londero, and Nielsen et al. all

Table 2TBARS measured by various authors

Authors Concentration of TBARS measured by various a

EDTA plasma (mol/l)

Knight et al. (1987) 0.60 0.21 (men)0.54 0.20 (women)

Nielsen et al. (1997) 0.41 1.29 (men), 0.025 and 0.975 fractals0.33 1.22 (women),0.025 and 0.975 fractals

Londero and Lo (1996) 0.85 0.25Khoschsorur et al. (2000) 0.70 0.15Carbonneau et al. (1991) 0.429 0.048 (total)

0.382 0.049 (bound)0.043 0.007 (free)

Templar et al. (1999) 0.11 0.03Seljeskog et al. (2006) Own method

Chromsystems assay

952 E. Seljeskog et al. / Clinical Bused orthophosphoric acid, Sinnhuber et al. used hydrochloricacid, whereas Carbonneau et al. used perchloric acid. We tested8 different acids at three different concentrations for the initialacidifying step of plasma (Fig. 3). It seems that the stronger theability of an acid to oxidize, the more bound MDA will beliberated and more MDA may be produced by PUFA oxidation.Perchloric acid with the lowest pKa released the highest amountof MDA by hydrolysis. Hence, for our new HPLC method,perchloric acid became the acid of choice.

MDA is also formed by decomposition of lipid peroxidesduring the acid heating stage of the TBA assay. In studies withperoxidizing fish oil as much as 98% of the MDA that reacted inthe TBA assay was formed during the acid heating [19,35].Deproteinization and removal of lipids early in sample treatmentmay thus influence heavily on measurements.

The method of Wong et al. demanded elaborate columncleaning procedures for stable elution over a longer time [26].This may be due to the fact that Wong only employed a NaOHmethanol solution as a precipitation agent. Comparison of pro-tein precipitation efficiency done by Polson et al. [36] showedthat methanol was less efficient for protein precipitation thanTCA. Our method comprised both methanol and TCA preci-pitation, which may facilitate an increase in the precipitationefficiency, as the two techniques have different modes of proteinprecipitation.

Previously, Agarwal and Chase [37] published an HPLCmethod for the measurement of TBARS in plasma. They addedthe antioxidant butylated hydroxytoluene (BHT) to stabilizetheir samples. They have neither tested different acids in theprocess of derivatization nor the effect of different anticoagulanton the concentrations of TBARS. They have not compared theirHPLC method with any commercially available kits.

Wong et al. used EDTAblood and found TBARSreference values for men and women with mean SD at0.60 0.13 mol/L. Knight et al. measured TBARS in EDTA(0.58 mol/L), citrate (0.88 mol/L), heparin (1.13 mol/L),and serum (0.79 mol/L) and established age and sex reference

ors Referenceno.

Serum (mol/l) n

0 0.79 0.23 23 [32]87 1.16 13 [6]6

4 [27][22]

0.454 0.066 (total) [20]0.398 0.068 (bound)0.042 0.008 30

[25]0.61 2.02 (men), 0.025 and 0.975 fractals 1270.36 2.17 (women), 0.025 and 0.975 fractals 1230.07 0.25 (men), 0.025 and 0.975 fractals 1270.08 0.23 (women), 0.025 and 0.975 fractals 123

hemistry 39 (2006) 947954values for men and women in EDTA plasma. Nielsen et al. alsodifferentiated between age and sex as Knight et al. and definedreference intervals by the 0.025 and 0.975 percentiles asrecommended by the IFCC [38] (men: 0.41 and 1.29 mol/L;women 0.33 and 1.22 mol/L in EDTA plasma). Nielsen et al.further refined the chromatograms by eliminating interferingpeaks. Khoschsorur et al. used a method similar to Nielsen et al.and reduced the retention time from 9.67 to 2.53 min (EDTAplasma values of 0.70 0.15 mol/L).

We measured TBARS in serum from 250 men and womenand found a total mean value of 1.0860.43 mol/L using ourHPLC method and 0.1350.04 mol/L using the Chromsystemsassay. The values found using our HPLC method are slightlyhigher that Knight et al. and slightly lower than Nielsen et al.

The lower values by Knight et al. may be due to the weakeracid used in the initial acid hydrolysis, whereas the higher valuesby Nielsen et al. may be due to the lack of a final precipitation byTCA.

The low values measured by the Chromsystems assay may bedue to a first deproteinisation step as done also by Carbonneau etal. and Templar et al., who measured 0.429 0.048 and 0.11 0.03 mol/L, respectively. This is in accordance with the findingsby Sinnhuber et al. that much of the MDA is formed during the

Biocheating stage of the derivatization process, such that if we removeproteins before heating, MDA measurements will be lower.Carbonneau et al. measured higher values than Templar et al.,which should be explained by Carbonneau's use of PCA as aprecipitating agent rather thanTCAbyTemplar et al., as PCAmaynot equal TCA as a precipitating agent.

Correlation between our HPLC method and the ChromsystemsMDA assay

The Chromsystems MDA assay weakly correlated to ourHPLC method for the detection of TBARS in serum. A log-converted scatter plot of values measured by the two methodsgave an r2 of 0.064.

Table 1 shows TBARS measurements with the two methods,values grouped by age intervals and sex and a mean total.TBARS concentrations measured with the two methods differby a factor of around eight, where Chromsystems assay give thelowest values (1.09 0.44 mol/L by our HPLC methodcompared to 0.13 0.04 mol/L by the Chromsystems kit).

Although the two methods differ in initial deproteinisation,there could still be expectations of a better correlation betweenthe two methods. Why there is such a weak correlation remainsto be solved.

The mean serum values measured by the Chromsystemsassay, however, correlates with the EDTA plasma measure-ments done by Templar et al. They measured EDTA plasmaMDA (TBARS) levels to 0.11 0.03 mol/L. They deprotei-nised plasma as a first step, as in the ChromsystemsMDA assay.And as serum values theoretically should yield higher valuesthan EDTA plasma values, it fits nicely with the slightly higherChromsystems assay serum measurements

Conclusion

Our HPLC method for the measurement of TBARS isaccurate, fast, and reliable; therefore, the present method shouldbe considered for use in clinical studies. A weak correlationbetween the concentrations of TBARS measured in 250samples by our method with a commercial kit may be due tothe different methods of derivatization.

Acknowledgments

We wish to thank Dr. Anne Bakken for reviewing themanuscript.

References

[1] Horton AA, Fairhurst S. Lipid peroxidation and mechanisms of toxicity.Crit Rev Toxicol 1987;18:2779.

[2] Poon HF, Calabrese V, Scapagnini G, Butterfield DA. Free radicals andbrain aging. Clin Geriatr Med 2004;20:32959.

[3] Marnett LJ. Oxyradicals and DNA damage. Carcinogenesis 2000;21:36170.

E. Seljeskog et al. / Clinical[4] Schmitt-Schillig S, Schaffer S, Weber CC, Eckert GP, Muller WE.Flavonoids and the aging brain. J Physiol Pharmacol 2005;56(Suppl 1):2336.[5] Balaban RS, Nemoto S, Finkel T. Mitochondria, oxidants, and aging. Cell2005;120:48395.

[6] Nielsen F, Mikkelsen BB, Nielsen JB, Andersen HR, Grandjean P. Plasmamalondialdehyde as biomarker for oxidative stress: reference interval andeffects of life-style factors. Clin Chem 1997;43:120914.

[7] Cavalca V, Cighetti G, Bamonti F, Loaldi A, Bortone L, Novembrino C,et al. Oxidative stress and homocysteine in coronary artery disease. ClinChem 2001;47:88792.

[8] Berliner JA, Navab M, Fogelman AM, Frank JS, Demer LL, Edwards PA,et al. Atherosclerosis: basic mechanisms. Oxidation, inflammation, andgenetics. Circulation 1995;91:248896.

[9] Parthasarathy S, Steinberg A. Cell-induced oxidation of LDL. Curr OpinLipidol 1992;3:3137.

[10] Esterbauer H, Wag G, Puhl H. Lipid peroxidation and its role inatherosclerosis. Br Med Bull 1993;49:56676.

[11] Witztum JL. The oxidation hypothesis of atherosclerosis. Lancet 1994;344:7935.

[12] Steinberg D, Parthasarathy S, Carew TE, Khoo JC, Witztum JL. Beyondcholesterol. Modifications of low-density lipoprotein that increase itsatherogenicity. N Engl J Med 1989;320:91524.

[13] Markesbery WR, Lovell MA. Four-hydroxynonenal, a product of lipidperoxidation, is increased in the brain in Alzheimer's disease. NeurobiolAging 1998;19:336.

[14] Mattson MP. Cellular actions of beta-amyloid precursor protein and itssoluble and fibrillogenic derivatives. Physiol Rev 1997;77:1081132.

[15] Niedernhofer LJ, Daniels JS, Rouzer CA, Greene RE, Marnett LJ.Malondialdehyde, a product of lipid peroxidation, is mutagenic in humancells. J Biol Chem 2003;278:3142633.

[16] Esterbauer H, Schaur RJ, Zollner H. Chemistry and biochemistry of4-hydroxynonenal, malonaldehyde and related aldehydes. Free Radic BiolMed 1991;11:81128.

[17] Moore K, Roberts LJ. Measurement of lipid peroxidation. Free Radic Res1998;28:65971.

[18] Nair V, Cooper CS, Vietti DE, Turner GA. The chemistry of lipid per-oxidation metabolites: crosslinking reactions of malondialdehyde. Lipids1986;21:610.

[19] Sinnhuber RO, Yu TC. Characterization of the red pigment formed in the2-thiobarbituric acid determination of oxidative rancidity. Food Res 1958;23:62633.

[20] Carbonneau MA, Peuchant E, Sess D, Canioni P, Clerc M. Free and boundmalondialdehyde measured as thiobarbituric acid adduct by HPLC inserum and plasma. Clin Chem 1991;37:14239.

[21] Hong YL, Yeh SL, Chang CY, Hu ML. Total plasma malondialdehydelevels in 16 Taiwanese college students determined by various thiobarbi-turic acid tests and an improved high-performance liquid chromatography-based method. Clin Biochem 2000;33:61925.

[22] Khoschsorur GA,Winklhofer-Roob BM, Rabl H, Auer Th, Peng Z, SchaurRJ. Evaluation of a sensitive HPLC method for the determination ofmalondialdehyde, and application of the method to different biologicalmaterials. Chromatographia 2000;52:1814.

[23] Largilliere C, Melancon SB. Free malondialdehyde determination inhuman plasma by high-performance liquid chromatography. AnalBiochem 1988;170:1236.

[24] Lykkesfeldt J. Determination of malondialdehyde as dithiobarbituric acidadduct in biological samples byHPLCwith fluorescence detection: comparisonwith ultraviolet-visible spectrophotometry. Clin Chem 2001;47:17257.

[25] Templar J, Kon SP, Milligan TP, Newman DJ, Raftery MJ. Increasedplasma malondialdehyde levels in glomerular disease as determined by afully validated HPLC method. Nephrol Dial Transplant 1999;14:94651.

[26] Wong SH, Knight JA, Hopfer SM, Zaharia O, Leach Jr CN, Sunderman JrFW. Lipoperoxides in plasma as measured by liquid-chromatographicseparation of malondialdehydethiobarbituric acid adduct. Clin Chem 1987;33:21420.

[27] Londero D, Lo GP. Automated high-performance liquid chromatographicseparation with spectrofluorometric detection of a malondialdehyde-

953hemistry 39 (2006) 947954thiobarbituric acid adduct in plasma. J Chromatogr A 1996;729:20710.[28] Nordic Reference Interval Project Bio-bank and Database (http://wip.furst.

no/norip/) (Accessed October 2005).

[29] Solberg HE. RefVal: a program implementing the recommendations ofthe International Federation of Clinical Chemistry on the statisticaltreatment of reference values. Comput Methods Programs Biomed 1995;48:24756.

[30] Working Group of CITAC and EURACHEM. Guide to Quality inAnalytical ChemistryAn Aid to Accreditation. CITAC/EURACHEM.31. 2002. CITAC/EURACHEM.

[31] Yagi K. Assay for serum lipid peroxide level and its clinical significance.Lipid peroxides in biology and medicine. Academic Press, Inc.; 1982.p. 22342.

[32] Knight JA, Smith SE, Kinder VE, Anstall HB. Reference intervals forplasma lipoperoxides: age-, sex-, and specimen-related variations. ClinChem 1987;33:228991.

[33] Aruoma OI, Kaur H, Halliwell B. Oxygen free radicals and humandiseases. J R Soc Health 1991;111:1727.

[34] Pryor WA, Stanley JP. Letter: a suggested mechanism for the production ofmalonaldehyde during the autoxidation of polyunsaturated fatty acids.Nonenzymatic production of prostaglandin endoperoxides duringautoxidation. J Org Chem 1975;40:36157.

[35] Halliwell B, Gutteridge JMC. Free radicals in biology and medicine. 3rded. Oxford University Press; 1999. 408 pp.

[36] Polson C, Sarkar P, Incledon B, Raguvaran V, Grant R. Optimization ofprotein precipitation based upon effectiveness of protein removal andionization effect in liquid chromatography-tandem mass spectrometry.J Chromatogr B Analyt Technol Biomed Life Sci 2003;785:26375.

[37] Agarwal R, Chase SD. Rapid, fluorimetric-liquid chromatographicdetermination of malondialdehyde in biological samples. J ChromatogrB Analyt Technol Biomed Life Sci 2002;775:1216.

[38] Solberg HE. The IFCC recommendation on estimation of referenceintervals. The RefVal Program. Clin Chem Lab Med 2004;42:7104.

954 E. Seljeskog et al. / Clinical Biochemistry 39 (2006) 947954

A novel HPLC method for the measurement of thiobarbituric acid reactive substances (TBARS). A c.....IntroductionMethodsChemicals and reagentsSterile water was used for all the solutionsHPLC instrumentation

Biological materialsStandard and sample preparationAnalytical performanceThe assay from Chromsystems Instruments and Chemicals GmbHStatistics

ResultsHPLC analysis and method developmentCalibrationPrecision and recoveryHydrolyzing acidAnticoagulants and storageThe Chromsystems assaySerum reference intervals for TBARS

DiscussionCorrelation between our HPLC method and the Chromsystems MDA assay

ConclusionAcknowledgmentsReferences